Why did some video games apparently use "interlaced" video modes?

Question

I was under the impression that "interlaced" video modes was only a thing for remote television content because it saved bandwidth to only send 50% of the data to the homes, so you could fit more TV channels in the same bandwidth (air or cable).

I thought that all video games, being local to the customer's home with a video cable between the console and the TV, used "progressive" mode, meaning "not interlaced", since there was a dedicated video cable between the two units. Why would it have to send "half the data" in that scenario?

But then I watched a video on YouTube where it was casually mentioned that some games used "480i" instead of "480p", without further explaining it. Apparently, this was done very late, so it was not some kind of early 1980 video game console technique either.

Why would any game console or individual game send an "interlaced" picture to the TV when it's right there above the console? I assume that the picture must suffer in some subtle way from this "only every other line of image data", but it must be far more noticeable for video game content compared to a movie or TV show.

Answer 1

Indeed, early devices such as C64, NES, or IBM PC with CGA adapter did not use interlacing, but simply sent 240p to the TV. And later devices such as the Amiga could send either 480i or 240p.

But TVs were not 480p capable, only 480i or 240p. So it was not possible to use 480p.

For example, Amiga 500 can send either interlaced 480i for hi-res graphics and progressive 240p for low-res graphics, and the game is free to select the mode anytime. This can be used for example show a static title screen in 480i for high resolution and the moving game action in 240p for lower resolution without flickering. The reason why switch to 240p, instead of sending 240p content as two 480i fields, is to avoid the 240-line content from flickering, as it would look like as if the same 240-line content is jumping up and down by half a line for each 60 Hz field.

While early gaming devices used progressive 240p signal only, later on, some games on newer gaming devices used interlaced signal to display even the game in interlaced 480-line resolution.

Then some background:

TVs can only accept a video signal with single horizontal frequency of about 15.734 kHz (or 15.625 kHz).

Standard TV programs are interlaced 480i (or 576i) to have high screen update rate of 60 Hz (or 50 Hz) to prevent flickering and to achieve a high resolution of 480 lines for static images. But it also means that the image is really sent as two 240-line (or 288-line) fields at 60 (50) times per second, with half a line offset, so basically a full frame of both interlaced fields is sent at 30 (25) Hz.

Gaming consoles are free to send any kind of signal they want but the TV must be able to lock onto it. So gaming consoles must also use same horizontal rate and vertical field rate to send the video.

It means that the progressive signal that is comparable to 480i is not 480p, but 240p. There is no flickering as only one field of video is sent so the video lines drawn are on top of each other. The difference is that if game screen has 240 lines but sends it as interlaced 480i, there can be some flicker seen due to the same lines being drawn as interlaced, the picture seems to jump up and down by one line of 480i (half line of 240p).

(I used 480 and 60 Hz since that's what the question asked. For differences between 60 Hz and 50 Hz video, replace 480 with 576 and 240 with 288 and 60 Hz with 50 Hz).

To rectify some misconceptions that seem to be very common:

• Signals sent to TV do not need to be interlaced
• Using RF modulator does not mean signal must be interlaced
• There is a difference between interlaced signal and progressive signal formats, regardless of what content is sent
• Interlacing means sending 480-line content as 240 even lines in one field and then the remaining 240 odd lines between them in another field, with field rate of 60 Hz. Signal has 525 lines divided into two fields of 262.5 lines. A full frame is thus drawn at 30 Hz, but that usually not important.
• Progressive signal does not mean sending 240-line content twice, as two interlaced fields that are offsetted by half a line in respect to each other. This is of course possible, but it's still interlaced signal, and this would look flickery.
• Progressive signal means sending integer amount of lines per field, so that there is no half-line offset, so lines of each field are drawn on top of each other without half-line offset, and there is nothing drawn between these fields like in an interlaced signal.
• Devices like Commodore 64, NES, and IBM CGA card all send out progressive signal of 262 lines per field.
• Devices like Amiga or PlayStation 1 can switch between sending 240p and 480i signals, i.e. 262 lines or 262.5 lines per field.
• Link proving progressive 240p signal exists and devices use it
• Link proving NES video timing is progressive
• Link proving C64 video timing is progressive

Answer 2

TL;DR:

It's not about what a console can deliver as it is what a TV can display. Classic (pre digital) TV sets could only receive and display interlaced frames. So no sense in producing a non interlaced one.

