15

As I had understood things, "interlaced" video was nothing but a clever way to save data bandwidth by only sending every other line, swapping between starting on line 0 and 1 for every frame, in order to exploit the fact that TVs worked physically in a way which made this work.

But why would a game console or anything else that is connected locally to that TV be sending it an "interlaced" rather than "progressive" signal? Since there is no bandwidth being taken up in the air, what is the point?

TechMoan today mentioned that his old hand-held video camera had a "1080 interlaced" LCD monitor, which made me remember that I wonder about this. Also, it's extra weird in this case because an LCD screen is not the same technology, and you can't just skip every other line of pixels in this context. So there is something I'm not getting about why "interlacing" is used, since it is apparently not just for saving radio frequency bandwidth.

I am also confused as to why they chose the terms "interlaced" and "progressive" instead of just "half" and "full" lines.

1
  • Comments are not for extended discussion; this conversation has been moved to chat.
    – Matt Lacey
    Nov 29, 2022 at 17:31

5 Answers 5

36

nothing but a clever way to save data bandwidth by only sending every other line

There's more to the story. Interlaced video originates with CRT televisions. Standard analogue television is always interlaced. Interlacing does not "only send every other line". It sends two half-frames at twice the basic frame rate, which are offset from each other by a scanline. The result is a higher subjective frame-rate.

You don't really need a framerate much higher than 25 / 30 fps -- similar to film. But at 25/30 Hz refresh, a CRT will have very noticeable flicker. Longer persistence phosphors would help, but would make for ghosting. Increasing the frame rate to 60 fps at ~500 lines would double the bandwidth. It was in this context that the savings for bandwidth by using interlacing is relevant.

Interlacing and de-interlacing are non-trivial operations and are usually done with considerable digital signal processing. A traditional CRT TV cannot handle non-interlaced video. That's why your local video signal would be interlaced. Some devices (e.g. 80s game consoles) would use half the lines every frame. That's technically 50/60 fps progressive, but the TV is still interpreting it as if it were interlaced; the lack of frame alternation by a line often meant there were dark bars between scanlines on a large screen.

In the digital era, I do believe interlacing makes no sense. But the standards of analogue TV are ~70+ years old, were very widespread in defining video standards when analogue TV was then norm, and their legacy is still felt in the digital era.

5
  • 1
    The TV does not interpret a progressive signal as interlaced. The progressive signal has larger gap between scanlines because the signal is not interlaced so nothing is drawn between the lines of progressive frame. So they are offset by a half scan line, not full scan line. The CRT TVs can handle progressive just fine, e.g. C64 and NES send progressive.
    – Justme
    Nov 26, 2022 at 19:37
  • 4
    digitall interlacing would be useful for showing interlaced material coming from analogue recordings. Nov 26, 2022 at 21:19
  • @ThorbjørnRavnAndersen It not only would be, it actually is useful and done already by TVs and video player computer programs to be able to show interlaced material on progressive displays.
    – Justme
    Nov 26, 2022 at 21:33
  • 3
    @Justme Some CRT TVs didn't handle progressive. For a high school science fair around 1984, I prepared a passive demo running on my VIC-20, and a friend brought his kitchen TV to connect to it. That's when we found out that this TV would just display flickering out-of-sync garbage from the VIC; we had to bring in another TV to make things work. Some months later I learned that this was about interlacing: the VIC was progressive by default but supported interlaced output by flipping a bit in a register.
    – Nimloth
    Nov 27, 2022 at 13:32
  • @Nimloth That's unfortunate, but generally, TVs are very insensitive devices, and not all VIC-20 VIC chips even supported interlaced output. But who knows if the VIC-20 horizontal syncs during vertical blank were similarly flawed as on some PAL C64 models.
    – Justme
    Nov 27, 2022 at 14:17
16

As I had understood things, "interlaced" video was nothing but a clever way to save data bandwidth

Not really, it's a way to give more timely resolution, on TV 'half' pictures are not two halves of one picture - as in one picture taken and then send as two parts - but two independent pictures displayed with an offset. That's why deinterlacing of movies does, without further processing, show zagged borders(*1).

The confusion may come by mixing up the term 'Frame' and 'Picture'. A 'Frame' is a data structure. One 'Frame' holds two independent 'Pictures'. The TV set will display them consecutive giving a 50 (60) Hz picture rate, but ofsetted by one line height (*2). The whole idea is to get a 50 (60) Hz picture rate and a virtual 576 (480) vertical resolution while only having to spend bandwidth for half of that.

