Talk:Interlaced video/Archive 1

Latest comment: 16 years ago by 38.105.193.11 in topic Just a thought
Archive 1

More disadvantages

Another disadvantage of interlacing is that it can only be displayed as is and only on a CRT, without causing problems. Trying to display each field or frame twice to reduce flicker (which occurs at less than around 70 fps) will probably make the interlace artifacts too visible and ruin the image, unless complex deinterlacing (that might cause other problems) is performed first. (So called "100 Hz TVs" are common in Europe today.) Also, the same problems will occur when interlaced material is shown on a flat-screen display like an LCD or plasma monitor, so interlace is not suitable for video distribution today.

That should be mentioned in the article. Actually I think this is a bigger disadvantage of interlacing than the "major" one about reduction of vertical resolution and flickering as mentioned in the article already.

ABostrom 01:37, 2004 Dec 20 (UTC)

I agree with Atlant that the "a" belongs with "frequency" Pdn 03:01, 16 Mar 2005 (UTC)

Terminology Origin

The page currently says "It is sometimes called the top field (since the top line is numbered one, which is odd)". I'm not experienced in this terminology so maybe it is the correct origin, but a more plausible explanation to me would be that odd lines are called the top field because if you split the screen into 2-line rows, the odd rows are the ones on top in each pair. 129.110.240.1 00:00, 19 Mar 2005 (UTC)

Flicker

I don't know if the illustration is correct or misleading, but the interline twitter explanation surely is bogus. The logic is flawed.


I am not an expert on the subject, however my understanding is that the interlace scanning method had a greater level of flicker than non-interlaced (progressive) scanning methods. Here, however, the author states that "..This takes a finite length of time, during which the image on the CRT begins to decay, resulting in flicker. An interlaced display reduces this effect.." which seems to contradict what I have read at other sources, including the wikipedia article on progressive scan. Maybe someone who knows for definite could rectify this.

There are several aspects to flicker which is why one can get different answers depending on the exact question you ask.
If you think of the screen "as a whole", then a screen that is written from top to bottom 60 (50) times per second appears to flicker a lot less than a screen that is written from top to bottom 30 (25) times per second. So considering the screen as a whole (say, looking at it from a distance), interlace definitely helps suppress this sort of flicker.
But then we have the problem of some very small detail on the screen that occupies only one scan line (say, part of some white line-art shown on an overall black screen). Because the object is only one scan line high, this object is only illuminated with every even or odd field, so even with interlacing, the object still flickers at 30 (25) Hz. So in this case, interlace doesn't help the flicker. And if the object is displayed on some sort of (say) uniformly grey screen, the object will appear to flicker much more than the screen as-a-whole. Because of this, systems that generate captions and titles for TV usually take a lot of care to not draw objects that are just one scan line high.
(Did this explanation help?)
Atlant 00:06, 12 May 2005 (UTC)
Flicker is the effect that involves the entire display. Interlace drastically reduces flicker, due to the fact that the refresh rate can be doubled for a given raster scanning format, while keeping the signal bandwidth constant. Interlace also improves motion portrayal - because the refresh rate (sampling rate) is doubled. Tvaughan1 02:14, 23 October 2006 (UTC)
The effect that happens when alternate scan lines are imaging an object with horizontal line detail that is on the scale of the scan lines, therefore making every other scan line light and dark is called "interline twitter". It is an artifact that is associated with interlaced video only... but it is not commonly observed, because most scenes that are being shot with a video camera don't contain this type of detail. Interline twitter can be reduced by filtering the vertical detail in the signal, but this is only done in certain instances... such as in a television studio with very high-end cameras (where the camera is still, and someone might have a shirt or a suit with a pattern that causes interline twitter). If a camera is outdoors, and in motion, and shooting normal every day scenes, interline twitter is seldom observed (due to the somewhat random nature of the scenery, and the fact that the camera is not stationary). Computer generated animation or graphics should be filtered properly when it is converted to interlaced video, as it is not so random, and twitter may occur. Tvaughan1 02:14, 23 October 2006 (UTC)

Comparing interlace to non-interlace

There is a confusion that runs through this article when comparing interlace to non-interlace. What exactly is being compared? That is, what factors are changing and what factors are being held constant? (Scan rate, lines per frame, etc).

The muddle starts in the first paragraph. "The method causes less visible flickering than non-interlaced methods." This is true under some conditions and not true under others. It all depends on what is being compared.

Another problematic statement is, "The major disadvantage of interlacing is the reduction in vertical display resolution." One can just as easily say that the major advantage of interlace is the increase in vertical resolution. Again, it depends what is being compared. If you compare an N-line non-interlace system with a N*2 line interlace system (a very reasonable comparison to make), which has more vertical resolution? Mirror Vax 09:06, 14 August 2005 (UTC)

The problem here is that in many places the article compares regular interlaced scanning to a hypothetical 30 frame progressive scan rate that has probably never been seen since the 1920s, and would flicker like a strobe light. No one advocates this. Anything currently calling itself "Progressive scan" must rely on either frame repeating built into the set, (as in HDTV) or conversion to PAL/NTSC before broadcast. (as in film transfers and "film look" video) For the rewrite, all such comparisons should be made between technologies of similar time periods and realities. Algr

I'm certainly no expert on this subject, but it seems to me the first item in "The alternatives to interlacing" is inaccurate. "Doubling the bandwidth used and transmitting full frames instead of each field" describes a progressive input at the same frequency as the interlaced one, which should certainly give you a better image. I.e., instead of showing only half the pixels in an image every 60th of a second, you would be scanning and transmitting all the pixels every 60th of a second. How could that not result in better picture quality? You would not have twitter, nor combing. (BTW, I'm not saying progressive 60Hz broadcast is practical or desirable) cortell

More cleanup needed!

I was going to put the POV-section and inappropriate tone headers on the "Interlacing as a data compression technique" section, but I decided to take a crack at 'em myself. Some parts I left alone (save for fixing typos) because I couldn't immediately come up with a better wording. On the other hand, some parts (such as the "like a tank" bit) were easy to replace. I hope that the changes I made didn't introduce any technical inaccuracies. Meanwhile, this article still needs a good going-over to improve the tone and remove spelling and punctuation errors. It may also be in need of technical clarification, as another person commented, but I'll leave that to those with more knowledge of the subject than I. --Jay (Histrion) (talkcontribs) 19:02, 1 December 2005 (UTC)

Removal of the "Interlacing as a compression technique" section

I object against the removal of the "Interlacing as a compression technique" secion in the following changeset:

(cur) (last) 12:10, 3 December 2005 Smyth (→Interlacing as a data compression technique - As the article makes quite clear, there never were any full frames.)

Also, the rationale presented by the editor is bogus. Of course there were full frames. The image focused by the lens on the back of the vidicon tube or a CCD, and recorded in its sensors is a full frame. In both cases only half of the lines of sensors are read out and encoded, the other half is simply reset when preparing for the next readout cycle.

The lines were ignored because 60 fields per second produces less eyestrain on a CRT than the alternative 30 frames per second. Saving bandwidth has nothing to do with it at all, and comparing interlacing's 2:1 "compression ratio" with MPEG's 22:1 is ridiculous. Much of the other text I removed was either out of place (discussion of how deinterlacing is computationally intensive) or completely wrong (stating that progressive devices are twice as complex as interlaced ones, or that interlaced video is impossible to edit).
But the article was missing some discussion of how interlaced signals interact with digital compression, so thanks for adding it. – Smyth\talk 17:11, 4 December 2005 (UTC)
The interlacing is clearly intended to save bandwidth when transmitting video-based material (50/60 Hz), because 50 progressive frames take twice the bandwidth of the 25 interlaced. I wasn't talking about the film-based material (24/25 Hz, as well as NTSC 30 Hz), where interlacing is serving a different role, flicker avoidance. The point of interlacing against MPEG comparision was that the digital compression is more suitable for saving bandwidth than the interlacing when transmitting video-based material. However I admit, that this could be put in a more clear manner, without the detailed comparision.
And of course the progressive devices are twice as complex, becase: MPEG decoder requires 1) twice bigger frame buffers, 2) twice more CPU power for block transformation, twice more CPU power and bandwidth for encoding the transformed blocks into stream, and twice more capacious and twice faster storage; however it is stupid to limit the frame format to the capabilities of the current technology, because the technology will develop tomorrow, but the recording/works of art will be spoiled by the limitations today.
I will reedit the article taking this into account and assuming that your misunderstanding was solely due to the unexplained video- against film-based material difference.Greengrass
I don't think that anyone has ever seriously suggested that capturing 50 or 60 progressive frames a second is a sensible use of either bandwidth or storage. The proper comparison is between 2N fields/sec and N frames/sec, not 2N fields/sec and 2N frames/sec. – Smyth\talk 19:59, 4 December 2005 (UTC)
Progressive image is universally considered superior to interlaced, whether it is filmed at 24 frames a second or 50 or 60 frames a second. The bandwidth or storage use are of little importance if the picture looks bad. Anyway, one cannot say I did not covered in depth the topic of bandwidth.Greengrass
60 progressive frames per second IS sensible and is precisely what ABC and ESPN are doing with their 720p sports broadcasts. 24-30 frames per second is never adequate for sports. Interlace creates obvious line crawl around the lines on the field, even at 1080i. Algr 09:12, 27 January 2006 (UTC)

Interlace illustration

I'm new here so I thought I'd put this picture up for review before putting it up on the real article:

| 

This picture illustrates how sharpness is lost due to interlace. The left and center images are identical except for interlace. But the center image has too much flicker because objects in the picture tend to line up with just one field. In the flag, for example, there are areas where one field hits mostly red lines, while the other hits mostly white.

Real interlaced transmissions deal with this by blurring out such details, as seen in the image on the right. However, while a line doubler might restore such image to progressive scanning, the necessary blurring would remain.

A Technical note: This animation runs at 30 hz, not the 50-60 hz that real monitors use, and may appear even slower on your monitor. The effect of interlace is therefore magnified.Algr Dec 16, 2005

Interesting, but it needs lots of caveats about how the interlaced frames will be torn to pieces because they aren't synchronized with your monitor's refresh rate. – Smyth\talk 13:20, 17 December 2005 (UTC)
Well, it will be a while before we get readers who have never seen an interlaced TV set, so I think a warning like "The effects of interlace are magnified for illustrative purpuses." should do. In 20 years everyone will be able to run the animation at 50 hz, and it will let people know what interlace used to look like. Algr Dec 17, 2005
The problem is, the effects of interlace aren't strictly "magnified", because you don't get frame tearing on an interlaced TV set, and so the animation makes interlacing look much worse than it actually is. Would it still be instructive if you reduced it to 10Hz, or even less? – Smyth\talk 11:24, 18 December 2005 (UTC)
Tearing huh? I'm not seeing that on my screen - maybe because I have an LCD? Here it is at 10 hz: Algr 16:54, 18 December 2005 (UTC)

| 

Tearing happens because browsers generally do not bother synchronizing the display of GIF animations to the vertical refresh. — Jukka Aho 03:42, 31 October 2006 (UTC)
Much better. But why are the colors different in the "progressive" image? – Smyth\talk 20:04, 18 December 2005 (UTC)
I dimmed the progressive image to match the loss of brightness caused by black lines over the interlaced version. Strangely, one or the other looks brighter depending on what angle I view the monitor at. The brightness matches when I view the illustration at a 90° angle. (Face on to the monitor.) Algr 22:26, 18 December 2005 (UTC)

I've decided I'm not to happy with my drawing. If someone wants to offer up something better, I can put it under the scan lines. A good image should have near horizontal lines or texture approaching the width of the scan lines.Algr 08:46, 24 January 2006 (UTC)


Get rid of this image. In a properly designed interlaced system a static image would look nearly the same as the progresssive one. This is Dis-information. 221.28.55.68 21:08, 23 March 2006 (UTC)

Has there ever been a "properly designed interlaced system"? VGA exists because the effects of interlace couldn't be dealt with. (The Amiga tried to do it. It worked for video production but was a liability everywhere else.) Why is progressive scan output of DVDs such a big selling point then? Algr 21:53, 23 March 2006 (UTC)
The problem with the Amiga (and its tv-compatible interlaced video modes) is that it does not apply any kind of vertical filtering to contrasty, thin, horizontal details that only occupy a single scanline. These kind of details cause unsightly twitter because they will essentially disappear from the screen for the duration of every other field. Modern video encoder chips (such as those used for "tv out" purposes on PCs) employ filtering techniques for reducing the twitter. They work by smearing the one-line-thick horizontal details over a taller area (the surrounding lines in the next field) and by reducing the contrast of the offending, thin detail when compared to its surroundings. The most advanced chips implement all this in an adaptive and parametrical way: there are levels and thresholds you can set, and the chip will react differently to different kinds of image content – preserving as much detail as possible but processing the picture where necessary. (For example, bright, thin horizontal details will get a more aggressive treatment than the dim ones.) The results are usually pretty good: the twitter is considerably reduced by this technique.
The Amiga would have greatly benefited for this kind of design. These kind of advanced filtering methods, however, weren't feasible nor affordable at the time the Amiga's graphics chip was designed, so the Amiga does absolutely no filtering in the vertical direction, and suffers from considerable twitter problems in its interlaced modes – at least with certain kinds of “pathological” image content.
Even though artificial, computer-generated images can be very difficult for interlaced systems, at least when not filtered properly, it should be kept in mind that they are a bit of a special case: different from natural images shot with a video camera. A video camera will usually limit the vertical resolution in order to avoid aliasing, and natural images rarely have thin horizontal features that would exactly align with the sample sites of an image sensor, anyway. (That said, occasionally you see problems with striped shirts, etc., but this has been already discussed elsewhere on this talk page. Sometimes you can also see badly telecined film material which has lots of [wholly unnecessary] twitter problems because of lack of filtering.) — Jukka Aho 03:42, 31 October 2006 (UTC)

Because movies are shot on film at 24 frames per second, and because film is a progressive scan capture format. Video will always look better if it isn't converted between progressive and interlaced systems. If the capture-storage-transmission-display system is either all interlaced or all progressive scan, the video quality will be optimal (all other things being equal). That is what is wrong with this "illustration". It purports to show what interlaced video looks like... but you can only see interlaced video on an interlaced display... not on your PC monitor which is progressive scan. Tvaughan1 05:22, 1 October 2006 (UTC)

Thank you for illustrating. It is helpful. Avé 23:26, 23 March 2006 (UTC)

I made a new comparison animation, but it don't think that I made it right here it is:
File:Progressive vs interlace.gif
Make a new one please...--Finest1 02:27, 25 June 2006 (UTC)

It looks like you scaled the lines after they were placed on the image, because they look much too fine, and rather irregular. Also, what is the copyright status of that image that the lines are on? Algr 08:58, 25 June 2006 (UTC)

Uh... I don't know. Delete it!--Finest1 19:22, 28 June 2006 (UTC)

You can not display an interlaced signal on a progressive scan monitor, then declare that interlaced video is inferior... unless you don't know what you are talking about. This illustration is dis-information. It mainly shows the effect of aliasing, which happens only on digital sampling systems that are implemented improperly (without the necessary low-pass filtering below the Nyquist frequency). The illustration is naturally displayed on computer screens which are progressive scan displays. As stated by another user above... in a properly designed interlaced system a static image would look nearly the same as on a progressive scan system.Tvaughan1 05:22, 1 October 2006 (UTC)

"without the necessary low-pass filtering below the Nyquist frequency" - But that is precisely what I have been trying to explain to you in Image talk:Resolution chart.svg. Low pass means blurring out high frequencies, which is why 1080i can never look as sharp as 1080p. Of course the effects of interlace are magnified on a 10 hz gif, but that is a useful way to show the reader what we are talking about. The illustration on the right shows the filtered version of the image, which has less detail. The middle image is not filtered, so it could be de-interlaced to the same detail as the progressive image on the left - however it looks worse. Camcorders from Sony (VX 2100, TRV 900) and Panasonic (AG-DVX100) can record video without low-pass filtering, such as is shown in the middle. This looks harsh and very twittery when viewed on a TV, but Panasonic recommends it when video is to be printed to film, or viewed on a PC display. Algr 06:38, 1 October 2006 (UTC)
Also, the only technical problem with simulating interlace on a progressive monitor is that you loose half your brightness, since half the pixels are black at any one moment. I've compensated for that by dimming the progressive image by an equal amount. So it is a perfectly fair comparison. The main purpose of that image is not to "prove" anything, but to show the reader what to look for when he/she examines real TV sets. Algr 20:10, 1 October 2006 (UTC)

No, this illustration is not accurate or useful. Video systems are designed to display pictures in motion. This illustration shows a static image. With interlaced video both fields are displayed on the monitor simultaneously... one field doesn't blank while the other is displayed! An interlaced display would look identical to the progressive scan illustration on the left. The image was generated using PC software... not with a video camera or any video equipment. The image purports to show something that no real-world video system will show you. Both interlace and progressive scan displays rely on certain properties of the human visual system. Motion is simulated by showing a sequence of static pictures. The human visual system perceives the motion and the detail all at once. Interlace video systems display two fields for every frame, giving smoother motion and enhanced spatial detail. Do you have an HDTV? Go up close to it while watching any HD TV program. You will never see details that are down to the pixel level. If you go up close you will see macroblocks from the MPEG compression. This is not a "real" problem because the system was not designed to be viewed from 1 foot away... NTSC is designed to be viewed from a distance that is about 6 times the picture height away, and HDTV from about 3 times the picture height. Video systems are just that... systems. You can't evaluate a system by looking at just one part of it. Interlace signals improve signal to noise because they improve the utilization of the available bandwidth from camera sensors (CCDs, CMOS, etc.) to transmission systems. Anti-alias filtering on interlaced video is absolutely trivial compared to the discrete cosine transform that happens in the MPEG compression. There is no way that detail at the pixel level would survive the MPEG compression that is a standard part of HDTV. It's funny that most people are watching 1080i HDTV and they never experience the pulsing, flashing striped image that this illustration attempts to illustrate. Tvaughan1 00:06, 3 October 2006 (UTC)

To illustrate the problem, I've modified this image to show what a progressive scan display on a CRT would look like, assuming that the persistence of the phosphors was approximately 1/60 of a second... which is what would have to be the case in order for every other interlaced field to be dark, as portrayed by this image. With such a monitor only half of the progressive scan picture would be displayed in 1/60 of a second. So, in the same time that the one field of the interlaced scan is displayed you will see only half of the progressive scan picture. Tvaughan1 03:06, 3 October 2006 (UTC)

 

Of course, this is nonsense, as progressive scan monitors have a refresh rate that is typically 60 Hz or higher. In other words, they repeat fields of video. But interlaced video doesn't look like this either, as phosphors persist, and the human eye averages all of the light received during the 2 fields that make up a single frame of video, as well as averaging the motion of the objects between fields or frames. If the human eye didn't average the visual presentation of the fields, movies that are shown at 24 frames per second would look strobed. Because interlaced video will update the motion of objects and the camera (panning or zooming) 60 times per second, while progressive scan video of the same bandwidth updates motion only 30 times per second, interlaced video gives you a more accurate representation of the motion portrayed in the video. In other words, interlaced video reduces flicker while improving spatial detail. Tvaughan1 03:06, 3 October 2006 (UTC)

New discussions are supposed to be at the bottom, so lets move things there. Algr 06:04, 5 October 2006 (UTC)

2

 

I'm sorry Tvaughan1, but you are the one who doesn't know what he is talking about, as this high speed exposure clearly shows:

As you can see, the phosphors DO NOT persist. A mere two inches above the electron beam, they have already lost most of their glow. (That distance represents approximately 1/500 of a second.) Each field is long gone when the next one is scanned. The problems you site with viewing interlaced video on progressive displays are PRECISELY BECAUSE early attempts at this would incorrectly have both fields visible simultaneously - resulting in "mice teeth" and the smearing of moving objects. So my graphic correctly shows (within the limits of the frame rate) what actually happens on an interlaced monitor. The effect is enlarged for illustrative purposes, but if you look closely at an NTSC screen you will see the same thing.