Similar console developers would have been unwise to create consoles and content that could not be displayed on what Joe Aerage had as TV. Sales might not have been geat.

(In addition, therms like 480i or 480p are only defined in hindsight from a digital POV, and have only a restricted meaning when it comes to analogue TV)

---

In Detail

I was under the impression that "interlaced" video modes was only a thing for remote television content because it saved bandwidth to only send 50% of the data to the homes, so you could fit more TV channels in the same bandwidth (air or cable).

No. Well, not directly. Of course, everything in transmission is always about bandwidth, as that's the shared resource. But it wouldn't have been a big deal to double channel width in the early days.

Interlaced 50(60) Hz delivers 25(30) pictures full pictures per second. In theory more than enough for moving pictures s cinema does fine with 24 pictures per second. Except, TV works complete different from cinema. A cinema frame is displayed on the screen once and in al regions (almost) at the same time, but in TV a dot paints the picture line by line on the screen.

Doing this without any additional measure would result in a quite flickery picture, despite a frame rate equal or better than cinema.

The chain of thoughts looked a bit like this:

• First stept: Make the screen disperse the beam energy over time (aka afterglow), so lit parts stay longer lit.

• It should not stay lit too long, as that would make any motion blurr

• With 25 (30) fps this would still result in an uneven, badly lit picture with a noticible fade (*1).

This is, BTW, the similar what makes pure 24 fps cinema flickery, like known from old silents. To reduce the inter frame flicker each frame later got displayed two or three times (*2) before moving to the next.

• Drawing a revived picture twice would have worked as great on TV, except a CRT got no storage, so it had to be send more often than 25(30) times per second

• Thus doubling the transmitted frame rate was the way to go.

• Doubling allowed to reduce the persistence allowing to reduce blur

• Doubling improved fast motion as well

So far a nice solution. A picture with 200-300 horizontal lines is quite fine for TV and would satisfy all needs. Producing TV content at 50(60) Hz frames would have been no issue. But then there were movies. They were produced on 24 Hz, so transmitting every picture twice would work quite as good as genuine TV production, but waste about half the bandwidth.

Adding interlace allowed to increase resolution without increasing bandwidth.

Thanks to the wonders of human sight movies now used the 50(60) fps to display their 24 fps in higher resolution, while genuine TV content got real 50 (60) fps in kind o an enhanced resolution.

I thought that all video games, being local to the customer's home with a video cable between the console and the TV, used "progressive" mode, meaning "not interlaced", since there was a dedicated video cable between the two units. Why would it have to send "half the data" in that scenario?

Because that's what the TV set expects as input and is able to display. A (classic, CRT) TV set does not combine two 'half' pictures into a single one to be displayed at double fps, but displays each 'half' on it's own, the second offset by a line. It's the way a CRT hardware works.

But then I watched a video on YouTube where it was casually mentioned that some games used "480i" instead of "480p", without further explaining it. Apparently, this was done very late, so it was not some kind of early 1980 video game console technique either.

Because games need to be compatible to the TV hardware (and standards) out in the field. It's not a great idea to produce a console that is can only work with the very latest displays - at least not if one intends to sell more than a few.

I assume that the picture must suffer in some subtle way from this "only every other line of image data", but it must be far more noticeable for video game content compared to a movie or TV show.

Not really, rather the other way around, as a console game is able to adjust. Keep in mind, 480i or 480p does only tell part of the story, as that number only names a frame size and structure, not the frame rate. Interlaced will usually create 50 (60) fps, while the same as progressive can be 25 (30) or 50(60) fps.

A classic TV set expects 50(60) 'half'-frames per second, so a console can use this as

• 50 frames per second with 'half' vertical resolution, or
• 25 frames per second with 'full' vertical resolution.

So any game that is about fast action, like a race game or a (complex) shooter will create each frame separate and get a smooth 50(60) fps with at cost of some resolution (288/240 horizonal). In contrast a game that is more about beautiful pictures, or lots of content may go ahead and do 25(30) 'full' frames using full 576 (480) horizontal lines (*3).

Oh, and of course games could go ahead and do fast content with high resolution - with the same drawback as TV content had with motion artefacts.

---

*1 - Yes, that is in fact related to the wandering dark bar that is visible when an unsynchronized camera films a CRT TV.

*2 - Or more exact, the douser that covers the film movement between frames also cuts one (two) more times within a frame, transposing the flicker to 48 or 72 Hz, where it gets way less noticible.

*3 - In reality this wasn't uses as often due memory limitations to generate and hold pictures that large.

Answer 3

CRT TVs were designed to handle interlaced signals, where the TV alternates by receiving odd scanlines and even scanlines on alternating frames. The so-called progressive mode was invented when some hardware designer noticed you could start the even and odd frames on the same scanline, so the CRT's electron gun overwrites the previous frame's lines instead of drawing between them. This doubles the frame rate and halves the vertical resolution, both of which were useful for game machines (since there usually wasn't storage space for high-res graphics). This is also why old game consoles exhibit the scanline effect, where there are little gaps between rows of pixels.

Interlaced signals don't have this problem because those gaps get filled on the subsequent frame, at the expense of making the image more jittery (an effect known as interline twitter), which is usually smoothed out by applying a blur, but video games didn't generate such blur. So that's another reason games used progressive scan.

Answer 4

To be exact: "Interlacing" is not just a method for bandwidth saving, but mainly for increasing vertical resolution.

EDIT: Bandwidth saving and increasing resolution are just the different sides of the same coin, see a comment by Justme.

E.g.: The TV screen (in Europe / 50 Hz) was divided into 625 lines. The first picture frame contains the odd lines (1, 3, 5, ...) from the upper left corner, the second picture frame contains the even lines (2, 4, 6, ...), started from the top middle of the screen. Two frames give one full image, so the European TV has 50 frames per second or 25 full images per second.

Video games, as well as home computers, did not use interlacing, they generated only "odd" picture frames, not respecting neither interlacing nor shifting. Old TVs were more robust and tolerant to the video signal, so they can handle such "non-standard" signals with no big problems. So video games have got 50 pictures per second, but only 312 possible lines to display. In fact, less, because some of them are outside the viewable area.

The main reason to use interlacing in a video game could be increasing the vertical resolution up to 600 (approx.) lines. On the other side, interlacing has some drawbacks, as image flickering. You have to use more subtle video generating circuits, handling with the even/odd frame distinction, etc.

N.B.: The USA and Japan and some other countries used 525 interlaced lines (262.5 with no interlace) and near 60 Hz frame rate (59.94 Hz in fact). Those 525 lines were not fully visible, only the 480 lines were visible, so "480i". In fact, there was no "480p" mode in the 80s', so technically it WAS 480i, but used as "240p"...

Answer 5

Two reasons come to mind:

1. The image was sent to the display (the TV) using an RF modulator. This essentially acts as a low power TV station. Since the TV expects broadcast channels to be interlaced, the signal sent from the RF modulator must be interlaced as well.

2. You alluded to this in your question: Reduced bandwidth. By only displaying every other line, the number of pixels that need to be drawn per frame is cut in half. Instead of displaying a 640×480 image 50 or 60 times per second, one can instead display a 640×240 image at the same rate.

Answer 6

Justme’s answer (and commentary with supercat) explains this the best. I’ll just add a couple of notes:

1. With standard CRT TVs, the choice was never between 480i and 480p but between 480i and 240p. 480p would have required doubling up the horizontal scan rate (the electron beam sweep rate from left to right and back) whereas 480i and 240p have approximately the same horizontal and vertical refresh rates and bandwidth, save for a very slight difference stemming from the removal or addition of a half line. (480i and 240p essentially draw the same number of scanlines per second whereas 480p would have required drawing two scanlines in the time period where 480i or 240p only draws one, considerably changing the required design specifications for the TV in question.)

2. As stated many times by Justme, CRT TVs are fairly tolerant toward signals which deviate from the exact broadcast timing standards. If they weren’t, devices such as the domestic VCRs (which often have quite variable time base due to the tape transport not being that accurate) would have been impossible. There are limits to this tolerance, of course, but the slight deviations used by TV-connectable home computers and video game consoles fall well within those limits.

3. Modern LCD TVs, on the other hand, are often less tolerant. This is basically just laziness on the manufacturer’s part. The manufacturer may have “forgotten” about 240p and possibly never tested how their product will synchronize with the non-interlaced signals generated by older computers and game consoles. (All in all, analog signal inputs are now disappearing from new TVs. As time marches on, you will need a specialist device to handle the necessary analog signal capture, upscaling, and generation of the desired simulated CRT raster/scanline effects.)

4. 480i / 240p are the nominal “active” (picture-containing) scanlines for 525/60 systems (North American System M, commonly referred to as “NTSC”). The rest of the scanlines are spent in vertical blanking, waiting for the electron beam to move back to the top of the screen. Scanlines belonging to the vertical blanking period are not shown on screen and may carry data, such as closed captions. The corresponding active scanline values for the European 625/50 systems (System B/G/I with PAL or SECAM color encoding) are 576i / 288p. In these systems, the vertical blanking period typically carried Teletext content.

5. The NTSC vertical refresh rate is actually not 60 Hz but 60*1000/1001 Hz. It used to be 60 Hz back in the days of black & white broadcasts but the addition of color (as an afterthought) required adjusting the vertical refresh rate slightly. This was, again, well within the tolerance of the CRT TV sets. The European 50 Hz systems implemented color in a different way and did not require such adjustment to the field rate.

6. People often assume 480i would mean ~30 fps (29.97 fps) imagery drawn onto the screen in two passes (the odd and even lines of the “frame” drawn separately; the “frame” being an individual, full-resolution still image).

   While you could generate video imagery in this specific way, this is not how video cameras do it. Video cameras (TV cameras) have always recorded motion in each pass, giving interlaced video the motion resolution of 60 Hz (59.94 Hz). It is only when you shoot still scenes (or scenes set against a still background, or still objects against a moving background) when the vertical resolution of the individual fields seems to combine into a single image, even though this is just clever trickery of engineering and the brain: the fields belonging to the same “frame” are captured from different points in time.

   Broadcast (SD)TV (video) is just a succession of alternating odd and even fields. “Frame” is not that meaningful a concept in interlaced video, except when editing (you obviously cannot splice together two odd fields but need to retain the correct field succession and cut at frame intervals) — or when considering some aspects of the analog color encoding standards.

7. In fact, the old, video tube-based TV cameras (preceding CCDs) recorded motion not only at field intervals but as continuous scans across the scanline raster, in the exact same scanning pattern as how the electron beam in the CRT-based TV receivers drew the images on the CRT screen. If a vertically-oriented rod passed the camera field of view in the horizontal direction during a single video field, the old-timey TV camera would have captured a distorted diagonal image of it. Things in motion were wobbly in much the same way as images are wobbly when shot with a “rolling shutter” CMOS DSLR. On the other hand, the electron guns of the CRT TVs displaying these images from a live studio show were, at each moment, essentially synchronized to the scanning image capture of the tube-based TV cameras in the studio — which is kind of neat, if you think about it.

8. Old TV-connectable computers and game consoles avoided interlacing probably for a multitude of reasons: 240p was a better match for their technical capabilities (limited video memory), a non-broadcast-standard signal was simpler to generate, and interlacing is not all that desirable in the first place as it makes all the horizontal lines noticeably jittery and tiresome to look at for computer use (a major headache e.g. for word processing).

   The jitteriness of an interlaced signal is a big issue especially when you have a limited palette (limited number of simple primary colors with no shades to smooth things out) and need to display non-natural, static, crisp imagery (such as text and charts), and have no advanced video processing or filtering capability.

   Back in the 1980s, some computer monitors were specially designed for working in interlaced modes and featured “slow phosphors” which retained the image from a single field scan longer than the usual, reducing both the interlace flicker and any perceived flicker related to the field refresh rate in general. They were marketed as being more ergonomic than standard monitors. But as you can guess, slow phosphors only suited relatively static screen content as they caused moving objects to have a noticeable motion blur trail behind them. Also, scrolling any content on a slow-phosphor monitor was a pain. It was better to browse text files one screenful at a time.

9. As for TV-connectable computers and computer-generated interlaced signals, the Amiga used to be one of the rare exceptions of its era, offering broadcast standards-compatible interlaced modes and genlock capability (capability of synchronizing to an external video signal). But even on this system, the interlaced modes were primarily used for video titling applications, not much else. The rest of the day-to-day use mostly happened in 240p/288p — but with vertically elongated, non-square pixels, so you could cram more stuff per scanline and implement more serious (80-column) applications than what the previous generation of TV-connectable computers (the 40-column-or-less 8-bitters) were typically capable of. The Amiga treats interlaced modes in a similar fashion to video cameras: motion updates in every field: adjacent odd and even field pairs that nominally belong to the same “video frame” do not come from the same point of time in the framebuffer state.

10. Later (CRT SD)TV-compatible game consoles — such as the original Xbox — embraced interlaced modes as a way of utilizing the perceived (SD)TV resolution to the max. But the Xbox implemented them using a video scaler chip (“TV out” chip) which sat in-between the framebuffer and the TV and had an advanced, adaptive filtering capability, making it possible to smooth out horizontal details over adjacent odd/even scanlines, this way alleviating the flicker. The Xbox also had a GPU capable of accelerated 3D graphics and shading with millions of colors, so the imagery it generated was closer to that produced by video cameras. The typical applications (games) also did not rely on the user having to intently watch static text screens or charts for long periods of time so the usage was different from that of a TV-connectable generic-purpose computer.

11. Some commentators assume 240p to have been inferior to an interlaced signal due to the scanlines of a non-interlaced signal (allegedly) having these huge, unsightly gaps between them. However, it would be a mistake to assume that a non-interlaced signal on a CRT TV would look anything like an image on an LCD screen with every second line missing. Scanlines are drawn closer to each other than is their “height” so the gaps are not quite that drastic.

    A better quick approximation of scanlines on an LCD screen is drawing every second line as a 50% dimmer interpolation from the line above to the line below. The images drawn by an electron beam on the CRT phosphors through a shadow mask are not tidy LCD pixel rows with exact dimensions or borders: the electron beam that draws the scanlines has a certain spread. Also, the brighter the image content is, the more the beam will spread.

12. There might also be a cross-pondian difference. On a European 625/50 CRT TV set, there are 19% more scanlines packed to the screen than on an American 525/60 CRT TV, so the gaps between them can be thought of being less pronounced, both in interlaced and non-interlaced video raster. That is, the scanlines are sparser to begin with on American TVs, and get yet sparser when scanned in a non-interlaced fashion. (But maybe the electron beam is also defocused more at the factory?)

    I do not have much personal experience on 525/60 (NTSC) systems but I have watched my fair share of computer images on a 15 kHz 625/50 (PAL/RGB) computer/video monitor. Scanlines were never too noticeable in the 625/50 video raster, even in the non-interlaced modes. 525/60 might be somewhat different case, though, due to the fewer number of them.