But why would a game console or anything else that is connected locally to that TV be sending it an "interlaced" rather than "progressive" signal?

Because 'interlaced' is the way the TV worked. What would be the point to deliver a signal to a TV it can't understand?

Beside that most consols/home computers didn't work interlaced, but presentet the same 'half' picture twice, giving a more stable display.

Since there is no bandwidth being taken up in the air, what is the point?

It's the same bandwidth taken up on the cable. It was a long way until TV became fast enough to handle what is now called progressive.

I'm not getting about why "interlacing" is used, since it is apparently not just for saving radio frequency bandwidth.

Because it's the technology the whole production and distribution chain worked. Same way as you would have no success transfering data using a high speed USB storage if there is no high speed USB interface - or no USB at all

I am also confused as to why they chose the terms "interlaced" and "progressive" instead of just "half" and "full" lines.

(*2)

'Progressive' is a term defined in hindsight to mark the new way of handling electronic motion picture. Also, it wouldn't be 'half' and 'full', as the lines transferred are always full, it's rather 'even' and 'odd' lines.


*1 - Well, for TV productions that is, as movies scanned from film will have (most) pictures scanned twice, as film is tone in 24 pictures/s, thus recombining them from TV will (taking account of the additional inserts) will again reproduce full resolution.

*2 - Note that the term line is used in two different ways when describing a TV picture. There are

  • TV-Lines(TVL) describing horizontal resolution, not vertical. They are essentially describing Bandwisth. And
  • Scan Lines which are created as part of the picture scan process serializing the content.
7
  • Like you say, most home computers did not send interlaced signal, they did send progressive. TVs can understand and show progressive even if the standard TV broadcast is interlaced. Movies may mean actual cinema movie shot at 24p, so in PAL countries both even and odd fields would be taken from same frame of movie. Content shot with TV camera indeed would have completely independent even and odd fields temporally, which gives the so called soap opera effect.
    – Justme
    Nov 26, 2022 at 21:40
  • @Justme The fact of movies scanned from file having two halfs of the same picture transfered are due the slow(er) frame rate of the source material, not due the TV part. Also, Home computers producing twice the same half frame does still not construct a progressive picture - that depends still on the TV set.
    – Raffzahn
    Nov 26, 2022 at 21:47
  • 1
    A standard CRT will phase lock to any vertical retrace timing within tolerance. Vertical retrace is supposed to be placed so as to include 312.5 or 262.5 lines per field, producing an interlaced image, but you could place it at 312.25 and have an interlace repetition period of four fields, or 312.125 and have a period of eight fields. That home computers mostly went 312, 311 or 313 (or NTSC equivalents) for what is effectively a progressive signal just makes sense in terms of how much data a home computer can store and manipulate.
    – Tommy
    Nov 26, 2022 at 23:55
  • 2
    What you call "pictures" here is usually called "fields" in my experience.
    – Hearth
    Nov 27, 2022 at 17:43
  • 1
    @Hearth No. Picture means one single image as content unit to be encoded within a frame. A field is only describing part of the data structure called frame, not the relation to content. One frame can hold either two halves of the same picture in either field - like when scanned from a film. Or hold two independent pictures one in either field, like when taken from a TV camera.
    – Raffzahn
    Nov 27, 2022 at 19:01
14

CRT TVs basically understood one single SD video format which is sent over antenna (or cable) and that format is local to the geographic are you lived (NTSC, PAL, SECAM for most people). They did not work with ED or HD formats. So the TV will only show signals that they can lock on. They can't show arbitrary signals that deviate much from standard they are built to.

It means line frequency of 15.625 kHz and field rate of 50 Hz for PAL/SECAM and line frequency of 15.734 kHz and field rate of 60 Hz for NTSC. Note that I use the terms PAL, SECAM and NTSC here incorrectly, as they are actually methods of colour encoding to video signal, but normal people use them to refer to 50 and 60 Hz video signals.

So you can either send a 240p or 480i signal to an NTSC TV and 288p or 576i signal to a PAL TV. The operation of the TV is not really exploited, you simply send it an interlaced or progressive signal and the TV will follow and show the signal as sent to it.