"while progressive scan video of the same bandwidth updates motion only 30 times per second"

Again you are going out of your way to miss the point by nonsensically comparing ANALOG bandwidth, when only the DIGITAL transmission from station to tuner is limited. You are comparing 60i to 30p, and ignoring that 60p is also part of the broadcast standard. Motion only makes things worse for interlace, because as soon as an object starts moving, the two fields no longer line up and 1080i effectively becomes 540p, while 720p has a new set of 720 lines every 60th of a second. In any case, CRTs are virtually obsolete for HDTV - everything is DLP, LCD or Plasma, and these sets are only hindered by interlace. Algr 20:15, 4 October 2006 (UTC)

The point was, progressive scan on a CRT would look like the image that I made, under the assumptions that you must have in order for interlaced video to look like your illustration. If you remove these assumptions, than interlace and progressive look nearly identical (like the image on the left). So which is it? Does your illustration illustrate a CRT? What would your progressive scan image really look like on this monitor? Is the monitor an LCD? If so, why would it blank every alternate field? It wouldn't. While I understand that you were trying to illustrate how interlace works, this illustration makes interlaced video appear to have a lot of flicker... just the opposite of what happens in the real world. Tvaughan1 01:27, 5 October 2006 (UTC)
Algr - Why compare 1080i60 to 1080p30? They are identical in terms of bandwidth. You can't compare apples to oranges. Think for a minute - Who is broadcasting 1080p? Who is shooting 1080p video at 60 frames per second? Sure, I'd prefer 1080p60 to 1080i60... who wouldn't? It's twice the detail! The only problem is; it's not available. It requires twice the bandwidth. It's not practical or affordable at this point. Do you know why? It's not a problem with the display technology. 1080p60 cameras don't exist. Broadcast standards have to be considered as a system. The most expensive HDTV cameras in the world can shoot 1080i60 or 1080p30 in full 1920x1080 pixel resolution (recording to HDCAM SR, or hard disk via HD SDI). If you want more accurate motion portrayal and greater spatial detail, you choose 1080i60 over 1080p30. For a given bandwidth you get a better picture. Of course, if you own a (native progressive scan) LCD monitor you might not think so, but that's another issue.
You're just not getting interlace. The 2 fields DON'T NEED TO LINE UP. They occur AT A DIFFERENT TIME... 1/60 of a second difference. They are captured at different moments in time, and they are displayed at different moments in time. You just keep trying to understand interlace by thinking in progressive scan mode. You can't display both fields in the same still frame... they aren't still frames, they are interlaced VIDEO frames. You seem to be stuck on the problem that occurs when you want to display a video signal that was captured as interlaced video on a progressive scan monitor. Yes... this is a problem... but this is not the way interlaced video was designed to work. Interlaced video is designed to be displayed on an interlaced monitor. Just because you happen to own an LCD monitor doesn't mean that interlaced video sucks, or doesn't work well.
As to the (off topic) point that CRT is obsolete... Do you know what "color space" is? CRTs have a much better color space than LCD. Tell me... if you were outfitting a professional HDTV broadcast or video post production facility, what would you use for your reference monitors? Pick a make and model... money is no object. I'd pick something like the Sony BVMA-32E1WU. I wonder why an "obsolete" 32" CRT would sell for $37,000. I wonder why Hollywood movie studios, post production facilities and broadcasters use CRTs instead of LCD monitors for any critical work (sure... they use LCDs for editing... but not color correction or critical monitoring).
As for your flickering "illustration" of interlaced video... interlace REDUCES flicker (versus progressive scan). It also avoids motion judder... an artifact of progressive scan capture systems. You can't "demonstrate" interlaced video on a computer monitor that is displaying progressive scan images (like a Wikipedia page showing your animated GIF). People don't see flickering on their CRT television sets... interlace works just fine... nothing like your "illustration".
You're also missing the fact that the human eye has a dynamic range that is orders of magnitude more sensitive than the camera that took the picture of the phosphors, and that the human eye / brain connection works through chemical and electrical impulses that can't distinguish light level differences at 1/500 of a second. It's a trick! But guess what? It works. To mimic the human visual system you have to set your shutter speed at around 1/20 of a second. But CCD camera sensors are a very poor stand-in for your retina and brain. Let me repeat - you CAN NOT evaluate video systems by taking a still picture, or looking at a still frame. Tvaughan1 23:01, 4 October 2006 (UTC)

I disagree with "24-30Hz is a sufficient video frame rate for smooth motion"

The statement "24-30Hz is a sufficient video frame rate for smooth motion" doesn't ring true for me. 50i video is much smoother than 25p. I can't think of a good alternative though.

I think you may be confusing the field rate with the frame rate. Otherwise, an entire film industry disagrees with you :-) (There's no doubt that 50 frames/second looks better than 24 frames/second, but "sufficient for smooth motion" is a lesser criterion.)
Atlant 12:45, 12 February 2006 (UTC)
What I suppose I mean is that 50i material (i.e. sports) is clearly smoother than and easily distinguishable from 25p material, which has a definite "stutter" (modern drama). I wouldn't say that 25p is sufficient for smooth motion, but it's enough to maintain the illusion of motion. 12p is even enough sometimes (at least when you're watching Wile E. Coyote chase Road Runner) to maintain that illusion, but that doesn't mean it's smooth.
What you are describing is called high motion, which I wrote about. I agree that the second bit about more frames being wasteful is wrong, so I've change this whole bit. BTW, you might want to sign your posts by typing four of these "~" at the end. Algr 16:04, 5 March 2006 (UTC)

Just a thought

As a hobbyist video editor i get to deal quite a lot with interlance and it's artefacts in the video footage. I can recommend one very good and reliable source of information about interlace: http://www.100fps.com I learned great deal about interlace from there. And indeed, lots of this in this wikipedia's article is messy. Usinf 100fps as reference we could correct alot of things. Robert 22:26, 1 March 2006 (UTC)

: Thanks for the reference.  I'll include it. Algr 16:06, 5 March 2006 (UTC)

The site is down. —Preceding unsigned comment added by 38.105.193.11 (talk) 21:03, 11 January 2008 (UTC)

Needs more work

This is supposed to be an encyclopedia! If you don't know, don't guess! I am not enough of an expert to author a complete rewrite, but I know much of this is totally wrong.

  • Example from "Application" section:

"The film solution was to project each frame of film twice, a movie shot at 24 frames per second would thus illuminate the screen 48 times per second."

WRONG! Think about it a little bit before posting. MOVIES ARE SHOT (AND SHOWN) AT 24 FRAMES PER SECOND. Period.

"Illuminating the screen 48 times per second" implies displaying the frame, then (unnecessarily!?) blanking, then displaying the same frame again.

All of this discussion about flicker is confusing and misleading. The perception of flickering doesn't depend on frame rate or field rate...it simply depends on HOW LONG THE SCREEN IS DARK or dimmed. More than maybe 1/60th of a second (give or take a zero...I'M JUST GUESSING. Somebody please look it up before moving this guess to the article) makes the viewer see flicker. Modern movie projectors can flip to the next frame very fast, blanking the screen for ever-smaller periods as the technology improves. I'm GUESSING that they're keeping the frame illuminated 75% of the time or more these days, thus taking less than 1/96th of a second to blank, move to the next frame, and unblank.

And, CRT makers can control the PERSISTENCE of the phosphor (how long it continues to glow after being struck by the electron beam during the scan.) They tune to the application. If the phosphor is too fast for the frame rate, flicker will be seen; if it's too slow, moving objects will leave trails.

  • The first part of the "Application" section is confusing too, and goes against what we see in real life:

"When motion picture film was developed, it was observed that the movie screen had to be illuminated at a high rate to prevent flicker. The exact rate necessary varies by brightness, with 40 hz being acceptable in dimly lit rooms, while up to 80 hz may be necessary for bright displays that extend into peripheral vision."

Firstly, movie theaters show movies at 24hz, not 40hz or 48hz, as explained above. Secondly, the wide screen at the theater extends way farther into one's peripheral vision than a television or desktop monitor, unless you have your face up to the glass. Thirdly, refresh rate is completely irrelevant to "flicker" as explained above. Fourthly, I don't think I agree with the assertion (guess?) that "the rate necessary varies by brightness". Brightness of what? The display? The environs? The display relative to the environs? If it's the latter (which seems to make the most sense) then movie screens are by far the brightest vs. a dark room. Movies have a 24fps frame rate, and are certainly "acceptable in a dimly-lit room."

The whole "Application" section needs to be rewritten/expunged.

  • The section about interlacing as a compression method leaves me highly skeptical too. Some people guessing again, is my guess. I vote it be ripped out. My understanding is, there's no compression advantage to interlaced video. Higher field rate, lower resolution....twice as many half-size images to compress as progressive-scan. Halving the vertical resolution is unlikely to cut the compressed field size in half vs. a full frame; so if anything, there's probably a net loss.

At any rate I have no idea how to fix it. Rip out whole sections and rewrite them, maybe? But certainly the WRONG datum about 48 illuminations per second on movies needs to be gracefully done away with somehow.

Somebody help me out here, I'm afraid to even touch this article.

""Illuminating the screen 48 times per second" implies displaying the frame, then (unnecessarily!?) blanking, then displaying the same frame again." I believe that is exactly what is done, the reason being that blanking at 48Hz is less noticable than at 24Hz. Of course it would be better to actually display different frames, but for cost reasons film is limited to 24fps. I'm not sure what this has to do with interlacing, though. Mirror Vax 04:45, 20 March 2006 (UTC)
To the unsigned commentor: You are being quite rude with your tone, and you are also quite wrong. Please don't touch the article. I specifically wrote that article to address some common misperceptions that I have encountered. Do some open minded research and you will find that the situation is quite different then you have in mind. Algr 15:49, 20 March 2006 (UTC)

... My apologies. I did research it, and movie projectors DO blank at 2x the frame rate, or even 3x. So that part is OK. And, sorry if my tone was offensive. I'm new to this; I'll try to keep it scholarly and objective in the future. ...I'm still dissatisfied with the article as a whole though. --shyland

I'm still planning on working on it. I only rewrote the top part so far. Algr 04:33, 21 March 2006 (UTC)

I edited the title of this section to make it less offensive... I'll keep my hands off the article. Shyland 16:52, 21 March 2006 (UTC)

long-persistence phosphors

which used "slow" or long-persistence phosphor to reduce flicker.

This isn't correct. Long persistence phosphors were never used in consumer devices - only for things like radar. On a TV set, the phosphor dims out within 30 scan lines or so, (about an inch) so this has no effect on interlace either. See the photo here: Refresh rate Algr 06:14, 15 May 2006 (UTC)

Misunderstanding of afterglow and 'persistence of vision'.

  • There is no significant afterglow on a CRT TV display. I once took a photograph at high shutter speed to test this, and found that the picture faded over about a quarter of the picture height, in other words in 1/200th of a second. Nor is 'persistence of vision' the simple thing it seems. I believe, from experiments, that it is processed image content in the brain that persists, not the image on the retina. Interlacing works because of this. The brain does not combine subsequent frames; if it did we would see mice-teeth on moving verticals, as we do on computer images or stills, which in fact we don't see on a CRT display. The brain also cannot register fine detail on moving images, hence I think litte is lost on a proper CRT interlaced display, while motion judder is reduced as well as flicker. Modern LCD and Plasma displays seem to me to be fundamentally unsuited to interlaced video since they necessitate de-interlacing in the TV with inevitable loss. In theory, it is not impossible to make an interlaced plasma or LCD display, in which the two fields were lit up alternately, but in practice this would halve brightness, even if the response was fast enough. In view of this, I think it is a great pity that the 1080i HD standard was created, since it is unlikely ever to be viewed except as a de-interlaced compromise on modern displays. If 1080p/25 (UK) were encouraged worldwide, then de-interlacing would not be needed. 1080p/25fps requires no more bandwidth than 1080i of course, but it has the advantage of being close enough to 24fps to avoid 'pull down' on telecine from movies (in the UK we just run movies 4% fast and get smoother motion.) It also fits well with the commonly used 75Hz refresh rate of many computer monitors, which would just repeat each frame three times for smooth motion. In high-end TV's processing using motion detection could be used to generate intermediate frames at 50 or 75Hz as was done fairly successfully in some last-generation 100Hz TV's. Reducing motion judder in this way as an option is a better way of using motion detection than de-interlacing, because it can be turned off or improved as technology progresses. I note that the EBU has advised against the use of interlace, especially in production, where it recommends that 1080p/50fps be used as a future standard. --Lindosland 15:26, 23 June 2006 (UTC)

odd field drawn first

This is not true, although it has been adopted in some applications in later years and IT and frame based stores and sensors were developed. The original PAL specification does not talk of this.

Interlaced tube cameras produce 50 321.50 line pictures, each different being starting 20msecs after the previous and there is no sense in which either field is first.

When film is scanned at 25fps for 50Hz television (European PAL) there are always two fields derived from one film frame. In the first telecines, the first of the two fields could be either odd or even. It makes no difference visually which.

In fact the manufacturers telecines were modified by some broadcasters (eg BBC) so that the first field scanned was always odd: this was because in the 1960s and early 70s it was common to re-record TV programs to film and it was important that each recorded film frame corresponded to on original film frame only.

Tony Gardner Gardnan 21:12, 5 November 2006 (UTC)

Interlace causes/prevents flicker?

Maybe part of the problem here is that the word "flicker" is being used to describe two completely different phenomena. One of them is caused by interlace, and the other is prevented by it.

In CRTs, interlace prevents strobing flicker.

In ALL displays, interlace unavoidably causes up/down juddering that is most obvious on absolutely static images. Non-CRT displays usually attempt to deal with the problem in ways that create artifacts of their own...

Today, we don't NEED to mangle the video in order to reduce strobing flicker, because we have framebuffers. This makes it possible to just display the same image, twice as often. Isn't this essentially what film projection is doing already? Snacky 03:22, 5 October 2006 (UTC)

The problem with Tvaughan1's argument about interlace preventing flicker is that he is comparing a real system (example: NTSC) with a pathological one that no one would ever use. (480 progressive lines, 30Hz refresh rate.) But if you want to get an NTSC set to scan progressively, it is actually very easy: Connect it to a Super Nintendo! (Or other videogames of that era or older.) Those old games generated a slightly nonstandard signal that caused the scan lines of each video field to scan directly on top of the previous one, rather then between them. This effectively makes each field a complete frame: It's 240 progressive scan lines at 60 frames per second. Since SNES and NTSC are both real formats using the same analog bandwidth, THAT is the logical comparison to make. The progressive SNES signal eliminates all judder and twitter, and has the same large-area flicker as NTSC. It's not as sharp as a properly interlaced signal, but the difference isn't as large as the numbers 240/480 would suggest. Algr 06:27, 5 October 2006 (UTC)
Huh? Please stop "characterizing" my responses. Other contributors to the article can read and characterize my statements all by themselves. This is not one person's article. Wikipedia articles are supposed to be a collaborative effort. We need to discuss things calmly and rationally... and we need to listen to one another (actually reading what the other person writes on the discussion page). Just make your arguments. Where did I compare a real system to a "pathological system that no one would ever use"? You need to actually read what I wrote, and try to understand it a little better.
Interlace reduces flicker. Flicker is the strobing effect of a monitor, caused by successive fields or frames being displayed. The "other flicker" that Snacky is describing is not flicker... it's twitter. Twitter only occurs when the subject being videotaped has detail that approaches the vertical resolution of the format. For most subjects and situations, twitter isn't an issue. You most often notice twitter when a person being videotaped is wearing a striped shirt with very narrow light and dark stripes. Television on-air "talent" is trained not to wear narrow striped shirts or dresses... but sometimes they forget, or a non-professional is being interviewed and they happen to be wearing clothing that causes twitter. By changing the level of zoom, the videographer can avoid line twitter. For most subjects, it isn't an issue (because the camera and/or the subject is in motion, and the subject doesn't have detail that alternates light and dark in lines that approach the format resolution). So, 99% of the time twitter is not an issue. Flicker is rarely an issue anymore these days, due to the fact that displays are able to use refresh rates of 60 Hz and greater. But any suggestion (or illustration) that tends to lead readers of this article in the direction that interlace causes or increases flicker is simply wrong. That is why I have a major concern with Image:Interlace10hz.gif. The interlaced portion of this illustration is a perfect illustration of flickering, but not of interlaced video. You CAN'T illustrate interlaced video on a progressive monitor, and everyone viewing this illustration is using a progressive scan monitor to view the illustration. Furthermore, the illustration is original work, not verifiable, and not from a neutral point of view. Tvaughan1 17:35, 5 October 2006 (UTC)


Tvaughan1, this entire argument stems from you "characterizing" my http://en.wikipedia.org/w/index.php?title=Image:Resolution_chart.svg graphic to mean something it doesn't, so you are one to talk. You've asked me to listen, but you ignore everything I've said, and don't even read your own references properly. Both times you've mischaracterized what people you quote are saying, and ignored my references. [1] You have no reference for interlace having less flicker then progressive, no reference for interlace being better at motion, and no reference for interlace being as sharp as a progressive display with the same line count. And every post you've made has clear errors it it. Here is another one: "Flicker is rarely an issue anymore these days, due to the fact that displays are able to use refresh rates of 60 Hz and greater." - "These days"? TVs in the 1930's refreshed at 60 hz, there was no other option because they had to be synced to line current.
-
"The illustration is original work, not verifiable, and not from a neutral point of view." This also describes just about everything that has ever appeared on the "featured picture" section of Wikipedia's main page. Another of my works will be there in a few weeks. Algr


From Digital Video and HDTV Algorithms and Interfaces, Charles Poynton, copyright 2003, Elsevier Science

"A sequence of still pictures, captured and displayed at a sufficiently high rate, can create the illusion of motion. Many displays for moving images emit light for just a fraction of the frame time: The display is black for a certain duty cycle. If the flash rate - or refresh rate - is too low, flicker is perceived. The flicker sensitivity of vision is dependent upon the viewing environment, and the larger the angle subtended by the picture, the higher the flash rate must be to avoid flicker. Because picture angle influences flicker, flicker depends upon viewing distance. ...

Twitter - The flicker susceptibility of vision stems from a wide-area effect. As long as the complete height of the picture is scanned sufficiently rapidly to overcome flicker, small-scale picture detail, such as that in the laternate lines, can be transmitted at a lower rate. With progressive scanning, scan-line visibility limits the reduction of spot size. With interlaced scanning, this constraint is relaxed by a factor of two. However, interlace introduces a new constraint, that of twitter."

I don't want to quote too extensively ... this is copyrighted material. But these concepts are all explained thoroughly and accurately in several books I own... "Digital Television Fundamentals" explains it on page 5...... "Modern Cable Television Technology" explains it on page 33 (a 1000 page book). These concepts of scanning and flicker and spot size are fundamental - they are discussed right at the beginning of any good book on video or television. As User:shyland mentioned... "this is supposed to be an encyclopedia... if you don't know, don't guess". If you are a video enthusiast, don't just surf the Internet Googling various topics... do some basic homework - buy a good book and read it! This whole article on Interlace is not accurate, nor is it written from a neutral point of view. Interlace has advantages as well as disadvantages. Try finding any mention of advantages of interlace in this article. If interlace was so bad, wouldn't we have discovered this by the 1990s? Wouldn't the Advanced Television Standards Committee have left interlace behind? They discussed the topic thoroughly, and decided to continue to utilize the technique. Tvaughan1 01:44, 6 October 2006 (UTC)
Yes, I know that TVs in the 1930s had a field rate of 60Hz... thanks to interlace solving the problem of flicker!! My point was, without interlace, you could send the same level of detail at 30 Hz and have a big problem with flicker. Until electronics could be developed that could scan an entire monitor with good detail progressively at 60Hz or faster, you had to use interlace to avoid flicker (and still have good resolution). And no... televisions never, ever "synched to line current". TVs synchronize to synchronization pulses within the broadcast signal (horizontal and vertical synch pulses). In other words, TVs slave to the NTSC or PAL master signal. Tvaughan1 01:44, 6 October 2006 (UTC)

Advantages of Interlace!

I recommend that the article include some discussion on the advantages of interlace. Although the historical section explains some advantages, there is a big section entitled "Problems caused by interlacing", but no corresponding section pointing out the advantages. Not exactly a neutral point of view. Rather than just editing the article, I'd like to challenge other editors to do their homework on the subject, and contribute their thoughts.

First, interlace allows you to double the field rate for a given bandwidth. This is not a minor, trivial point. It means you can sample the moving picture (the scene... the subject... the action) at twice the sampling frequency. You know the human hearing range is approximately 20 - 20,000 Hz. That's the frequency response of your ear and brain... the human hearing system. So what is the frequency response of the human visual system. You know your eye and brain average the effect of a sequence of still pictures, or a sequence of pulses of light. But you also know that there is this problem with flicker (a noticeable strobe effect) that can occur at frequencies as high as 50 Hz or 60 Hz, or sometimes even higher, depending on the intensity of the source, ambient lighting, and how close you are to the source. Did you know that movies that are shot on film at 24 frames per second do not do a good job of portraying motion? Interlaced video shows motion much better than 24P movies, or 30P video (yes, there is such a thing)... because motion is sampled at 60 Hz. Anyone who understands digital sampling theory knows that doubling the sample rate gives you better frequency response. Would you rather listen to music sampled at 22.05 KHz, or music sampled at 44.1 KHz?

Second, interlace allows a CRT to have a reduced spot size... half the size, as a matter of fact. Interlace allowed the designers of television as we know it to double the spatial detail on the screen. This is because if you create a CRT without interlace using scan lines that are too close, the gaussian beam profile overlaps too much. Since alternate fields are mostly faded due to the short persistence of the phosphors on the screen, interlaced lines can be spaced closer.

Third, although interlace introduces a potential problem with interline twitter, this is not a commonly seen problem. Only very high-end cameras (or CGI systems that are generating video from animation) actually reduce vertical resolution of the signal in order to prevent possible problems with interline twitter. This aliasing effect only shows up under very particular circumstances. Videographers can prevent it by avoiding subjects with fine dark and light horizontal stripes like a striped shirt. The problem is miniscule in the big picture... not worth the hype it is getting in this article.

Fourth, although conversion between interlace and progressive (or vice versa) introduces artifacts (due to the fact that alternate fields of interlaced video are not time-aligned with each frame of progressive scan video), this isn't a "fault" of interlace. Both formats are equally at fault for not matching the other perfectly. Interlaced video is designed to stay interlaced forever, right through to the display. So is progressive scan video. I realize that the computer world is a progressive scan world... you are reading this on a progressive scan monitor right now, and you may even be a video nut like me who often edits video on your PC. But I like videotaping football games, and I realize that my best option right now is to use interlaced video to shoot these games. The action wouldn't look nearly as good with any progressive scan system that I could afford, or that would be practical for me. This is the case for the VAST MAJORITY of videographers in the world. They use interlace... because it works better, for a given price. Would I like to shoot football games in 720P60? Sure! I'd love to. How much are those camcorders? Well, true native progressive scan camcorders are just becoming available... starting at around $9000. So how do I distribute these highlight videos to all the players on the team, and their parents? HD DVD? Blu-Ray Disc? WMV-HD? Oh wait... they don't have any way to watch these formats hooked up to their TV.

Fifth, let's talk exposure time. As each field is exposed, objects in motion (or the camera's panning or zooming) can cause "smear"... motion blur. In low-light situations you want a longer exposure time, in order to minimize noise. In an interlaced camera in low-light situations the exposure time can be brought down to 1/60 of a second. A progressive scan camera could use an exposure time down to 1/30 of a second, but as any photographer will tell you, 1/30 of a second is definitely too slow for anything that is moving (images are blurred). So what happens if you shoot at 30P instead of 60i with an exposure time of 1/60 of a second? You get motion stutter... it's not as smooth. This is because the scene is being sampled for 1/60 of a second every 1/30 of a second. For instance, a hockey puck moving quickly across the frame would appear as a series of dashes, rather than a smooth moving object. Sports videographers will tell you - you can't use 24P or 30P... it's either 30i or 60P.

Sixth, there is the issue of bandwidth. Refer to [2]. There are all kinds of considerations when you design video systems, and interlace was not just a good idea back in the 1930s... it is still a good idea given all of the tradeoffs that are inevitably a part of designing a video system. Of course, 60P requires twice the bandwidth of 30i for a given spatial resolution. In the world of HDTV, the choice is between 1280x720 60P or 1920x1080 30i. You can get twice the spatial resolution with interlace, for a given bandwidth. These are the kind of real-world tradeoffs that everyone must make. So, you do the smart thing... you use your EYES, and see what looks best. Of course, your eyes may not agree with my eyes, or someone elses... video quality is a subjective thing. We are always going to have to make this tradeoff when we balance how we use available bandwidth. More display resolution... a higher frame rate... more bits/pixel... less lossy compression... interlace...? You say bandwidth is plentiful... why is this an issue? The uncompressed bandwidth of a 1080i60 video signal is about 1.485 gigabits per second. The signal comes to you in the US (over the air) at around 19 megabits per second, or up to 38 megabits per second via cable TV. The chroma is sub-sampled, and then MPEG-2 is used to compress each field spatially, and groups of pictures temporally. And you were worried about a little bit of vertical low-pass filtering to avoid interline twitter? Trust me... it pales in comparison to the compression that is needed to bring you your digital HDTV signal. Once more... interlace doesn't decrease resolution... interlace is what enables 1920x1080 instead of 1280x720. It buys you increased spatial resolution at the same field rate and overall bandwidth of a progressive scan signal... or vastly improved motion quality if you were considering 30P versus 30i.

Lastly - back to point #1... the accuracy of spatial detail in a moving picture. Remember... the eye and brain average the picture. 1080i HDTV is specified to have 1920x1080 pixel resolution. Just because there are 2 fields per frame, and fields are displayed alternately doesn't mean that 1080i has any less resolution than 1080p. The eye and brain make a "moving average" out of the interlaced fields. Not just the first top field and bottom field, but the first bottom field with the second top field, and so on. This moving average technique is pretty slick... it works great. Sure... the guy trying to sell you a full 1080p HDTV monitor will tell you lots of stories... except the fact that there are no 1080p broadcasts, and almost no 1080p content to be had anywhere. Sure... some day soon we will be able to see 1080p24 movies... this doesn't count... it's not video, per-se... it's film that has been transferred to video. Film shot at 24p will always look best at 24p. So, back to broadcast video. 1080i60 has all the resolution and twice the motion sampling frequency of 1080p30... for the same bandwidth. Let's say you are watching a hockey game, and someone shoots the puck across the ice. Would you rather the puck be sampled 60 times per second, or 30? Which picture will have more detail? Broadcasters could easily be transmitting these programs in 1080p30, but they choose 1080i60... for a reason. Tvaughan1 03:33, 6 October 2006 (UTC)

interlace comparison

Tvaughan1, you are making a dishonest comparison when you claim that an interlaced signal has both more detail and better motion rendition then progressive. In fact, interlace is a compromise, sacrificing one to gain the other.

I never said that interlaced has both more detail and better motion rendition at the same time... read more carefully. Tvaughan1 02:53, 21 October 2006 (UTC)
You need to _write_ more carefully. You list these two advantages one after the other with no hint that utilizing one negates the other. Good writing is not about what you can get away with, but what the uninformed reader will understand when he or she reads your text. Your writing leaves the reader with no choice but to assume something that you admit not to be true. You've repeatedly criticized me for misrepresenting what you say - but this stuff IS in what you wrote, whether you realize it or not.Algr 20:49, 23 October 2006 (UTC)

The problem is that in mid thought you switch what you are comparing interlace to. With identical bandwidth, you could have 480/60i, 480/30p, or 240/60p. When you claim that interlace has better motion rendition, you can only be comparing 480/60i with 480/30p. But in that comparison, the interlaced signal has WORSE detail then the progressive one. Similarly, when you claim that interlace has more detail, you must be comparing 480/60i with 240/60p. But again, 240/60p has better motion rendition then 480/60i because their is no line crawl or twitter. In either comparison, interlace LOOSES to progressive in either detail or motion rendition. So it is dishonest for you to claim that interlace wins both ways.

Your understanding of the "loss of detail" due to interlace is exaggerated and flawed. There is NO loss of horizontal resolution, and very minimal loss of vertical resolution (no loss, in most cases). Tvaughan1 02:53, 21 October 2006 (UTC)

When you move the discussion to HDTV, you have also thrown in a false premise by insisting that analog bandwidth be equal. In the history section it makes sense to discuss bandwidth, but HDTV transmissions are digitally transmitted, so there is no such limit to analog bandwidth. The rules are totally different: If your digital format is Flash, you could transmit 1080x1920 animations over a 56k modem in real time. Interlaced signals convey no benefit at all to progressive displays, which are now the standard for HDTV. Again you try to have it both ways by assuming that movies will be transmitted in 720p (thus sacrificing detail to gain motion that they don't use) and sports will use 1080/30p. (Thus sacrificing motion that they do need.) But simply reverse the two progressive standards above and you have the best solution for each kind of content. Of course progressive will look worse if you go out of your way to use the wrong standard for a given task, but that is hardly an unbiased comparison, is it?

The only way to make a fair comparison between 2 video formats is to compare similar alternatives... to level the playing field. Would you compare DV to DVCPro HD? Of course not. With similar bandwidth, you have 2 viable alternatives. I never claimed that interlace conveys a benefit to progressive material such as programs that were shot on film, nor did I claim that interlace video signals provide any benefit to progressive displays. It doesn't. You are also misunderstanding how MPEG-2 compression works. It doesn't take much bandwidth to say "repeat the previous frame" (after 24P material is converted to 1080i60). 24P film material is very efficiently encoded as 1080i60. Tvaughan1 02:53, 21 October 2006 (UTC)

And finally, I have repeatedly given you four references showing that interlace looks nowhere near as sharp as progressive with the same number of pixels, and yet you keep insisting that interlace means twice as many pixels means twice as much detail. Interlaced signals yield no benefit whatsoever to anything shot on film, or in film-look video, because 1080/24p uses less bandwidth, and is also much sharper then 1080/60i . Even with sports content, their are plenty of people who say that 720/60p looks better then 1080/60i.

Your references are either totally non-authoritative (some guy's website... ), or you are misunderstanding the references you are quoting. Your assertion that interlace looks nowhere near as sharp as progressive with the same number of pixels is incorrect, and it illustrates how you continue to totally miss the point of interlace. Because interlace cuts the bandwidth in half (for a given spatial resolution and refresh rate), we can use this bandwidth for something... like increasing the spatial resolution (from 720p to 1080i resolution, for instance). Tvaughan1 02:53, 21 October 2006 (UTC)
Tvaughan1, you haven't provided a single reference stating that 1080i looks anywhere near as sharp as 1080p. You keep spinning around in the same circle: You site pixel count to prove that sharpness is only determined by pixel count. Algr 20:49, 23 October 2006 (UTC)

Finally, your alteration of my animation is based on this same dishonest comparison, and nonsensical insistence on equal bandwidth when the point of digital compression is to render analog bandwidth irrelevant. Do you even know how bandwidth works in a plasma display? It is radically different then in CRTs, the signal has many hundreds of "frames" per second, with many tracks running in parallel. Algr 16:39, 20 October 2006 (UTC)

What?!!! "the point of digital compression is to render analog bandwidth irrelevant?" This is just wacky. It makes no sense at all... nor does a discussion of how plasma displays work. Tvaughan1 02:53, 21 October 2006 (UTC)
This is why you fail. Interlace was king in the analog days, but that era is over and you haven't kept up. Why tie yourself in knots to preserve analog bandwidth when a 1024p (SXGA) display sells for $99? Bandwidth (even digital bandwidth) is NOT what limits the sharpness today's displays. Coding efficiency and screen hardware density are today's challenges. Algr 20:49, 23 October 2006 (UTC)
Algr - you are misreading and misquoting me. I never said any of the things that you say that I said. Your animation is totally wrong... not illustrative of interlace... it illustrates flicker, which interlace reduces. It is misleading, as was all of the discussion comparing interlace to MPEG compression... they are 2 totally different concepts, and shouldn't be compared. Are you more interested in owning this article or improving it? Do you own a single decent book on video? Do you own a decent video camera? Have you ever shot or edited progressive scan video? Are you, or have you ever been a member of the NAB, SMPTE, or SCTE? Have you ever worked in the broadcast television or film industry? Tvaughan1 02:53, 21 October 2006 (UTC)
No, yes, yes, yes, no, yes. Algr 20:49, 23 October 2006 (UTC)

I'm sorry but repeated edits like this one are unacceptable POV-pushing. Tvaughan1, I like a lot of the stuff you have to contribute but I think you should cool it. The way you've conducted the ... um, extreeeeeeeeeeeeeemely long debate on this talk page is, unfortunately, not productive. Neither is it productive to reverse the meaning of an entire section by removing everything you don't agree with and replacing it with stuff you do. Nothing personal. More important than whether you're right or wrong, these tactics are not going to improve the article. Snacky 03:37, 21 October 2006 (UTC)

Snacky - did you read what I contributed before you just reverted it wholesale? I spent a great deal of time working on improving the article. Before I edited the article I posted a great deal to the discussion page, then waited a couple of weeks for comments. I didn't reverse the meaning of an entire section at all... I broke the section up into 2 sections - advantages and disadvantages of interlace. The paragraphs I deleted consist of a very confusing and inappropriate comparison of MPEG compression and interlace. The 2 methods are totally different concepts. Although there are effects on MPEG due to interlace, comparing interlace to MPEG is totally inappropriate for this article. Does MPEG allow you to improve the refresh rate, or the spatial detail? Tvaughan1 18:00, 21 October 2006 (UTC)

Neutral point of view?

Interlace is a technique that has received wide support by television experts, even with the development of HDTV. Broadcasters have a choice between interlace and progressive scan formats, and at least half choose interlace. This article doesn't describe interlace accurately. The interlace gif image is a terrible illustration... the "interlace" portion is flickering, and interlace doesn't introduce flicker, it eliminates it. None of the advantages of interlace are described in the article, and attempts to describe the advantages have been reverted. The concept of interlace is fairly technical, and obviously not well understood. This article doesn't help anyone coming to Wikipedia for an understanding of the concept... rather, it reinforces misconceptions about interlaced video. Tvaughan1 02:04, 23 October 2006 (UTC)

Yes, the article (and even this talk page) is confusing. Please excuse if my non native English in not 100% clear, but I'll try to explain my point of view in simple terms!
Interlace is a technique and as such not good or bad.
The confusion comes from mixing a image capture technique with the way it is displayed.
Why not separate the article into Interlace -> capture and Interlace -> display ?
Because the advantages and problems of each are different. Examples:
"when displaying 576p25 from a DVD (for example a PAL movie) on a 576i50 CRT TV is there any disadvantage on having a interlaced display?" Here the issue would be the interlaced display!
"when displaying 576i50 from a DVD (for example a PAL video) on a 720p50 display is there any disadvantage on having a interlaced source signal?" Here the issue would be the interlaced source! Ricnun 14:56, 27 October 2006 (UTC)

Ricnun - Thank you for your thoughts on this article. I think you said it best... "Interlace is a technique and as such not good or bad." Still, given the circumstances of each person or company's situation, the choice must often be made between using an interlaced video format or a progressive scan format for capturing, producing, transmitting or displaying video content. Unless this article (Interlace) is well organized and well written, it does a disservice to those who come to Wikipedia to improve their understanding of the concept of interlaced video. The goal of Wikipedians should be to improve understanding, not to push a particular point of view. The article shouldn't be "for interlace" or "against interlace", it should thoughtfully describe the technique and all of the benefits, drawbacks, and considerations. The technique of interlace should be understandable as a single concept. Display of interlaced video is very similar to the capture of interlaced video, so separating the description of interlace into 2 articles would not be helpful, in my opinion. However, the (Interlace) article should be organized in a way that describes all of the aspects of interlace in different sections - much better than it does now. Tvaughan1 17:00, 28 October 2006 (UTC)

Bandwidth

Interlace is still a useful and relevant technique today because bandwidth is nearly always limited. Not at the display, but everywhere prior to the display.

Bandwidth is a prime consideration for broadcasters of all types. Perhaps some day in the future, with fiber to the premises and fully switched networks bandwidth will be less of a consideration, but cable TV operators, over the air broadcasters and satellite networks all consider bandwidth to be a limited resource. A cable TV operator (Multi-system operator, or MSO) understands that their hybrid fiber / coax system can carry a limited number of channels; typically about 100 channels, or more. The more channels that a cable system can carry, the more revenue a cable operator can generate... through premium channels, video on demand, and different tiers of subscription rates.

Bandwidth is also an important consideration when designing a video camera, or a video signal format (that is stored to tape, magnetic, or optical media). The higher the bandwidth required, the more expensive the system. Interlace is a valuable technique to improve video quality, for a given cost and bandwidth. For instance, interlace allows a 1920x1080 pixel signal to be broadcast with the same refresh rate (60 Hz) as a 1280x720 pixel signal. While some would argue that 720p has a better picture quality than 1080i, more than half the broadcast television industry has chosen 1080i over 720p. The truth is, when you are watching your HDTV program at home - MPEG compression causes the "real" resolution of both of these formats to be somewhat less than the theoretical resolution suggested by the pixel count. MPEG compression is a separate issue, however. HDTV can be recorded in full, uncompressed resolution by professionals using the latest equipment. Full 1080i60 has twice the spatial resolution of 720p60.

A valid comparison of interlace versus progressive can only be made when comparing systems of equal bandwidth. Everyone knows that 1080p60 is better than 1080i60, as long as 1080p60 has twice the bandwidth (or 480p with twice the bandwidth is better than 480i). Because bandwidth is a precious commodity to equipment designers/sellers and cable system operators, interlace is still relevant, useful, and often preferred.Tvaughan1 04:09, 24 October 2006 (UTC)

The real problem with interlace

There are only 3 real problems with interlace.

First, interlaced video is designed to be captured, stored, transmitted and displayed on video equipment specifically designed for interlaced video. Whenever interlaced video needs to be converted or displayed in a progressive scan format, the quality will suffer. Depending on how the conversion and display is handled, the result will either be motion artifacts (if 2 interlaced fields are displayed at the same time), or a reduction in resolution. Up to half the resolution can be lost if the simple technique of displaying only alternate fields is used. If a sophisticated deinterlacing system is used, the resulting video will have a small reduction in resolution.

Second, interline twitter can occur. This is not normally a real issue, and it can be safely ignored for most practical purposes. There is no "mandatory" pre-filtering of the vertical resolution, and there is never a reduction in horizontal resolution associated with interlace.

Third, if you attempt to show interlaced video in slow motion without first deinterlacing, motion artifacts will be visible. This is only an issue if you wish to show normal 60 Hz interlaced video in slow motion. This is why professionals use high frame rate progressive scan systems to capture action that will later be viewed in slow motion.

By far the largest problem is the first issue. Manufacturers and consumers have fallen in love with "flat screen" television displays, which most often are native progressive scan displays. If you own a progressive scan monitor, you will never prefer interlaced video, because it will always have to be deinterlaced (this is handled in the monitor), and the resulting quality will suffer some degradation. It seems clear that some of the Wikipedians who have been actively contributing to this article are biased in favor of progressive scan for this reason... they prefer progressive scan displays for their own, perfectly valid reasons. Perhaps they even own a big-screen progressive scan television, or two. This is a good personal reason to prefer that television companies produce their content in a progressive scan video format... it will look better on your TV. However, the resulting disparaging comments towards interlaced video ("Interlace was king in the analog days, but that era is over and you haven't kept up.") is clearly not a valid, neutral point of view based on the real merits of the two alternatives. Tvaughan1 04:09, 24 October 2006 (UTC)

Interlace and MPEG - 2 different techniques

Interlace and MPEG should not be discussed as alternatives, as they are in the current version of the article. Interlace is not simply a compression technique... it is a display technique. Yes, interlace cuts the bandwidth in half for a given spatial resolution and refresh rate. But you can't compare interlace to the equivalent progressive scan format with twice the bandwidth... this isn't valid. Interlace is used in order to double the refresh rate, or double the spatial frequency. In other words, first we cut the bandwidth in half, then we use the bandwidth we saved to accomplish something.

MPEG is a digital compression technique, interlace is not. MPEG provides both spatial and temporal compression (commonly).

MPEG is not an alternative to interlace, progressive scan is. MPEG compression can handle field encoding of interlaced video very efficiently. Because MPEG provides temporal compression, there isn't a great deal of difference in encoding efficiency between interlaced and progressive scan video (of a given spatial resolution). While MPEG compression is affected by the choice of the native video signal format (interlaced or progressive scan), a discussion of how MPEG compresses interlaced signals versus progressive scan is more appropriate for another article, not this article.

The point is - Interlace is a display technique, not a compression format. The discussion of interlace versus MPEG, or effects on MPEG compression are not on-topic for this article. Tvaughan1 04:09, 24 October 2006 (UTC)

The EBU position on Interlace and HDTV transmission

I am not a spokesman for the EBU, but at the IBC media conference in Amsterdam in September 2006 they gave a very convincing demonstration that 720p50 is noticeably superior to 1080i25 for direct to home transmission. Whilst many of the arguments in this chapter are valid when watching studio or low compression (high bit rate) HDTV pictures, the reality of HDTV transmission is that it is going to be largely commercial, and the pressure is (paradoxically perhaps) to compress and therefore compromise the quality in order to squeeze in more channels into the available bandwidth. Accountants and revenue rule this business. The demonstration showed that 720p50 pictures give far better quality at lower bit rates than 1080i25. Thus for the same quality level you get more channels using 720p. The tests were rigorous and fair. The conclusion was clear.

Incidentally, it is difficult to disagree with the EBU position expressed for several years, that interlace has no place in a world where progressive displays are replacing CRTs. Equally it is difficult to understand why broadcasters are opting for 1080i transmission. The whole legacy philosophy (make the new compatible with the old) is a real stone around the neck of advancing technologies.

Since all modern displays (computer, LCD, Plasma, DLP) are inherently progressive, and convert interlaced transmissions to progressive before display, just what is the point of interlaced transmission?

Good quality conversion of progressive to interlace is relatively easy, but interlace to progressive requires more complex (=expensive) techniques to do well.

For more information, go to the sites below. The demonstration was given by the senior engineering personnel of the European Broadcasting Union in the exhibition hall, not the conference. It used AVC MPEG4 compression codecs on large identical side by side 1920 pixel plasma displays. The pictures were critical and varied, and viewing was close. There is more detail at http://www.ibc.org/cgi-bin/ibc_dailynews_cms.cgi?db_id=23146&issue=2. The issues are fully covered in the technical sections at ebu.ch, and probably in the trade magazines (e.g. tvbroadcastengineering and TVBEurope)

The EBU conclusions include 1080p50 as the preferred format for production. Much publically declared development by the manufacturers and researchers will probably make this a reality. The EBU state that, pragmatically, 720p50 is best for transmission in compressed channels at the moment. Whilst accepting 1080i for European broadcast, they do not recommend it for production and transmission. They hope that 1080p50 will be a broadcast format in the future as well. How you square that with the accountants they don't discuss.


For information, the EBU is the European Broadcasting Union. Most European public broadcasters are members. They are very active in setting European and world wide technical standards for broadcast television and radio.

Tony Gardner Gardnan 08:56, 2 November 2006 (UTC)

The EBU article and concludions are available by going to http://www.ebu.ch/en/technical/trev/trev_frameset-index.html, select issue 308. More details at http://www.ebu.ch and http://www.ibc.org —The preceding unsigned comment was added by Gardnan (talkcontribs) .

Please sign your posts (4 tilde characters at the end will automatically add your user id, date and time). Tvaughan1 18:02, 27 October 2006 (UTC)
I'd be interested to read the reports that you mention. Do you have a direct link to a paper or conference report? I couldn't find anything on these websites. Who gave this presentation? Tvaughan1 18:02, 27 October 2006 (UTC)
Certainly the different frame and field rates of the PAL world are a significant factor that must be considered. Progressive scan at 60 Hz requires 20% more bandwidth than 50Hz (before MPEG compression). I would have to check into this further, but I would guess that the MPEG compression picture sequence would be quite different for 1080i25 and 1080i30, with more predicted (P) and bi-directionally predicted (B) frames in a group of pictures for 1080i30. In any case, the quality of MPEG compression can vary widely, depending on the implementation. I have personally tested a wide variety of MPEG encoders (for DVD, mainly), and I have seen a wide variety of results, depending on the bitrates, the source material, the encoding settings, and the model/version of the encoder. MPEG-2 uses an algorithm that can be implemented with many different settings and methods, and some encoders are clearly better than others when it comes to determining motion vectors and allocating bits. In any case, the effect of interlace on MPEG encoding efficiency is a complex topic also... but I think it should be discussed in a separate article, not in the article that describes interlace. Tvaughan1 18:02, 27 October 2006 (UTC)
It is certainly a valid point that broadcasters should understand what types of signal transmission standards will work best for their intended audience. If most of their customers are using progressive scan monitors, this would be a valid point to consider when choosing a broadcast standard. This is probably not the case today, in terms of the installed base of primary television receivers in homes... yet a majority of television receivers sold this year will be native progressive scan, and so the installed base of displays is changing. Still, it isn't safe to conclude that all TV displays will be progressive scan in the future. Some formats, such as CRT can be configured to operate with interlace or progressive scan. Some newer display technologies will also be able to handle either form. This is a valid issue to consider, but not a reason for the "interlace" article to have a great deal of discussion trashing the technique of interlace. The article in its current form does not have a Neutral Point Of View... there is a great deal of explanation as to why interlace is bad, but precious little on the merits of interlace. And there are a number of points that are confusing and misleading, such as the sections that compare interlace to MPEG, or describe interlace as unneeded due to lossy digital compression techniques. I'd like to see the article cleaned up and focused on the technique of interlace. Tvaughan1 18:02, 27 October 2006 (UTC)
  • Comparing interlace with video compression is perfectly valid because they both do the same thing - increase the number of pixels/scan lines possible at a given data rate, and they both have the same drawback - pixels/scan lines can't always be used to their theoretical potential. Algr

Interlace is a scanning method, not a compression technology. Fundamentally, interlace provides no compression at all. You either use it or you don't. Tvaughan1 22:52, 27 October 2006 (UTC)

  • Except for a single line in the introduction, there is no criticism of interlace at all until the second half of the article, and this starts with older computers being unable to generate a good interlaced signal. This hardly seems unbiased to me. (BTW, someone botched up the history when they moved it to England - I'll go fix it.) Algr

Why is a whole section devoted to the problems caused by interlace, but there is no corresponding section devoted to the benefits of interlace? Tvaughan1 22:52, 27 October 2006 (UTC)

  • "Some newer display technologies will also be able to handle either form." Such as? HDTV sales only started getting big in the last two years or so when progressive scan took over. I don't mean that as a cause and effect relationship, (price and quality had more to do with it) but it makes it likely that progressive HD sets already outnumber tubes in homes, and will continue to further dominate. There DOES seem to be some technological problem with scanning 720p on large screen tubes, since so few of them did it. (Whereas 17" PC monitors easily reach 1080p and have no problem with switching scan rates.)Algr 20:59, 27 October 2006 (UTC)

Laser HDTV and SED show a lot of promise, and they could easily replace LCD and Plasma if this promise is met. It would be trivial to have these displays operate in either progressive or interlace modes, just like a CRT can. My 34" CRT television has no problems displaying 720P. Just because some display technologies can't display interlaced video properly doesn't mean that interlace is not a valuable technique (actually, that fact should be considered more of a "drawback of LCD or Plasma" than a "drawback of interlace"). If most consumers displays can't display interlaced video decently, it would certainly be a factor that broadcasters should consider... but it still would not mean that interlace has no potential benefits when considered theoretically, apart from practical considerations of other techniques and technologies. This article should focus purely on interlace as a display technique... benefits, drawbacks, effects on other aspects of video systems. Practical effects are worth mentioning, but as an aside, not in the main discussion of the theory of the technique. Tvaughan1 22:52, 27 October 2006 (UTC)

This is NOT what interlaced video looks like

| 

or

| 

Interlaced video doesn't look like this. What this image demonstrates best is a phenomenon known as flicker, which interlace essentially eliminates. Anyone looking at this image would conclude that interlace has greater flicker than progressive scan. Take a look at any standard CRT television, and then look at this image... an interlaced video signal doesn't look like this. The brightness and color of the progressive scan side of the image don't match the interlace side, and the progressive scan side doesn't flicker at all (thanks to the high refresh rate of the computer monitor you are using). The image is misleading and unhelpful. You simply can't "represent" interlaced video on a progressive scan monitor (which is what nearly everyone in the world viewing this image is using). It should be deleted from the article. Tvaughan1 22:23, 27 October 2006 (UTC)

I would say the illustration is, in some ways, representative of the “twitter” problem in thin horizontal details, which is caused by interlacing. (I have written more about this problem above – in the section where the Amiga computers and their interlaced, tv-compatible video modes are discussed.) Nonetheless, there are several problems with it:
  • There is no reliable way to make a GIF animation cycle at exactly the same rate at which interlaced video would actually be refreshed on a tv screen. (And even if there was, which rate would that be? The PAL rate or the NTSC rate?)
  • Even if there was a reliable way of representing the animation at the correct refresh rate, there is usually a mismatch between the refresh rate of a computer monitor (PC graphics mode) and the refresh rates of tv standards. This mismatch will cause aliasing problems in the temporal dimension: at regular intervals, the first or the second frame of the animation will be shown a longer or shorter duration than the preceding frame (essentially, sometimes one of the frames in the animation will be duplicated, sometimes one of them will be omitted.) In real interlaced video, each field takes the same amount of time, and no fields are duplicated or missing.
  • Browsers do not usually bother synchronizing their screen drawing operations to vertical blank. As this kind of synchronization is missing, the frames in the GIF animation will regularly be changed in the middle of a screen refresh, and in the worst case, in the middle of the GIF picture itself. This will result in tearing. (The current animation displays randomly appearing flashing horizontal lines on my CRT-based monitor, but if there actually was motion in the animation, it would be much more prominent.) This kind of tearing is not present in actual interlaced video.
  • Since a GIF animation can neither control the refresh rate on the viewer's screen nor synchronize to it, the comparison to the “progressively-scanned” image on the left is skewed. On a CRT-based display, the “progressively-scanned” image will appear at user's preferred refresh rate, which may be way higher than the rate at which the GIF animation runs, or way higher than the refresh rates at which tv systems work. (On an LCD display, the whole concept of a “refresh rate” is muddier, and would have to be simulated for the “progressively-scanned” sample image as well, which is probably impossible.)
  • Even if you could synchronize the GIF animation to the screen refresh rate, double-buffer it, and make the computer refresh the screen at exactly the same rate as a tv screen does, different CRT monitors have different amounts of afterglow. Modern computer CRTs generally use faster phosphors than TV CRTs, which makes the image more flickery-looking with low (tv-alike) refresh rates. Old PC VGA monitors, on the other hand, have noticeable trailing/afterglow for all moving objects, such as a mouse pointer, which is not like a tv, either. It is pretty much impossible to simulate (or prevent) these kind of things. Yet they will affect to how flickery the image is seen as.
  • It is not made clear to the viewer which kind of progressive system the interlaced sample animation is being compared to. For example, it is troubling that there is no display of a same-bandwith, same-refresh rate progressive system with half the vertical resolution and larger gaps between the scanlines. If Ballard's patent (U.S. patent 2,152,234, page 3, lines 26 to 42 in the first column) is to be trusted, interlacing was originally deviced as a way to boost the refresh rate and the vertical resolution of an existing TV system without requiring more bandwidth. While it is possible to make a reverse comparison, starting from a progressively-scanned full-resolution twice-the-bandwith system, and creating a half-the-bandwidth interlaced system out of that, the patent does not seem to represent the idea that way.
  • The sample image itself has pathological features which are not typically present in real TV signals, such as single-scanline tall horizontal details. It is good to make a mention of these kind of features being problematic for interlaced systems, but an image with those problems is not a typical interlaced image. The filtered (anti-aliased) version on the right better represents natural interlaced images as shot with a video camera and seen on tv, but it should be mentioned that images like the non-filtered version are pathological cases rarely encountered in the wild, at least as far as normal television goes.
It should be clearly pointed out in the article that the whole image is only an inaccurate simulation; a mock-up that may represent some aspects of an interlaced image with some accuracy, but not the others. The animation is not totally useless, in my opinion, but as of now, misleading. The main flaw of the image is that there are always three different progressive scan systems that are related to a given interlaced system, but only one of them is illustrated in the picture – and even that one is displayed in most viewing environments as a disproportionally stable version when compared to the slow-mo refresh rate of the “interlaced” mock-up, without mentioning this on the page.
It should also be mentioned that the current image mainly represents the twitter effect, and even that in slow motion when compared to the real refresh rates on a real tv screen. (Slow motion is not necessarily a bad thing in itself, and perhaps a really slow-motion animation of interlaced and progressive scanning on a CRT screen could be useful as well, with a simulation of fading phosphors and all? It might take some time and effort to create that kind of animation, though.)
The animation could also be more useful if it illustrated all three related progressive systems for comparison, and if there was some motion in the picture, so that the clear difference between an interlaced system and the same-bandwidth, same-vertical-resolution progressive scan system could be seen. (Smooth motion in interlaced system, half-the-temporal-rate motion in the progressive system.)
Last but not least, perhaps the animation could be recreated as a short, repeating or cyclical segment from actual, natural video – one shot with a video camera – but with the slow-mo field drawing simulation. It could still be cropped to a tiny size. — Jukka Aho 06:46, 31 October 2006 (UTC)
The Interlace illustration NEEDS to be in slow motion in order to make it's subject clear. Even if it was possible to run the animation at 50 or 60 hz, it would be harder to see the subject. Models of atoms are rarely life-sized either. The purpose of the central image with it's single line details is to help explain why interlace causes a loss of detail - a critical fact of the article. If you don't blur these details out, they will flicker. My camcorders, (Sony VX2100) do not filter details when used in still-photo mode, and so I see obvious flicker in almost every shot that is in focus. Panasonic's AG-DVX100A cameras describe this as having "thick" and "thin" lines - Thick is for normal video, (what any video camera does) and thin has more vertical detail, but flickers without further processing such as printing to film.

Comparing interlace to progressive

The main flaw of the image is that there are always three different progressive scan systems that are related to a given interlaced system I'm not sure what you mean by this. Elaborate?Algr 10:29, 31 October 2006 (UTC)

Take, for example, the 625-line 50 Hz PAL system. The related systems are as follows:
  1. The original interlaced system:
    • 312.5-line interlaced fields
    • 50 Hz vertical refresh rate
    • 15.625 kHz line rate
  2. Progressive variant A – the same bandwidth, (nearly) the same refresh rate, half the vertical resolution:
    • 312-line progressive frames
    • ~50 Hz vertical refresh rate
    • 15.625 kHz line rate
  3. Progressive variant B – the same bandwidth, half the refresh rate, the same vertical resolution
    • 625-line progressive frames
    • 25 Hz vertical refresh rate
    • 15.625 kHz line rate
  4. Progressive variant C – twice the bandwidth of the original interlaced system, the same refresh rate, the same vertical resolution:
    • 625-line progressive frames
    • 50 Hz vertical refresh rate
    • 31.250 kHz line rate
These are not only theoretical constructions. Variant A is typically used by old game consoles, 8-bit home computers, and the like. Variant B is kind of virtual – it does not really exist as a scanning format, but its digital equivalent is used as a storage format: for example, “PAL”-formatted Hollywood movies are typically stored as a 25 fps (essentially progressive, although not necessarily flagged so) MPEG-2 stream on a DVD. Variant C is the “progressive PAL” scanning format (ITU-R BT.1358) — Jukka Aho 13:24, 31 October 2006 (UTC)

I see. You might also want to throw in:

  1. Progressive variant D – the same bandwidth, the same refresh rate, same pixel aspect ratio, 2/3 vertical resolution line count:
    • 416-line progressive frames
    • 50 Hz vertical refresh rate
    • 20.830 kHz line rate

This would be to PAL what 720p is to 1080i. The nearest real-world equivalent would be EGA's 350/60p mode.

Also, you could use the same relationships to produce 4 matching interlaced systems for any proposed progressive one:

480/60p variants:

  • A -960/60i
  • B -480/120i
  • C -480/60i
  • D -720/60i

This is more then just a footnote to an image, you'd need a whole section on it's own to explain it. What do others think? Is this valuable? Algr 17:16, 31 October 2006 (UTC)

Seeing as only the author of the image is able to defend it, and others agree that it has serious problems, I'm going to delete it. I know this will upset Algr... that's not my purpose. The image is located in the "description" section of the article, and it does not help to describe interlace, especially given the fact that another image already graphically illustrates the technique. The image attempts to represent the video that is displayed by an interlaced signal, and it fails in this effort. The image represents interlaced video with twice the luma per visible scan as progressive scan (which is never the case), and with resulting differences in color (chroma)). The image is original work, and unsupported by any referenced sources. A modified version of this illustration might be useful and acceptable to demonstrate interline twitter under the "Problems caused by Interlace" section. Tvaughan1 20:32, 16 November 2006 (UTC)
No one "agreed it had serious problems", we just said that the comparison is complex and you need to be clear on what you are comparing. The illustration is simply "Variant C" listed above, which is the same as comparing NTSC to VGA. This to me is the most natural comparison because it includes two real systems that viewers today might have in their homes, and since both advertise "480 scan lines", we could say that all else but the presence of interlace is equal. Algr 06:51, 17 November 2006 (UTC)
Algr - who besides you has claimed that this illustration is acceptable for Wikipedia? A number of other Wikipedians have described the image as flawed and misleading. Your illustration is original work, uncited, and not from a NPOV. It doesn't describe interlace - it describes your point of view of interlace. Tvaughan1 15:46, 17 November 2006 (UTC)
I addressed Jukka Aho's concern with the added text. Who else are you talking about? You are the one who is POV pushing by trying to portray 480i as equivalent to 480p, despite overwhelming evidence to the contrary. (How many references have I given you now, six?) No one else except you, Tvaughan1, has recommended it's removal. Algr 17:25, 17 November 2006 (UTC)

It's time to clean up this article!

The entire section entitled "Problems caused by interlacing" describes no problems caused by interlacing. Rather, it compares interlacing to MPEG compression.

The worst offending sections are... For the purpose of reducing the bandwidth necessary to transmit the video-based material, interlacing is inferior to the modern digital block-based compression techniques for the following reasons:

Interlacing should be compared to progressive scan, not to MPEG. The following points show a distinct lack of understanding of how MPEG compression works...
Interlace and MPEG both do exactly the same thing. The only thing that interlace does that MPEG doesn't is reduce the bandwidth needed inside the CRT - but with LCDs everywhere and modern 1035p monitors selling for $99, this is a useless savings. Algr
“Interlace and MPEG both do exactly the same thing”? That's quite a bold statement – and as far as I can see, quite misleading, too, at least if not explained better. It is true that interlace can be seen as a bandwidth-saving method, if (and only if) the reference point is progressive variant “C” (see above) but I can't see any justification for making direct comparisons to MPEG. I also don't see why MPEG(-1/2) should get any special treatment in this article, at least over MPEG-4 (and related formats, such as DivX), Flash video (H.263), DV, M-JPEG, HuffYUV, the Sorenson codec, or a number of other formats.
That said, it is good to mention somewhere in the article the special considerations that field-based material causes for DCT-based compression algorithms, and it would also be good to mention how some video formats have special modes for storing interlaced material (for example, MPEG-2 employs a different DCT block scan pattern for interlaced video, and has special flags for repeating fields or indicating the top of bottom field to the decoder), but all this should either be made briefly and to the point, or if it seems to require a lenghty discussion, separated into its own section, or even into a sub-article of its own. There is no reason why DCT-based compression algorithms should be the primary focus of the “problems” section.
Overall, I feel that this article should, first and foremost, discuss interlacing and its history, benefits and disadvantages as an analog technique (for cameras and displays that "scan" the picture in interlaced fashion) and only then – in later sections – expand into topics such as interlace and digital video formats, interlace and computers, interlace and compression algorithms, interlace and modern progressively-scanned (or non-scanned) displays, etc. — Jukka Aho 15:35, 3 November 2006 (UTC)
   * Interlacing performs poorly on moving images, leading to saw tooth or combing distortion.
Not true at all - interlaced video performs better on moving images than progressive scan at a given spatial resolution and bandwidth. These distortions only appear if interlaced video is displayed on a progressive scan system. Interlaced was never designed to be deinterlaced, or displayed on progressive scan systems.
I can see this distortion easily on 480i sets with most programming, and can always tell by eye if a screen is 480i or 480p. Remember they were selling 480p sets in the early '90s before there were even DVD players to hook up to it. Maybe you don't have an eye for seeing it. Algr
Again, a personal attack. Can we discuss the subject? Your personal observations are not a published source. Interlace performs better on moving images than a progressive scan system of equal bandwidth and spatial resolution, due to interlace having twice the field rate compared with a similar progressive scan system. Tvaughan1 04:07, 2 November 2006 (UTC)
   * Fine horizontal detail is subject to twice as much flicker as the rest of the picture.
The effect that someone tried to describe is technically known as "interline twitter", and NOT flicker. On a properly designed interlaced video system "the rest of the picture" has no visible flicker.
   * Progressive MPEG is flexible and adaptive about which details of the image it compresses and how much, while the compression by interlacing does not discriminate about the perceptual complexity of the element of the image being compressed.
This is just nonsense. Citation? MPEG compression is just as flexible and adaptive for field based sources as frame based sources. In fact, macroblocks can be encoded as either for interlaced video, and after motion vectors are established macro blocks describe the motion of objects across multiple fields or frames.
   * The quality of consumer-grade deinterlacers varies, while the MPEG decoder is absolutely deterministic in regard to the decompression of the progressively compressed stream.
This suggests that interlaced video needs to be deinterlaced. Interlaced video was never designed to be deinterlaced. Comparing deinterlacers to MPEG decoders is nonsense. MPEG decoders are absolutely deterministic in regard to the decompression of interlaced (field based) streams also.
If current sales trends continue, in five years stores might not even have any interlaced sets any more. 1080i is already almost extinct as a display format. Of course you need to deinterlace! —Preceding unsigned comment added by [[User:{{{1}}}|{{{1}}}]] ([[User talk:{{{1}}}|talk]] • [[Special:Contributions/{{{1}}}|contribs]])
This fact is external to the subject of interlace. Interlace is a display technique. The article should describe the technique, the benefits, the drawbacks, etc. Wikipedians can become more informed and they can make their own choices as to what they want to buy. The article shouldn't be based on assumptions about what is necessary or unnecessary, good or bad. Suggesting that interlace is bad because the quality of consumer grade deinterlacers vary is not treating the subject from a neutral point of view. Interlace is obviously not designed to be deinterlaced. If someone knows that the display will be progressive scan, all things being equal one would want to choose a progressive scan camera. Of course, progressive scan cameras are generally much more expensive or inferior in quality at similar price points to interlace video cameras today, due to the bandwidth required by progressive scan systems at good frame rates. I think saw that User:Algr mentioned that he has a Sony DCR-VX2100 camcorder somewhere... an interlaced video camera that is only capable of 15 fps progressive scan (I have the VX2000). Why are progressive scan cameras so much more expensive for a given spatial resolution and frame rate? When you understand this, you understand the value of interlace. The problem that interlace helps to solve isn't on the display side... it's on the capture, storage, and transmission of video. Tvaughan1 04:44, 2 November 2006 (UTC)

The combination of interlacing with block-based compression technique inherits all the drawbacks of the interlacing, while also reducing the efficiency of block-based compression. Because interlacing samples every other line without prefiltering, it increases the amount of high-frequency components in the signal fed to the block transformation. This leads to lower efficiency of block transformation (i.e. DCT), or alternatively increases the amount of artifacts after decompression. This also decreases the effectiveness of the motion compensation technique, used in the interframe compression formats like MPEG.

Citation? Does the writer know how field-based MPEG compression works? Where did this information come from? Was it original work?

When vertical color compression (also called decimation or color subsampling) is included to the combined compression system, it is further effectively compressed by the interlacing. And vertical color subsampling is almost always included into digital and analog television systems (with the exception of broadcast NTSC, and [subject of controversy] broadcast PAL), all over the world. Thus with 4:2:0 color compression technique (i.e. half horizontal and half vertical resolution) the vertical colour resolution drops from 1:2 to 1:4, and overall color resolution from 1:4 to 1:8.

Simply not true at all. There is no such effect from interlace on chroma subsampling. Citation?
With 4:2:0 sampling, each four-pixel square shares a single color value. So if there are 576 progressive lines, the top line shares all it's color data with the line below it, yielding 288 color lines from top to bottom. However in an interlaced system, the "next line" is the third, not the second - the second line would be part of the opposite field, and it's data too far ahead or behind to be worked with. (Remember, no frame stores in the 1960's) As a result, with only 288 lines in a field, you get 144 color lines. Look it up yourself, we have already provided far more citations then you have. Algr 02:49, 2 November 2006 (UTC)
Let's assume a 4:2:0 system with a nominal resolution of 720×576. This system stores the color components as follows:
  • 720×576 pixels for the Y channel
  • 360×288 pixels for the Cb channel
  • 360×288 pixels for the Cr channel
Now, let's assume that we have two 720×576 frames, both subsampled to the 4:2:0 format, as analysed above. One of them is progressive (variant "B"), and another one is interlaced. Could you explain how there is, in your opinion, less data or "less overall resolution" in the interlaced/interleaved frame? (Yes, it is obvious that the individual fields have only one fourth of color resolution when compared to their luminance resolution – but that "one fourth rule" holds true for the whole frame, and over the length of the whole video stream, regardless of whether it contains interlaced or progressive [variant "B"] data.) — Jukka Aho 18:08, 3 November 2006 (UTC)
That is not correct. 4:2:0 chroma subsampling (or 4:2:2 or any other chroma subsampling scheme) has an identical number of chroma samples for interlace or progressive. 4:2:0 subsampling has 1 chroma subsample for every 4 luma samples, regardless of whether you are using interlace or progressive. I'll cite figure 10.2 from page 92 of "Digital Video and HDTV Algorithms and Interfaces" by Charles Poynton, which graphically illustrates these subsampling schemes. Tvaughan1 04:07, 2 November 2006 (UTC)

It is sometimes claimed that combining MPEG compression with interlacing reduces the amount of processing power required from the MPEG decoder almost in half. However, this argument does not stand when faced with the immense processing power needed for unobjectionable deinterlacing of the image after MPEG decompressor; and all modern displays but the (gradually disappearing) CRTs require progressive image as its input.

Citation? Where is this coming from? The first statement describes a possible benefit of interlace, not a "problem caused by interlace", and then the statement is disputed. Why include it in the first place if it isn't true?

Another argument is that combining interlacing with MPEG drives up the overall sweet spot of the compression system. (Note though, that the sweet spot does not get close to doubling, due to the inefficiencies described above.) Specifically, it makes it possible to transmit 1920x1080 60 Hz video over the broadcasting bit pipe chosen for the ATSC system. However, essentially the same effect on the sweet spot without the drawbacks of interlacing could be achieved by simply prefiltering high frequencies out before applying a progressive MPEG compression; or, less efficiently, by filtering out high-frequency components from the compressed MPEG stream right before injecting it into the broadcasting pipe. On the other hand, most DVB flavors (T, S) offer a suitable bit pipe already today, and a better terrestrial broadcasting technology could have been selected for ATSC too.

Again a statement that describes a possible benefit of interlace followed by statements refuting the claim. Again, more comparison of interlace with MPEG compression, which is inappropriate. What is being compared? Interlaced video versus what (specifically)?
The "problems caused by interlace" section should describe problems that occur when using interlace that don't occur when using a progressive scan video signal of similar bandwidth... an "apples to apples" comparison. If there are actual effects of interlace (versus progressive) on MPEG compression, this discussion might be appropriate for a section entitled "effects of interlace on MPEG compression"... but any statements in this section should be verifiable from authoritative sources. Tvaughan1 03:50, 1 November 2006 (UTC)
Tvaughan1, you are arguing in circles here, we have already addressed most of these points above. In this entire discussion you have only ever provided two citations, and one was a subset of something we already had. I've provided six myself, and you just ignore them. If you want to change the article, why don't you do some research yourself instead of just complaining. As I've said before, you have absolutely nothing to justify your insistence that 1080i looks as good as 1080p in any realistic setting. Algr 02:49, 2 November 2006 (UTC)
Arguing? Complaining? Please stop with the personal attacks. I am working on the interlace article, that's all. I suggest you review WP:OWN and WP:AGF. As an engineer with considerable expertise in digital video it is getting a bit frustrating to have valid points shot down with no validity, and to have serious contributions to the article reverted wholesale. If you continue to revert my contributions without any valid discussion or reason I'll have to ask an Admin to arbitrate. Tvaughan1 03:49, 2 November 2006 (UTC)
I don't think this can ratchet up to arbitration right off the bat, but mediation is a likely next step, and if you seek it, I, for one, would gladly submit to the process. I hope Algr would too. I'd like to see what he has to say. Snacky 05:09, 2 November 2006 (UTC)
If this discussion had more participation from wikipedians with some video expertise, it would be easier to arrive at a consensus, point by point. As you have noted, the discussion has been very long, yet incredibly unproductive... characterized by wholesale reverts of any attempts to edit the article. (see WP:OWN). If you look through the entire discussion, you will note a number of folks who have raised concerns on various points. Yet these concerns are shot down by the same person time and again. Who wants to join such a discussion? Very few have the stomach for it. So step up to the bar Snacky... do you have anything to say about any of these issues? Please... any help would be appreciated. Tvaughan1 05:25, 2 November 2006 (UTC)
This "problems caused by interlace" section does a very poor job of describing the problems caused by interlace. The two main problems caused by interlace are interline twitter and motion artifacts (a.k.a. interlace artifacts). Each of these issues should be described adequately and clearly. Interlace should not be compared with MPEG compression, as the two are entirely different concepts (as evidenced by the fact that video is often interlaced and encoded with MPEG). Effects of interlace on MPEG compression are a very complex subject, but not at all well treated in this article at the moment (in fact, they are grossly misunderstood by the article in its present form). Tvaughan1 04:24, 2 November 2006 (UTC)

1080i versus 1080p

Let me concede a point... because it seems to be something that is a real hang up. Is 1080i as sharp as 1080p? While both formats are described by the same SMPTE 274M standard, and both have 1920x1080 pixel resolution, if both are used with the same refresh rate the 1080p picture will have better spatial detail... both horizontally and vertically. Of course, the 1080p signal will have twice the bandwidth, prior to MPEG compression. And, of course, there is no such thing as 1080p broadcast HDTV, due to the bandwidth issue. And, of course, the bandwidth problem is a real issue for video cameras... there aren't any 1080p video cameras to speak of.

Why would 1080p60 have better spatial detail than 1080i30? It has twice the pixels per frame, and therefore motion is represented better (objects are updated 60 times per second with twice the pixels of 1080i for each update). Progressive scan also avoids the interline twitter artifact that can occur with interlace when the subject being shot has detail that approaches the vertical resolution.

Let's be clear. 1080p doesn't have better horizontal spatial resolution than 1080i... it has better horizontal temporal resolution. In other words, more frequent updates of the details of objects in motion. For a fixed camera and stationary objects, there is no difference in spatial resolution of 1080i and 1080p at the same frame rate. 1080p will also have slightly better vertical resolution for 2 reasons... no filtering to avoid interline twitter, and twice the pixels per frame (better motion portrayal). Again, this isn't an apples to apples comparison. The benefits of 1080p come at the cost of twice the bandwidth for a given refresh rate.

So, where is 1080p? Almost non-existent, at this point. But shouldn't 1080p much better for movies? Well, yes... but here you are talking about 1080p24, not 1080p60. If you are ready to invest in an HD DVD or Blu-Ray Disc player with HDMI 1.3 and a 1080p display, you might notice a tiny improvement versus 1080i. These disc formats are the only place you can get a good 1080p source.

So how does this point affect the interlace article? It is hard to describe this, because it is not an apples to apples comparison. If everyone agrees that a progressive scan format is better than interlace at the same resolution and refresh rate, it might be helpful to point this out, but only if it is mentioned that this comes at a cost of twice the bandwidth (before digital compression... the effects of interlace vs. progressive on digital compression is another matter).

Note - the convention for a scanning format should use the frame rate, not the field rate as the final number... 1080i30 has 60 fields per second... according to numerous sources. Tvaughan1 06:24, 2 November 2006 (UTC)

EBU appears to be using a slash character (“/”) between the p or i and the number that indicates the frame rate, such as “720p/50”. Some examples of this usage can be found in these articles:
The latter of the two articles also includes an animation that compares interlaced scanning to progressive scanning. The animation is a bit flawed in that it does not illustrate how the scanlines would fade away on a CRT-based display, how the electron beam would draw them from the left to the right, or how motion would be updated. What is more, the illustration also dumps several lines worth of picture on each of its simulated scanlines! The side-by-side comparison is interesting, nonetheless, since it clearly shows how the interlaced system has already swept all the way to the bottom of the screen once while the progressive scan system – variant “B”, in this illustration – has not advanced any further than in the middle of the screen. (Something like this is what I meant above when I suggested that a slow-motion “scanning” animation of an interlaced system and the relevant progressive variants, side-by-side, could be a useful addition to the article.) — Jukka Aho 23:37, 5 November 2006 (UTC)

content creator view

I was discussing HD vs SD TV with Weta Workshop today as it happens; the comment about future TV producers may not care about extra costs is pretty unlikely. It's market demand and the desire to produce the best possible content within the available budget that dictate what they produce in terms of resolution. The same will apply to interlaced/progressive choices.

212.183.136.194 08:38, 2 November 2006 (UTC)

Do you mean this comment? "This argument will become less relevant in the future with the progress of science and technology, and the TV producers of today or tomorrow may not be very sensitive to the capital costs." I would agree... this is not a fact... it is speculation, and probably not accurate speculation. Tvaughan1 00:20, 3 November 2006 (UTC)

Interlace and compression

Interlace was invented to solve a particular problem of visible flicker on CRT displays. 25Hz updating of CRTs is visible - 50Hz is not, at least in home viewing conditions. Essentially interlace gives twice as many pictures each with half as many lines - 50 X 312.5 line pictures in place of 25 X 625 line pictures.

50 X 625 line pictures (non-interlaced) would also solve flicker but consume twice the bandwidth. In this sense can interlace be described as compression. Modern displays are different to CRTs, and 25Hz progressive updating is acceptable, and produces no flicker. Efficency of MPEG2 and MPEG4 compression of interlaced and non-interlaced pictures is a complex issue, best solved by practical and realistic tests. In the context of direct-to-home HD TV transmission with compression, it seems that progressive is more efficient. Refer to the EBU results discussion above.

Gardnan 23:18, 3 November 2006 (UTC)

Compression implies a reduction in the quantity of data or the data rate. Interlace doesn't reduce the data rate of the video signal... it never captures the alternate lines in each field in the first place. In that sense, one could say that choosing a lower resolution or a lower frame rate is a compression technique. I think that a discussion of the effects of interlace on video compression is appropriate, in a sub-section of the article... but I don't think it is appropriate to refer to interlace as a compression technique. Tvaughan1 04:13, 4 November 2006 (UTC)
It is my understanding that progressive scan video of a given resolution and frame rate is more efficiently encoded than interlaced video... but I don't have any quotable sources on this at the moment (just advice from experts that I know). Does anyone have any good sources on the efficiency of MPEG compression for interlaced versus progressive scan material? Tvaughan1 04:13, 4 November 2006 (UTC)

Semantics: compression and interlace

Both MPEG type compression and interlace are means of maximising the perceived picture quality whilst minimizing the bandwidth. One affects the other - they cannot be separated. Interlace is often cited as an early analogue means of compression. It arose early last century because 25Hz vertical scanning produced visible and disturbing flicker, invisible (in the home) at 50Hz - however the channel bandwidths available did not have sufficient bandwidth for 50 times the full number of lines, hence interlace. Good idea in its time. Whether interlace is called compression or not is a matter of semantics. Idem 422 colour sampling, but thats another story! Gardnan 09:58, 5 November 2006 (UTC)

Interlace isn't compression by any reasonable definition of the word "compression". Compression takes an existing signal, bitstream or file and makes it smaller. Interlace is a raster scanning technique that is used on both acquisition and display. MPEG is not an alternative to interlace. The only alternative to interlace is progressive scan. Tvaughan1 23:15, 5 November 2006 (UTC)
Interlace can be thought as a bandwidth-saving technique, but only if it is supposed that there is some magical “source signal” that exists in the progressive scan variant “C” format, from which the said interlaced signal is derived.
Is it correct to say that all interlaced material originates in the progressive scan variant “C” format, and is then reduced (“compressed”) to a corresponding interlaced signal, using only half the bandwidth of the original signal? In my opinion, no. Interlaced signal, in its purest form, originates directly as interlaced signal in the camera. For example, when shooting interlaced video with a tube camera, or with a rotating scanning Nipkow disc camera, there never was an “original progressive signal” with twice the bandwidth. So in this sense, Tvaughan1 is correct.
Is it correct to say that interlaced scanning gives pictures that, in many ways, resemble a corresponding progressive variant “C” signal, even if the interlaced signal only needs half the bandwidth for creating that illusion? In my opinion, yes, since the nominal/subjective vertical resolution and the refresh rate are the same as in progressive variant “C”.
As has been discussed above, there is a perpetual problem in this discussion, and also in the article itself: time and again interlace is being compared to one of the three corresponding progressive-scan systems (which I took the liberty of naming “A”, “B”, and “C” in the above discussion), but you never really know which one, at least if you're not already familiar with the subject. A writer who is making some random comparison or statement about interlace isn't usually telling which one of the progressive systems he has in his mind, and may even suddenly change his viewpoint in mid-sentence. (For example, if compared to progressive scan system “B”, instead of “C”, interlace cannot be seen as a bandwidth-saving method – in that comparison, it's merely a different way of refreshing the screen. It all depends on your viewpoint and point of reference.)
Whatever interlace is, or is not, “compression” is a bit awkward term for it. “Compression” usually indicates that some kind of reverse operation (“decompression”) will take place at the receiving end. Interlace – as it is traditionally implemented in CRT-based interlacing tv-sets – does not have that kind of reverse “decompression” operation.
My suggestion: whereever this aspect of interlace is discussed in the article, we could settle for calling it a “bandwidth-saving technique”, and then explain it in terms of comparing it to a variant “C” progressive scan signal, while at the same time clearly explaining that...
  1. variant “C” (and not any other variant) is what we have in mind here, and
  2. interlaced signals do not usually originate as variant “C” progressive signals, but directly as interlaced signals (so we're, in a way, comparing interlace to an imaginary format that was never there in the first place when our interlaced signal was made)
  3. sometimes interlaced signals might be derived from variant “C” progressive signals, but that is not the typical case, and
  4. variant “C” is not the only possible directly-related progressive scan system that interlace could be compared to, and
  5. in those other comparisons to the other related progressive-scan formats, interlace cannot really be seen as a bandwidth-saving mechanism.
See? It's a complicated thing, with multiple aspects that all need to be explained. It's not going to be an easy task to explain that all in an understandable and concise way. — Jukka Aho 00:51, 6 November 2006 (UTC)
How did you come to pick those exact numbers – 50 Hz and 25 Hz? At least Ballard's interlace patent (U.S. patent 2,152,234) was not based on the 50 Hz refresh rate, but on a 48 Hz refresh rate. He chose that number to make his system compatible with 24 fps film. (It is all explained in the patent text, see the link for yourself.) — Jukka Aho 01:11, 6 November 2006 (UTC)
The TV field rate in Europe was generally locked to the mains frequency at least until PAL colour was introduced. Mains is 50Hz in Europe. I undertood that this decision was made so that any interference to the picture due to poor power supply smoothing - e;g; hum bars - would be static and less disturbing.
Of course, but if we're discussing the reasons of why interlaced scanning was chosen over any of the possible progressive variants (A, B, or C), surely we can't just assume anything about the refresh rates based on the current TV standards in any one country, since interlace was developed before the current TV standards were devised. Instead, we would need to find out who really originally invented interlaced scanning (see below, I already started a new section about this, since it has been suggested that Randall C. Ballard wasn't necessarily the first), and what was the original tv system it was applied to. Some points to consider:
  1. That original interlaced system may well have been some low-line count system with a lower refresh rate, just like Ballard's patent hints, and not necessarily a 50 Hz or 60 Hz system at all.
  2. Any other interlaced systems that came after that first succesfully implemented interlaced system may already have been designed to be interlaced from the get-go, without even considering a progressive alternative.
If that second points holds true (I'm not saying it does, but it's a possibility) where is the frame of reference then? Is it progressive variant A, B, or C? — Jukka Aho 17:44, 6 November 2006 (UTC)
One consequence was that 24fps cinema film was generally scanned and transmitted at 25fps, making voices higher and films shorter. Film buffs often noticed the latter and complained wrongly that a film had been cut, but rarely were there complaints about pitch.Gardnan 21:50, 7 November 2006 (UTC)


Of course interlace and MPEG are two different things. The definition of compression given by TVaughan and Jukka Aho above assume you have a recording or source of pictures requiring a certain bandwidth, and you then proceed to modify them in such a way as to reduce the bandwidth required.
I argue that TV system designers have or had a choice when standardising the transmission standards. They could have had a 50 progessive standard, made the cameras and telecines work in that, but that required too much bandwidth, so to reduce the required bandwidth they chose interlace.
Both reduce bandwidth compared to a reference standard. For me both are reasonably called compression techniques - so is 4:2:2 colour coding, PAL or NTSC U and V colour bandwidth filtering and some other techniques.
It is semantics and depsends on your starting point.Gardnan 10:11, 6 November 2006 (UTC)
In my opinion, we can't just assume the reasoning that went on in the engineers' heads before the current, popular (interlaced) "standard definition" tv standards were created. Instead of merely assuming these things, we would need to find a reliable historical source where the proceedings and reasoning of the TV standards committees are recorded. To my knowledge, the only such source that has been presented in the current article, is Ballard's patent text. For example, we don't have material that would reference to the proceedings of the (US) National Television Standards Committee (NTSC), or the body that standardized the interlaced 405-line system in the UK. We also have nothing about mechanical Nipkow disk tv sets or scanners/cameras (in Randall's patent, film is scanned for television broadcasts with a Nipkow disk that has an interlaced scanning pattern), and virtually nothing about the interlaced tv standards that were (possibly) in use before those standards, pre-NTSC. The whole “history” section in the article is a bit lacking since it does not have references and there's no clear, verifiable timeline on interlaced tv systems. (It should be noted, however, that one of the linked articles mentions interlace as a resolution-enhancement technique, just like Ballard's patent.) — Jukka Aho

Gardnan 09:57, 6 November 2006 (UTC)

Gardnan - while it seems trivial to argue over the definition of "compression", it goes to the very heart of the matter when it comes to the technique of interlace. So, if you have the patience, I'd like to see if we can reach a consensus on this point. You are correct in saying that TV system designers have a choice when choosing a TV signal standard, and that interlace is chosen over progressive scan in many cases. Choosing to use available bandwidth differently does not constitute "compression". For instance, US broadcasters today have a choice of HDTV standards. Some choose 720P and others choose 1080i... using the same bandwidth. Is 1080i a compressed version of 720p? Of course not. The broadcasters who choose 1080i use the same bandwidth to broadcast a signal with a higher spatial resolution. So, it isn't really just semantics... and yes, it depends on your starting point, or your frame of reference. But one can never assume a starting point or frame of reference... that interlace is chosen over the identical signal format with the same resolution and refresh rate. Interlace provided early TV system designers the opportunity to develop a TV system with a higher refresh rate, and/or a higher display resolution, versus any progressive display standard under consideration. Using available bandwidth differently is not compression. Would you agree? Tvaughan1 14:17, 6 November 2006 (UTC)
Chroma subsampling, however, is definitely a form of compression (generally, multiple color samples are averaged into fewer samples). This is especially evident in professional HDTV camcorders which can provide an uncompressed 4:4:4 signal via HD-SDI, while recording a sub-sampled 4:2:0 or 4:2:2 signal to tape. Tvaughan1 14:17, 6 November 2006 (UTC)

I think we agree on the underlying technical arguments. Frankly whether interlace is called compression or not is irrelevant so long as we recognize that interlace and compression are linked in important ways.

Note though that many modern CCD cameras are native progressive - indeed many HDTV studio cameras process 1080p50 internally. In this case the starting point is clear, and an interface in the camera processes the 1080p50 to make 1080i25 or 720p50. This I would think is compression also under your definition? For me, interlace is an old technique of little value in the new broadcast world, for all the reasons stated elsewhere. If modern cameras and film scanning are inherently progressive, and if modern displays are also inherently progressive, I just cannot see the point of ever in the chain changing the video to interlace? Hopefully third generation HDTV will head towards 1080P50 from beginning to end. Gardnan 12:42, 7 November 2006 (UTC)

Gardnan - yes, if a camera is truly "native progressive", reading all sensor elements for every field or frame, but then processing this as interlaced video, you might call this usage of interlace "compression". It would stand to reason that the camera could provide a higher quality progressive scan signal if the bandwidth to deliver the signal were available throughout the production, storage, and transmission chain. But that is a big "if"... it doesn't seem to be the case so far... progressive scan cameras are few and far between, and very expensive to comparable interlaced video cameras. I also agree that if you know the video will need to be displayed on a progressive scan system you are better off capturing the video in a progressive scan format. Again... a big "if" for television video, since there is still quite a large installed base of native interlaced television displays, and since we can't predict the future of display technology (it isn't safe to assume that either plasma or LCD will become the predominant format in the future, although they have become the dominant formats being sold today). It is easy for people to reach the conclusion that bandwidth is getting cheaper in all areas, and soon it will be cheap enough to simply move to 50 Hz or 60 Hz progressive scan video formats, and be done with interlaced video. But companies are in business to make money, and they don't give away things like bandwidth (even inside a video camera!). Even if the bandwidth to deliver 1080P60 is cheap and plentiful from the camera to the storage, editing, encoding, transmission, decoding and display systems, one could argue that this bandwidth could be used to deliver an alternative that is even higher resolution ( ~1280i30 ... 1440i30 ... ?). So, I'm not sure that the debate will end soon. Tvaughan1 19:51, 7 November 2006 (UTC)

I would argue firstly that all CCD cameras and modern film scanning are inherently progressive and that the interlace signal at the BNC or other connector of a camera is interlace only because the camera has included the necessary interface to output in interlace (and consequently to reduce the quality). The internal working is progressive until a late stage, converted to interlaced (if at all) only at the output stage. Why otherwise?

For bandwith reasons, perhaps? Why process any more data in the camera than is necessary, or any more data than is going to end up on the tape, or being broadcast? I am no expert on CCD chips and the technical details of their use in video cameras, but I have quite often seen it explained that interlaced fields are created already at the readout stage, and – this is what I find interesting – the scanline data is actually generated by combining two CCD sample site lines – one below the other – at a time. Here's one such explanation . — Jukka Aho 01:37, 9 November 2006 (UTC)

The EBU article referred to elsewhere is very relevant. I do not agree that the "ifs" are "big". All VGA and higher standard computer screens interfaces from the PC compatible 1980s machines are progressive. In Europe, it is increasingly difficult to find CRT displays in the general public shops - some major UK chains have formally announced that they don't sell CRT TVs anymore. Everything they sell is progressive, needing extra proessing to convert (note convert) to interlace: what's the point? Who would invent inerlace if it didn't exist at this moment in time, and why? What application? & would it be possible to obtain international and world standardisation which is so essential in broadcast to bridge the theory with the implementation. Another point - I enjoy this sort of debate but do third parties? or are we just adding noise (confusion) to a page which should be concise and clear? Shouldn't we compress our discussion? (or interlace?). Could we continue via private e-mail or elsewhere and not argue in public at least until we understand each other. There is much repetition of statements in the preceding exchanges. Also I think there are bigger issues than the detailed dispute here - a larger view is needed. It concerns me that business economics dominate the technical quality - but it must nevertheless be correctly taken on board. Also a good appreciation of standardisation procedures and why these are necessary before "something" can be sold are essential parts of this discussion and worth concentrating on.Gardnan 22:46, 7 November 2006 (UTC)

History: Who really invented interlaced scanning, anyway?

The article states that the inventor of interlaced scanning was an RCA engineer, Randall C. Ballard. A link to his original patent application is provided as well. (U.S. patent 2,152,234 – it is an interesting read.) But was Ballard really the first? In another discussion, it has been suggested that the inventor of interlace was really Fritz Schröter: ("Verfahren zur Abtastung von Fernsehbildern", DRP-Patent Nr. 574085, Anm. 27.09.1930.) Does anyone here have access to German patents from the 1930s? Fritz Schröter is also mentioned in the Wikipedia article about raster scan. — Jukka Aho 01:30, 6 November 2006 (UTC)

EBU and Interlace

There is a detailed technical discussion of interlace and progressive scanning in the broadcast world on the EBU site http://www.ebu.ch/en/technical/trev/trev_frameset-index.html. Select issue 308, and go to article entitled HDTV — EBU format comparisons at IBC 2006 by Hans Hoffman. The conclusions include the following statements, adequately justified in the text of the article: The 720p/50 signal should provide better movement portrayal and the 1080i/25 system should provide more detail via the higher horizontal resolution and Considering all factors, a 720p/50 signal seems to have more advantages than a 1080i/25 signal.

Incidentally interlace is elsewhere described by this author as a "trick" and he states "1080i/25 already suffers a first spatio-temporal “compression” in the baseband domain when interlacing is applied ...". Gardnan 12:24, 7 November 2006 (UTC)

Chronological division of the article?

Digital video broadcasting – and, consequently, subjects like MPEG-2, MPEG-4, and HDTV broadcast resolutions – pop up in these discussions quite often. They are important topics, and relevant enough to be discussed at length in an article about interlace, but I feel these subtopics are – at the same time – somehow skewing the viewpoints represented here, and steering the spotlight off of what, in my opinion, should be the main focus of the article: discussing interlace neutrally as a technique that has been used for television broadcasts for the past 60+ years, and will still be used as a tv broadcast technique for a long time into the foreseeable future. (The move towards progressive systems has arguably began, but interlaced video cameras, tv sets, game consoles, DVD players, SDTV broadcasts, vast amounts of interlaced archive material in the vaults of tv production companies, etc., are not going to disappear anywhere overnight.)

Hence, I suggest a more chronological approach. The article could be divided up in chronological sections in a such way that it would first explain...

  • Who invented interlace? (Was it Ballard? Was it Fritz Schröter instead? Was interlace invented independently in different countries or was there only a single inventor, copied by everyone else?)
  • What interlace meant in the 1930's (Why and how it was conceived, what kind of technological breakthrough it was back in the day?)
  • Which were the first publicly used tv systems to use interlace (Was NTSC or any of the current SDTV systems the first one? Not necessarily!)
  • How did interlace end up being so unanimously adopted as a major technique used by all tv systems worldwide? Who did these decisions? When? On what grounds? Can we cite the reasoning of these committees? Do we have references?

Then we would move on to explain the 1950–1995 period, during which interlace was already a given. Nothing much to say here, except for a mention about video games ("tv games"), home computers, and game consoles which often used progressive variant “A” on an ordinary tv set, instead of interlaced scanning. ("Office" computers and their tendency to use progressively-scanned monitors should also be mentioned, and they are mentioned in the article, even now. Then again, interlaced modes used to be used semi-commonly even on PCs during the early SVGA era.)

Finally, we would move to discussing the 1995-to-the-present-day period, where digital video compression, digital video broadcasts (plus HDTV, with or without interlaced modes), and displaying interlaced material on progressive displays, such as on LCD panels, are everyday issues, and where we're beginning to see that interlaced material, efficient digital video storage, and progressive displays don't always go too well together.

I'm not necessarily suggesting the above as a rigid general outline of the article (and that might even be against some structural definition of a good Wikipedia article, for all I know), but there should be some sort of division and organization of the topics to different time periods – at least in some ways like in the above, even if only internally in a couple of sections. Otherwise things will just get too interspersed and muddled.

The clear benefits of interlace are mostly in the past – but that's a long and glorious past, and that story has not really ended yet. In short-sighted present-day view – especially in the Western countries where progressive HDTV LCD panels, PC TV cards and whatnot have (on the larger timescale) just only recently become affordable and popular, there is easily a temptation to overemphasize the problems of interlace on progressive-scan displays, and forget about the history of the technology, but that is not fair handling of the topic. The burdens and drawbacks of interlace have only really been starting to become relevant during the latter half of the last decade. Both viewpoints should be discussed, but if they are discussed with no chronological separation into time periods at all, and with the primary emphasis on the present-day problems (such as HDTV resolutions etc.) only, I believe the reader will be left with a very confused and skewed view about the topic. — Jukka Aho 03:15, 9 November 2006 (UTC)

I think this is a great suggestion... although I would suggest that a good technical description of interlace should be the first section of the article - so that people who come to the article in order to understand what interlace is find what they are looking for right up front. One of the problems with Wikipedia is the fact that there are so many editors that contribute little bits and pieces to articles, and sometimes the articles get a little disjointed as a result.. they lack a smooth flow and a good layout. So, as they say... "Be Bold!". You should feel free to make some edits... or even rearrange the overall layout a bit. Tvaughan1 04:36, 9 November 2006 (UTC)
Personally I think it would be great if we could have a sub-section that discusses the effects of interlace on MPEG compression... but this is a very complex topic that is probably only well understood by a very small group of experts (engineers who develop encoders ... working at places like the Fraunhofer Institute, CinemaCraft, Main Concept, and Microsoft). Tvaughan1 04:36, 9 November 2006 (UTC)
Jukka Aho is right to say that native progressive scan devices is only now becoming affordable for home TV. However it has also became dominant very quickly - several European major shopping chains simply don't sell anything else. Important though, in Europe, is that this coincided in time with the freezing of HDTV standards. That is why in Europe it is possible to take a firm position in favour of progressive. The situation was different in the USA, when the ATSC HD standards were established years ago in a CRT dominated world, in which compression issues with interlace were less commonly understood, and interlaced HD was an option.
Gardnan 08:03, 9 November 2006 (UTC)
As I said earlier in this long thread, legacy issues are a stone around the neck in changing TV to home technologies. The opportunity to make a clean break is rarely possible - backward compatibility is nearly always required.
Gardnan 08:03, 9 November 2006 (UTC)
In my opinion the EBU is making a significant and important contribution with its strategic views and particularly its arguments for 3rd generation HDTV (effectively 1080p50 all the way from the studio to the home). The arguments are clear. Avoidance of encoding inefficency of interlaced pictures compared to progressive is an important factor in their arguments that this will be possible in existing channels in the future. Go to the technical pages at http://www.ebu.ch.
AlsoI strongly recommend the annex to the demonstration article in issue 308 of the Technical review - a clear presentation of the temporal and spatial spectrums of interlace and Kell factors answers many of the points in this section. Kell factors are very significant in TV scanning, and highly relevant to points made in this thread and the main article. It is a major ommission that it is not much talked about in the interlace sections.
Gardnan 08:03, 9 November 2006 (UTC)

Survey - please indicate your opinion on the Interlace illustration

I believe that the interlace illustration below does not illustrate interlace in any way that could be considered neutral. The image does not contain interlaced video - it is a hand-made animated GIF image. The image was carefully constructed in order to generate interline twitter. Actually, the overall effect of the interlaced portion of the illustration is more representative of flicker, which is not what interlaced video looks like. The image is an obvious attempt to demonstrate the problems caused by interlace, but it was not put in that section of the article - it was added to the "Description" of interlace section of the article. The supposed interlaced images have alternate lines that are black (this is not the way interlaced video looks, due to the persistence of phosphors on a CRT and the persistence of vision of the human visual system). The colors and brightness of the "progressive scan" and "interlaced" portions of the illustration are different. In short, this is a horrible illustration of interlaced video. The image is not from a neutral point of view, and it represents original work. There are no citations that can support this image. User:Algr, the author of the illustration is welcome to comment on why he believes the image is valid, helpful, or neutral. Tvaughan1 18:30, 17 November 2006 (UTC)

 


Straw Poll

Please sign your name using four tildes (~~~~) under the position you support, preferably adding a brief comment. If you are happy with more than one possibility, you may wish to sign your names to more than one place. Extended commentary should be placed below, in the section marked "Discussion", though brief commentary can be interspersed.



Discussion

I see this image as vital. It is the only thing in the article that actually LOOKS like interlace the way it appears on a real TV. A reader who closely examines a real set will likely see the aperture grille, dot crawl, and the texture of the faceplate as well as interlace. Without this illustration, how will the reader know which of these is what this article is describing?

Tvaughan1 has included misinformation in the top of this poll. There is no persistence of phosphors on a CRT into the next field. Each field is well under 1% brightness before the next field is scanned, as is shown here: [3] The P4 phosphor is one used for television, as shown here: [4]. Thus it is perfectly accurate to show the space between fields with black lines. Of course we have had this discussion above already here: [5] Algr 06:04, 18 November 2006 (UTC)

Interlace, as well as all video and film motion pictures, depends on the persistence of vision characteristic of the human visual system... not the persistence of phosphors on a CRT. Televisions are designed to be viewed from an appropriate distance (about 6 picture heights for standard definition, about 3 picture heights for HDTV). If you view any video display too closely you will see artifacts or details that you would not see at the proper viewing distance. As anyone with a standard "tube" television knows, this illustration is not what interlaced video looks like. Tvaughan1 15:44, 18 November 2006 (UTC)
I can see this on any interlaced set. It is VERY obvious on black-and-white screens, but in color it is hard to distinguish it from the aperture grille. You can see interlace exactly as shown here by looking closely at a color TV screen from an extreme angle - place your eye as close to the left or right side of the screen as you can, and play something bright. Finally, it is visible any time you have motion - in sports you constantly see the jaggies sliding up and down the lines painted on the field - it is just part of the 'look' of TV that everyone is used to. Algr 19:34, 18 November 2006 (UTC)
  • Bias disclosure: I agree with Algr on nearly every point except for his unwise choice to discuss "interlace as a form of compression," which is an idea I've always thought adds more confusion than light to the discussion. That said, the illustration basically captures the right idea. Is this really "what interlaced video looks like"? Obviously, the animated gif updates a lot more slowly than a real display would, and there are other practical issues in getting an always-perfect representation on the page. But it should be kept as a useful illustration. Snacky 03:06, 19 November 2006 (UTC)

Eugene2x's version of interlace.gif

Hi Eugene2x! Sorry about the revert. While I'll admit that my art left something to be desired, there are some more practical issues with yours. Could you fix them?

1) The picture of the clouds has almost no detail near the level of the scan lines, so the center image does not twitter. A major point of the image was to show people what twitter looks like, since it is hard to describe in words. I chose the image with a US flag because I knew it would be easy to put the red and white stripes where they would best illustrate the subject. When the flag ended up needing to be that small, I drew the soldier. Other things you could try are bricks, window blinds, trees with prominent branches, or a large building with prominent windows.

2) The aliased image on the right is not blurred correctly. You appear to have blurred the image AFTER adding the scan lines, and you also blurred it both vertically and horizontally. The correct way to blur it is to shift the whole image down one pixel, and then mix it 50% with itself in the original position. This replicates what CCD scanners do when they output in interlaced mode.

Thanks for helping! I look forward to seeing what you come up with! Algr 08:04, 22 December 2006 (UTC)

Ok, I'll try as best as I can. I thought yours looked better too! —The preceding unsigned comment was added by Eugene2x (talkcontribs) 19:46, 22 December 2006 (UTC).
How about this?

 

--Eugene2x -- ☺ Nintendo rox! 22:13, 22 December 2006 (UTC)

Yes, that's much better. I like the fourth picture, it makes a useful point. This could make a Featured Picture with a few more tweaks. Are you game for it? Algr 08:20, 23 December 2006 (UTC)


Hmm... Yup, I think I wanna make it a Featured Picture. So, what changes do I exactly need to make? ~~Eugene2x ☺ ~~ 05:52, 24 December 2006 (UTC)
Well, it would be nice if there were more details like the word "interlace" in the picture for the lines to twitter. Perhaps a horizon line, or some background figure with stripes like a building? Or you could add more text. "How interlacing effects detail" perhaps?
Also, the brightnesses of the progressive and interlaced images should match. The lines will naturally darken the interlaced image, but in real life the reduced detail lets you pump up the brightness on a CRT, so interlaced sets used to always be as bright or brighter then progressive. I don't know if you could brighten the interlaced image without touching the black lines, so the best thing to do here might be to darken the progressive images to match the interlaced ones. (This is all surprisingly tricky - I had to eyeball it on mine, and I don't think I got it quite right.)
I see you've added a new "Interlace scan" gif. It looks much better, but it seems you've duplicated an error that always bothered me about the old one. In the animation, the second field draws in while the first field stays in place, resulting in a moment when the whole screen is illuminated, and then both fields disappear together. In fact, the scan lines on TVs fade very quickly, and the first field is totally gone before the second field starts being drawn. You could show this by blacking the first field out as the second field is drawn in, and vice versa, so that at any given moment, half the screen is black.
Thanks for your help. It's a great improvement already! Algr 08:53, 27 December 2006 (UTC)

While I appreciate the effort Eugene2x made to improve the illustration, the image is not representative of interlace in any way that could be considered accurate or neutral point of view. This is evident in the way the "interlaced" portions of the image have a completely different luminance than the "progressive scan" portions. This illustration is original work. The problem with original work is that it can't be considered a reliable, peer reviewed source of information for this encyclopedia. How was this image created? How does this accurately represent video that would be captured with a video camera? It isn't video, and it wasn't captured with a video camera. To represent the "twitter" phenomenon fairly you can't use a CG image! CG images must be processed correctly to avoid severe twitter, and I don't think the average Wikipedian has the tools to do it. You can not "represent" interlaced video using a progressive scan signal (an animated GIF image displayed on a progressive scan computer monitor). The only way to fairly compare interlaced video to progressive scan video is to use 2 different monitors, side by side. Any attempt to represent the visual qualities of the 2 techniques on this page is not just fraught with difficulty, it is impossible (and it can not be done from a NPOV). The only thing that can be represented fairly is the scanning of alternate lines (the first image on the page), and what happens when you display interlaced video frames progressively (the bike rider image). Tvaughan1 00:07, 28 December 2006 (UTC)

Note that this is just a simulation. In no way are these examples of interlacing supposed to be perfect and extremely accurate. Besides, one could just look at an old CRT TV from the side and see the interlacing lines. ~~Eugene2x Sign here ~~ 01:02, 28 December 2006 (UTC)
Ok, I made another one:

 

Eugene2x - I understand where you are coming from, and I appreciate what you are trying to accomplish... but I think that you just revealed your bias when you said "just look at an old CRT TV from the side and see the interlacing lines". This is your personal experience, which doesn't count for much in Wikipedia. An old CRT TV is just one example of interlaced video. Do you have any professional quality CRT monitors? I do. They look pretty good... nothing like your illustration or previous efforts. All video displays are designed with the viewing distance and angle in mind. Viewing a CRT from the side is not an unbiased way to evaluate interlace as a video capture/display technique. How does your LCD monitor look from the side? My point should be obvious - these illustrations are clearly biased against interlace. If it isn't obvious, I don't know what is. Who would possibly prefer interlaced video over progressive scan with this illustration to show the difference? Nobody. But interlaced video clearly has applications when all constraints are considered from end to end (capture, encoding, storage, transmission, decoding, display). This illustration can't possibly consider the persistence of the human visual system that interlace relies on. The interlaced portion of the illustration will always be shown on progressive scan computer displays, and it will always have serious issues. It is darker, and if you attempt to compensate for this the colors will change. It isn't video, which is what we are supposed to be talking about when we discuss interlace. It's an animated GIF. It just simply shows a computer graphic that makes interlace look like crap, and progressive scan look good. Tvaughan1 03:36, 28 December 2006 (UTC)

As Wikipedia:Choosing_appropriate_illustrations points out, illustrations can be used for one of several purposes. The problem with the above illustration is that it attempts to illustrate the appearance of interlaced video and progressive scan video, and it fails miserably in doing so for all of the reasons already stated. The first image in the article only attempts to illustrate the concept, and it succeeds in doing so. Is it OK to inaccurately illustrate the appearance of a visual display technique? No. The whole point of the display technique is to improve the appearance of the motion picture display (given various technical constraints). When you inaccurately illustrate the appearance of the quality of the displayed video, you inaccurately portray the value of the technique. No amount of explanation can possibly correct the misperceptions that result from this type of inaccurate illustration. The explanatory text below the image attempts to portray the image as a demonstration of the concept, rather than of the appearance, but the nature of the image is clearly an attempt to illustrate the appearance (especially given the previous image). Tvaughan1 04:10, 28 December 2006 (UTC)

#2 :Eugene2x's gif

Hi Tvaughan1, I figured you'd be back at some point. Of course I don't agree with you about anything except that the brightness of the progressive images should be adjusted.

 

Complaining that the image isn't "true interlace" is rather like complaining that the piston image above does not make your monitor spout real fire, or that it moves impossibly slowly for a real piston engine. If this illustration moved at accurate speed, it would be useless, because the viewer would not be able to see how it worked. The same thing applies to our GIF. People know what a TV image looks like, but they won't know what we mean by "twitter" unless we have an image that shows it clearly. I explained this to you above before. For the sake of clarity, Illustrations almost always enlarge their subjects, slow them down or speed them up, or pick the most extreme or photogenic examples. Ours is no different. Actually, I prefer it at the 30 hz rate, but others have said that it takes on a "tearing" effect on older computers. (It looks very much like TV on my screen) But this illustration's purpose is to show people what to look for on a real TV so that when they do see it, (more subtly) they know what it is.

I do use professional quality CRT monitors, and they twitter exactly the same as any other TV. Indeed, it would be a problem if they didn't because you are supposed to use them to spot problems in the image. The only reason we brought up looking at CRTs from the side is because before you were insisting that no one can EVER see line crawl. I can see it on almost any image, but it just gets more obvious from the side.

I simply can't fathom what you mean by "original work". Anything that isn't original would be plagiarized. We have nothing but your POV that this image would create any false expectation in the viewer. Algr 05:57, 28 December 2006 (UTC)

Something like this could help: (Maybe with the horizontal lines a little darker?)

  Algr 06:16, 28 December 2006 (UTC)


The engine image is meant to illustrate a concept. The interlace image is obviously an attempt to illustrate the appearance of video using different techniques. Appearance vs. Concept. If this image was placed in the "Problems caused by Interlace" section, and limited to demonstrating the concept of Twitter, I would have no problem at all with it. It is currently in the "Description" section, where it is portending to illustrate the appearance of interlaced video vs. progressive scan video. Any average wikipedian would view this image and quickly conclude that interlace is vastly inferior to progressive scan under similar circumstances. In this illustration interlace looks like crap, and progressive scan looks perfect. The illustration demonstrates the issue of twitter, but it causes every alternate field of interlace video to be missing, and it causes the whole frame to flicker horribly. Meanwhile, on your monitor and mine, the progressive scan portion of the image is being refreshed at 72 Hz or better (in all likelihood). Progressive scan appears to be perfect. What does it really look like at 30 Hz frame rates with subjects in motion? It would flicker horribly, and the subjects would have poor motion portrayal. Appearance vs. Concept. This image is clearly an effort to illustrate the appearance of the video, given the titles in each section of the image, despite the disclaimer below the image that it only illustrates the concept. The interlace portion of the image suffers from problems that are insurmountable, given the situation (it will always be viewed through the Internet on PC displays). Tvaughan1 14:25, 28 December 2006 (UTC)
I think this should be okay:

  ~~Eugene2x Sign here ~~ 02:26, 29 December 2006 (UTC)

The bricks are good, but the attempt at a cube effect is a problem. In proper perspective, the lines on the sides and top ought to converge on the horizon. (So you wouldn't actually see the top.) And the composition with the ball seems crowded. BTW, The brightness matches just right now.
"Any average wikipedian would view this image and quickly conclude that interlace is vastly inferior to progressive scan under similar circumstances. "
And now you have revealed YOUR bias. The only way you can claim that "interlace looks better then progressive" is by a forced perspective that conflicts with industry standard terminology. Most people only know to measure sharpness by line count. Industry standard is to count the lines of both fields in an interlaced signal. Therefore VGA and NTSC are always presented as "similar circumstances" since they both have "480 lines" in standard industry terminology. But VGA looks vastly superior then NTSC under any possible circumstance. To make the comparison based on equal bandwidth you'd need a disclaimer the size of this whole article to explain what you were comparing to what. You'd might as well claim that NTSC and ATSC are identical because they both use 6 Mhz.
If the illustration were run at 60 hz, the third image would flawlessly recreate the appearance of NTSC twitter. Running it in slow motion is a common and easily understood illustrative technique. The twitter becomes more blatant, but it's overall quality is the same as what anyone sees on a regular TV. Anything else is your POV. Algr 07:46, 29 December 2006 (UTC)
First, it's disingenuous to quote me saying something that I didn't say. Please stop it. I didn't say "interlace looks better than progressive" in this discussion, yet you quote me as saying this. Furthermore, you simply change the question, then answer your own question. I wasn't asking about sharpness or resolution, nor was I comparing NTSC to VGA or anything else. I would vastly prefer that comments are directed to the topic, not the person. I'm asking legitimate questions regarding the neutrality and the appropriateness of the illustration. You can't defend the neutrality of the illustration by responding with accusations towards the person who asked the question. You need to respond to the questions. Tvaughan1 13:36, 29 December 2006 (UTC)
I'll ask the question again: As per Wikipedia:Choosing_appropriate_illustrations, WHAT DOES THIS ILLUSTRATION ILLUSTRATE? Does it illustrate the CONCEPT of interlace? Or does it illustrate the APPEARANCE of interlace? What would the average person conclude when reading this article? The explanatory text that follows the illustration says "This animation simulates a progressive and two interlaced images with the same line count". "Simulates (the appearance of)" the images? Tvaughan1 13:36, 29 December 2006 (UTC)
The illustration is clearly meant to demonstrate one of the problems caused by interlacing - interline twitter. At a minimum, it should be placed in the "problems caused by interlacing" section, next to the explanation of interline twitter. Where it is currently positioned, it is clearly attempting to illustrate the appearance of interlaced and progressive scan video overall, and it is extremely misleading. Tvaughan1 13:36, 29 December 2006 (UTC)
The illustration is of the APPEARANCE of interlace. It is slowed down for clarity, but is otherwise perfect. (Maybe we should try it at 15 or 20 hz to see if the tearing problem stays fixed.) In terms of neutrality, you are setting up an impossible standard by demanding that the illustration be exactly the same as something that is massively variable. We came up with four conflicting ways in which an interlaced signal and a progressive one might be considered "equal". [6] As far as the quoted statement, that is the only idea you have ever contributed to this article, so just because you didn't speak the words at that moment doesn't mean we shouldn't think that that is your objection. Algr 07:28, 3 January 2007 (UTC)
For the last time - please stop misquoting me, paraphrasing me, or summarizing my position. I speak for myself. Anyone can read what I have said and form their own opinions of what I meant. Please stick to the subject... the subject isn't me. Try to avoid using the word you entirely. It is impossible to have a decent discussion if the discussion is constantly getting personal. Tvaughan1 06:26, 5 January 2007 (UTC)
OK, now we are getting somewhere. Clearly, the appearance of the interlaced portions of the illustration isn't good at all. The supposedly interlaced portions of the illustration flicker horribly. They appear to have half the resolution of the progressive scan portions, because every alternate line is missing in each field (this isn't the way interlaced video appears, or is perceived, due to the way the human visual system works - the persistence of vision). In other words, the progressive scan portions have twice the bandwidth of the interlaced portions. Is this a neutral point of view? I don't think so. The images are not video, they are computer generated animations. Interlace is a video capture and display technique. Computer generated animation is not video, and it is best displayed on a progressive scan system. The images don't show real video (shot with real video cameras), containing objects in motion. If we compare progressive scan video to interlaced video from a truly neutral point of view we would have to compare systems of equal bandwidth. Both systems would have to be captured and displayed in their native format (but interlaced video can't be displayed on a web page that is displayed on a progressive scan computer display). The interlaced video would either have twice the resolution or twice the refresh rate. This illustration doesn't even come close to being accurate, neutral, or supported by any objective sources. It's totally misleading, and it doesn't belong in the article. Tvaughan1 06:26, 5 January 2007 (UTC)

STANDARD test image

Here's an image that uses a STANDARD test image (Barbara), and a STANDARD test pattern, and ONLY attempts to show the twitter/judder/whatyawannacallit. It's somewhat big, but that was necessary to show all the possible examples. I hope it manages to shut you all up. If you want to explain other effects or inherencies of interlacing, use another image. You can't show everything in one image clearly. --Mufunyo 20:08, 19 January 2007 (UTC)

 

Interesting. You've done it in a way that maintains full image brightness, and comparable color. Algr 05:54, 20 January 2007 (UTC)

Very good image. However, it may confuse people who don't know much about interlace. ~~Eugene2x Sign here ~~ 06:14, 27 January 2007 (UTC)

I don't think it can get any more confusing than the previous images shown here. --Mufunyo 18:44, 30 January 2007 (UTC)

This is better than the earlier attempts, but it is not video, it is a standard still image with pseudo-video processing. This still image was not captured with a video camera. Again, these types of images may be helpful to demonstrate a concept, particularly the problems caused by the conversion / display between progressive and interlaced formats. It is impossible to accurately represent the appearance of interlaced video on a web page, and so I submit that all such attempts should be abandoned, and all such images should be removed from the Description section of the article. Tvaughan1 22:10, 4 February 2007 (UTC)

This is not a video, and Wikipedia is not a movie. Wikipedia is a website, and these animated GIFs already push its medium to the outer boundaries of what can be done while still catering to most internet users. If this Wikipedia article were a Discovery documentary I would be able to explain interlacing nearly perfectly, showing things in full-, slow-motion and with still comparisons. I never said I wanted to accurately represent the appearance of interlaced video on a webpage, I merely intend to show the concept of interlaced twitter, and what can be done about it (using progressive scan, or lowpassing), which just so happens to remotely mimic the actual appearance of interlaced video. I thought the big disclaimer I posted earlier was clear enough? If you only demonstrate one aspect (twitter), you're clearly leaving out other important effects intentionally and thus never trying to achieve a realistic representation. But if you clearly mention what you are attempting to demonstrate, that is not a problem. Also, the problems highlighted in my GIF still apply to motion (though explained through a still image), and especially so now that TV station operators are turning off lowpass filters because the majority of the population has LCD TVs with some type of smart deinterlacing. In the Netherlands, computer overlays that announce the upcoming show or a delay or other schedule modification on the Dutch channel Veronica flicker extremely on interlaced TV sets. Television and interlaced broadcasting are more than just things that are captured with a video camera. As a sidenote, Barbara is highly likely to actually have been captured with a video camera, it is 787x576 which becomes 768x576 after cropping black borders, which in turn is standard resolution for square-pixel aspect ratio digitized PAL, the image has no high frequencies along the vertical axis (so you could say it has been low-passed), and even shows slight ringing along the horizontal axis, on the boundary between the face and patterned clothing. I just downscaled the image and sharpened it, because in full size it would not have been usable for this comparison. --Mufunyo 14:10, 5 February 2007 (UTC)

I completely agree, Mufunyo. Wikipedia is not supposed to be an online video-streaming source such as YouTube, it is supposed to be a place where many can share information with each other in the form of an encyclopedia. If anyone wants to see real interlace, they can just view an old CRT monitor or TV up close. Tvaughan1, the only thing you have done for now is look at an animation on the article and complain. Why not actually try to contribute to the article? ~~Eugene2x Sign here ~~ 05:14, 6 February 2007 (UTC)
Thanks for your contributions to the discussion Eugene2x. I've contributed many edits to the article. An entire section, as a matter of fact. Just take a look at the article's history. Lately, any edits I make are reverted... so I've thought it best not to get into a revert war. It's great to see people acknowledge that it is impossible to represent the appearance of interlaced video on a web page. The illustration with the orange balls that is currently in the description section of the article attempts to illustrate the appearance of interlaced video - and it fails completely. This is misleading and inappropriate. This illustration should be deleted, and no attempt should be made to illustrate the appearance of interlace in this article. The illustration shown above on this talk page (or portions of it) could be useful to demonstrate certain concepts, in the "problems caused by interlacing" section. Tvaughan1 19:43, 6 February 2007 (UTC)
It is impossible to represent the appearance of a cube on a web page, since screens have only two dimensions - should we eliminate all of those illustrations as well? Without some visual representation, the article would be mostly useless to anyone who didn't already know what interlace looked like. The orange balls illustration literally IS interlace, (not a simulation) because it has the same illumination pattern that a real CRT would have if it were illuminated at 10 hz with a tight beam. It is only POV that the illustrations do or don't look like interlace - I think they do.Algr 06:42, 7 February 2007 (UTC)
It's just amazing that anyone could honestly say that the illustration in question accurately represents the appearance of interlaced video. It's equally amazing how insane these discussions are. Stooping to the level of the discussion... why would anyone argue that the appearance of a cube can not be represented on a 2 dimensional page. It can. Photographs represent 3 dimensional objects on 2 dimensional surfaces all the time. Video does this. Perspective drawing does it. These are fallacious arguments... because X is not possible Y is not possible (though X had nothing to do with Y). Let's stick to the subject... interlace. The illustration in question attempts to portray the visual quality of interlace versus progressive, and all it does is portray interlace as crap and progressive as good. This is not NPOV... this should be crystal clear to anyone with an objective viewpoint. The illustration is ACTUALLY trying to illustrate a specific "Problem caused by Interlacing"... interline twitter. What if we create a new illustration that shows that for a given bandwidth, progressive suffers from poor motion portrayal, or poor resolution? Should we put that in the "Description" section of the article? Tvaughan1 01:39, 8 February 2007 (UTC)

Some LCDs use interlace

Did you know that some cheaper LCDs DO use a form of interlace? They use a striped pattern similar to CRTs, while others use a diagonal checkerboard pattern. A few years ago I read about a technology that alternately blocks one half of each LCD pixel, and then the other half, thus letting each pixel act like two pixels next to each other. But I can't find any references online now, so I can't write about it. (They may be calling this something other then "interlace", but the principal is the same.) I've seen this on many portable DVD players, and just a month ago I saw it on a 17" computer monitor. It doesn't look so hot - I think it runs at maybe 30 hz instead of 60 hz. But it probably shaves a few dozen dollars off the price of the screen.Algr 06:42, 7 February 2007 (UTC)

I think you mean dot inversion and row inversion, which is a form of power saving. Cellphones, PSPs, Nintendo DSs, portable DVD players, all use row inversion to preserve battery life. http://www.techmind.org/lcd/ --Mufunyo 11:09, 8 February 2007 (UTC)

An interesting link - I didn't know this stuff about LCDs. I'm not sure if this is what I was describing though, as this seems to be something common to ALL LCDs. What I am describing is only visible with certain ones. On those screens I can ALWAYS see a horizontal line pattern that looks just like the one on my .GIF. (I've only seen one monitor with a checkerboard 'interlace' pattern - it was a 17 inch monitor.) Algr 04:55, 22 February 2007 (UTC)

The high-powered inversion schemes are practically invisible (subpixel dot inversion), whereas the power saving schemes (row inversion) are visible by simply shifting your eyes across the screen rapidly or tilting the display (in the case of cellphones/other portable devices). Thus, it is very possible that what you are describing is row inversion, as it is much easier to notice than schemes meant for high-power desktop displays. --83.98.253.113 18:12, 22 February 2007 (UTC)

Confusion: interlacing does not reduce bandwidth

IMHO, the following sentence is erroneous :

Interlaced video reduces the signal bandwidth by a factor of two, for a given line count and refresh rate.

The bandwidth is related to the image rate, no matter the scan mode.

In progressive scan, you display the entire image twice (25 i/s at 50 Hz or 30 i/s at 60 Hz), so you have to store it, but you stil transmit 25 or 30 i/s.

The difference is that in interlaced mode, you can display the image in "real time" as it is transmitted, whereas in progressive mode you have to store it before displaying it.

There is no movie with 50 or 60 image per second. This would be useless, 16 i/s is enough to have an acceptable fluidity of movement.

I don't change it myself because my english is not so good and it would take too long to clean it all.

Best regards

cdang|write me 15:38, 22 March 2007 (UTC)

There are no movies at that frame rate, but I believe that some TV sports coverage is. – Smyth\talk 18:07, 22 March 2007 (UTC)

Yes, it does. I wrote that sentence, and it is accurate. For a given refresh rate (field rate... not frame rate), interlace reduces the bandwidth by a factor of two. This describes the video signal, not the display refresh rate (which could be anything). Tvaughan1 18:16, 23 March 2007 (UTC)

That's right. Also, progressive scanning does not necessarily require that an image be stored and projected twice. That is only true if it it transmitted at 24p - 30p, but 60p modes are available at up to 720p for broadcast, and 1080p for Blu ray disks.
Almost anything shot on video before the mid 1990's would be at 50 or 60 images per second, not 24 - 30. The difference in look is quite obvious. Some famous 60 hz programs: All in the Family, WKRP in Cincinati, Night Court, Survivor, most daytime soap operas, talk shows (Letterman & Oprah) unless they are showing clips from movies. 24 hz programming: Mash, I love Lucy, Law and Order, Sienfeld, Lost, Star Trek (all). BBC programming traditionally mixed 50 hz studio video with 25 hz location film, (Examples: Original Doctor Who, Monty Python's Flying Circus.) but in the mid 80's they started shooting everything on video, with 25 hz or 50 hz consistant for the whole program. Algr 21:59, 24 March 2007 (UTC)
OK, I don't know TVHD and did not think about video filming, so I apologise if I was a little bit rude — I have mainly experience in cinema (I used to be projectionnist). But I think this would need some explanation. I mean, it is the reduction of resolution that reduces the bandwidth (for 625 lines/50 Hz standard — 575 lines resolution —, you have 1 picture at 287.5 lines every 0.2 sec) and not interlacing itself. If you transmit at 288p-50 Hz or 575i-25 Hz, that's the same, isn't it?
Or?
Would we agree if I write: "The bandwidth is related to the image rate times the resolution"?
cdang|write me 09:14, 26 March 2007 (UTC)

This is a pointless argument. Nobody ever makes the choice between progressive and interlaced video at the same line count and refresh rate, because they are applicable to different situations. You might as well say that "12 bits-per-pixel images reduce the bandwidth to half of 24 bits-per-pixel, for a given resolution". This may be true, but it's not likely to be useful, because the 12bpp image simply has half of the 24bpp images's information removed, and so is appropriate for a different set of purposes. – Smyth\talk 19:26, 26 March 2007 (UTC)

I don't mean to be useful, I mean to be as accurate as possible. Look at the #Removal of the "Interlacing as a compression technique" section discussion above; if you simply say "interlacing reduces the bandwidth", someone can believe it is can be used for compression which is false. If you point out you reduce the Y- or time resolution (Y-resolution if you shoot at 50/60 Hz, time resolution for telecine), then the reader has a clear view.
IMHO.
cdang|write me 11:54, 27 March 2007 (UTC)
" If you transmit at 288p-50 Hz or 575i-25 Hz, that's the same, isn't it?" - Yes, those use the same amount of bandwidth, but the 575i image would usually look much better. In your example, analog bandwidth is equal, so interlace will give you a better picture then progressive. My example compares NTSC with VGA. Here line count is equal, (480 lines) but bandwidth is cut in half. We had quite a discussion about this before - the issue of what constitutes an "equal comparison" of interlace and progressive is quite complex, and gets into what technology is practical at a given time period. Interlace does NOT simply reduce time resolution - in fact from some perspectives it can increase it. The overall effect of interlace is rather complex. If this still isn't clear in the article, maybe it is worth a new section. Algr 06:01, 28 March 2007 (UTC)

'Digital' CRTs?

In History: "However, newer digital CRT displays have much better picture quality than LCDs, DLPs, or plasma displays..."

Where can I get one of these new super CRTs? Should there be a reference or at least an explanation for this?

Yikes! Almost everything in that paragraph was wrong. Removed. Algr 07:21, 6 June 2007 (UTC)
Actually, you didn't have to remove that paragraph, simply replacing the words "digital CRT" with "SED display" would have been sufficient. As for the original question about where to get them, I think you'd have to break in to a research lab -- the technology isn't mass-produced yet. 83.98.253.113 00:31, 30 August 2007 (UTC)

Interlace isn't restricted to CRT displays

Plasma displays based on the ALiS technology display alternate fields as well. See:

<http://www.plasmadepot.com/plasmatv/howplasmaworks.html>

for technical details.

Skeptical thinker (talk) 17:08, 24 November 2007 (UTC)

Interlace demo anim, again

All the above animated GIFs are a very bad demonstration of interlace, as they use a static scene, while in reality dynamic scenes are used (watch TV for a few seconds, if you don't believe me ;-)

I would also like to be very careful with statementas like "interlace saves ..." , "interlace gives ..." and similar. Because people wil take it as reference and then we will have another 50 years of pictures like this:

 
The same image as used in the article, click to look at the real size, as the interlace effect is lost while reducing the image 5 times for thumbnail size. NOTE: Thumbs are apparently illegal , here is the link to the picture

Remember, all modern displays are progressive in nature. To display interlaced content, they must convert it to progressive. So there is no choice really. Everyone uses (and will use) progressive, except some will convert it to interlace for transport and then back. Losing quality in each step. Interlace is good if it is used from the image capture through transport and final display. Otherwise it is a nightmare. The above picture is (actually not the best) good illustration. Another thing. 50i flickers so even if you have an interlaced display you still have to convert the interlaced content. Basically the "advantage" of interlace is that it creates a market for "sophisticated and smart deinterlacing systems". —Preceding unsigned comment added by Xerces8 (talkcontribs) 12:12, 25 November 2007 (UTC)

That's your opinion, I disagree on almost all accounts, but remember wikipedia is not a platform for arguing opinions. We are supposed to present document facts or opinions, not argue our own. Carewolf (talk) 15:07, 25 November 2007 (UTC)

With which of the following do you disagree or think that it is not a fact ?

  • interlace is used for TV/video (as opposed to for example computer display)
  • TV/video is mostly moving picture
  • comb effect results in deinterlacing in some cases
  • modern displays are mostly LCD or plasma type
  • LCD and plasma (except AliS) are progressive by nature
  • interlaced content can not be displayed on progressive display without conversion
  • interlace related conversions decrease image quality
  • 50/60 Hz (divided by 2 due to interlace) display refresh flickers

--Xerces8 15:16, 30 November 2007 (UTC)

  • Comb effect is the result of bad deinterlacing, but that's a bug in the deinterlacing.
  • Modern LCD display are perfectly capable of displaying in double frame-rate (50 or 60fps), which is needed for showing interlaced video.
  • Interlaced content can be shown at double framerate in a way that doesn't convert anything, but is sometimes known as deinterlacing anyway. The point is it doesn't decrease image quality.
  • As above, showing interlaced content on a non-interlaced display doesn't have to decrease quality.
  • Interlacing decrease flicker by using 50/60 frame per second, unlike progressive (usually 25/30 fps) which flickers more.

Carewolf 15:52, 30 November 2007 (UTC)


Can you tell more about this "double framerate" method ? It is not mentioned in the article, neither in the comments.

--Xerces8 12:48, 4 December 2007 (UTC)

Aha, you mean Line doubling ? Well, that way you lose the "advantage" of interlace (more resolution with same bandwith).

--Xerces8 (talk) 08:03, 6 December 2007 (UTC)

--Anon 03 December 2007 There is a broken link in the references section of this page. The Sarnof link. The problem is that the link in this page uses .htm as the file suffix. It should be .html

Thanks —Preceding unsigned comment added by 203.12.172.254 (talk) 00:56, 3 December 2007 (UTC)