So it depends on your game what you want to send. For title screen or static game graphics you might want the interlaced for more resolution, and for moving high paced graphics you might want flicker free progressive.

There is exactly same amount of pixels pushed per line and per field, so the signal does not depend on video hardware speed. However, it requires double the video memory storage to store 480 lines than 240 lines. And it takes double the CPU time to generate graphics for 480 lines than 240 lines. Higher resolution graphics require also more disk storage space.

Interlaced signal then just shows the even 240 lines in one field and the odd 240 lines in another field that are timed so that odd lines are drawn between lines of previouly drawn even lines.

You could just draw the same 240 lines twice, as both odd and even frame, but it might show as flicker. So for example an Amiga which can send both progressive and interlaced can select how it wants to send the signal.

If you ask about terminology, the terminology is progressive and interlaced for the signal. Progressive signal sends each line in n increasing numerical order, so in progression. Interlaced sends lines with even number first, and then returns back to draw the odd lines between the even mines. I don't understand why you would name them full or half lines. Sending full lines all the time means signal is progressive. Interlaced signals send also full lines, except the last line of a field is a half line, and it is necessary if you need to send 525 or 625 lines per full frame interlaced, you need to send 262.5 or 312.5 lines per field.

But a 1080i TFT could have 540 lines and each video field of 540 lines, both odd and even, are just drawn on the same 540 lines of TFT.

9

A real CRT is a continuous drawing area. On a theoretical perfect CRT, the painting of each field would touch every point of the area exactly once, creating an evenly-lit and completely continuous image.

Because the electron gun sweeps very close to horizontally, there's a discrete element to the vertical sampling.

In interlaced video, the sample points from one field to the next are offset by half a line vertically, and are painted with the same offset.

So the objective is to provide the brain with more vertical information, it being smart enough to combine what it learnt from one field with what it learns from the next.

It is a myth that the objective is to serialise one frame in two halves; each field is a unique sampling, captured at a different time. In completely-traditional television, the image is captured exactly as the raster runs so every point on screen is actually captured at a different time.

It is also a myth that gaps are left between the near-horizontal lines in one field so that they can be filled by the next; the rate of phosphor decay means that the screen is never completely lit even within a single field. Less than 1/5th of the screen is lit at any time, so the lines from the previous field are long gone by the time the next appears.

Interlaced video therefore is not so much a clever way to save bandwidth as a clever way to communicate greater detail within the same amount of bandwidth. Which might be splitting hairs, but I think captures the nature of the signal being generated — interlacing doesn't degrade something, it enhances it.

So why use it in a closed system? Besides complying with standards if applicable, in order to provide more vertical detail within the same amount of bandwidth — whether the limit be RAM speed or size, display scan rate, or the nature of the video connection.

The terminology of half and full lines also doesn't really make sense. All fields, interlaced or progressive, contain only full-height lines that are intended to paint the entire display. All interlacing does is offset each field by half the height of a painted line relative to the previous.


For as long as it remains online, check out this very high-speed video of a CRT displaying the most-definitely non-interlaced Super Mario Bros. Of note:

  • the lines are very close together; and
  • the phosphors decay really quickly.
2
  • 2
    There is an important point here: the CRT sweep is not exactly horizontal. In fact, over the length of a line it drops almost to the start of the line after next. This is hardwired in the control circuitry of TVs, and understanding this fact makes understanding how interlacing works much easier.
    – occipita
    Dec 4, 2022 at 0:05
  • 1
    To build on @occipita's comment: the slight diagonal is why interlaced video works because it means that the placement of vertical sync within the line affects vertical deflection when the raster re-enters the top of the screen. Placing the sync halfway through the line will cause reentry to be half a line higher.
    – Tommy
    Dec 5, 2022 at 17:53
5

Hardware with a given level of complexity and cost will be limited to processing some number of pixels per second. Pushing more pixels per second will require more complex/costly hardware.

Of the three choices 1080p@60fps, 1080i@2x30, and 1080p@30, the first will generally offer the best results, but at the highest cost. While there are a variety of reasons why 1080p@30 might be preferred to 1080i@2x30 when recording, they're generally far less important on playback. If one has a good recording but currently lacks the equipment to play it back optimally, the playback experience can be improved later by improving the playback equipment. If instead one has an inferior recording, the quality of the playback experience will forever limited by the quality of the recording.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .