2
\$\begingroup\$

I have a lot of old analog video equipment and my trusty Apple IIe Color Monitor is finally dying. Since CRT's aren't produced anymore, and because I just am interested in how it is done, I obtained a decent plasma monitor (I can't get enough of those deep black tones). I want to know if an analog signal can be deinterlaced without resorting to a FPGA or similar overkill.

Can something to get rid of the interlacing effect on non-CRT's be made with components like 74-series IC's and proc amps (and maybe a AVR microcontroller), or do I need to go out and get some specialized chip from some specialized producer? All I want for output is progressive-scan composite video (although a better signal would be acceptable, I would prefer composite).

The most complex digital video generation hardware I have ever worked with was that of a Commodore 64 (converted it to PAL by swapping IC's with a european one, and replaced a few caps). My other experience was just adding a analog filter to a NES system. So digital isn't really my thing. I hate to think this is a dumb question, and hopefully, even if there's no answer, I am not the only one who wants this.

\$\endgroup\$
5
  • \$\begingroup\$ I highly doubt there's a simple solution for deinterlacing... Given the length of Wikipedia's page on this regard en.wikipedia.org/wiki/Deinterlacing \$\endgroup\$
    – Dzarda
    Jan 30 '14 at 22:33
  • \$\begingroup\$ Back-to-back TVP7002 chips or similar style of setup might be about the "simplest" option, with a small micro just to set the chips into the correct modes. \$\endgroup\$
    – John U
    Jan 31 '14 at 12:53
  • \$\begingroup\$ that sounds promising... I will look into it, but I wanted simpler than that. \$\endgroup\$
    – Wyatt Ward
    Feb 1 '14 at 21:09
  • \$\begingroup\$ If your analog equipment can generate component video, these el-cheapo SCART to HDMI boxes actually work surprisingly well. They have a video decoder, scalar (with RAM) and HDMI serializer built in - it's amazing what you can get for 30 bucks: ebay.com/itm/… \$\endgroup\$
    – Zuofu
    Mar 2 '14 at 13:53
  • \$\begingroup\$ no, @Zuofu , it only does composite. \$\endgroup\$
    – Wyatt Ward
    Mar 8 '14 at 20:11
1
\$\begingroup\$

Interlaced scan composite video signal isn't easily convertable to similar progressive-scan signal. You need to store the whole frame into some kind of memory and resend it at different speed using progressive scan. The simpliest way you will need a DAC-ADC pair with Fq about 6.5 MHz, an SRAM or ADRAM unit capable of running at 13 MHz clock and holding two frames, and the logic to control both input and output address. This way you will be able to store two interlaced half-frames and resend them back as progressively-scanned frames.

Everything said above applies to monochrome images as I don't know how any color-coding system works with progressive scan.

The image produced is likely to have visible interlacing artifacts, especially if there are horizontally moving objects in the image. It's a quite difficult task to remove them, but if you really want to reduce them, the better option is to have a PC with a analog video capture card and a program that processes and displays the image captured.

\$\endgroup\$
7
  • \$\begingroup\$ The PC solution will apply signifigant latency to the video, which may or may not be a problem. \$\endgroup\$
    – pjc50
    Jan 31 '14 at 11:26
  • \$\begingroup\$ I have a capture card (cheap roxio video capture device, which i forced to work with linux after hours of research) and a PC with a i5 cpu and a nvidia gt440. I cannot really use this though, because there's no way to output composite video from it and I don't want to be doing analog -> digital -> analog. However, I love your first idea. If only i had some SRAM on me besides in my NES cartridges :p \$\endgroup\$
    – Wyatt Ward
    Jan 31 '14 at 23:33
  • \$\begingroup\$ There are three kinds of chroma signals I've seen with progressive-scan video. Some devices output the same chroma phase every line (Apple and Atari computers did this; so did early C64's), so a solid color is a vertical stripe pattern; some output 262 lines per field with alternating phases every line (Karaoke players do that, and I think the NES does too), so a solid color becomes a stationary checkerboard. Some output 263 lines per field with alternating phases per line, so a solid color becomes a 30Hz flickering checkerboard (VIC-20 and later C64's). \$\endgroup\$
    – supercat
    Feb 2 '14 at 18:55
  • \$\begingroup\$ Broadcast interlaced video, or video generated from a DVD player, will have 262.5 lines/field with chroma phase alternating per line; this yields a checkerboard that moves vertically. Video from a consumer-grade VCR will often have chroma phase that varies arbitrarily from one line to the next, yielding wavy patterns (a VCR will try to output one scan line every 227.5 chroma clocks, but tape head speed is seldom as well-regulated as the chroma clock generator. If a line happens to start 227.4 chroma clocks after the previous one, its chroma will be offset by 0.4 cycles from... \$\endgroup\$
    – supercat
    Feb 2 '14 at 19:05
  • \$\begingroup\$ ...the previous line (rather than by a perfect half-cycle). Consumer-grade tape formats like VHS convert chroma signals to a format that doesn't rely upon phase when recording, and on playback generate a phased-color signal from that; thus, if a tape is played multiple times, the chroma phase relationships that are observed on one play-through may have little relation to any that are observed on subsequent ones. \$\endgroup\$
    – supercat
    Feb 2 '14 at 19:10
0
\$\begingroup\$

Conceptually, there is no reason why an LCD or other such display should interpret a non-interlaced signal as interlaced, rather than simply doubling every scan line [upon receipt of two or more "even" fields, start assuming that each "odd" field matches the even one until an "odd" field is received; do the reverse upon receipt of two or more "odd" fields]. I've seen a few displays that handled such things nicely, but most of them get it wrong. Still, the easiest way to have a display that works nicely might be to find one that just works or tolerate behavior that's not 100% precise.

I'm not aware of any Apple //e software where interlace-related issues would be likely to cause a problem. Color would be a bigger issue on that machine. Trying to build a color decoder that really works nicely might be hard, but modifying the motherboard to output a decent-looking component video signal should be pretty easy (basically tack-solder wires to a few points and run them to a circuit with a few discrete-logic chips and resistors).

The biggest places I would expect interlace-related problems are with Atari 2600 games, since many of those will display different images on alternate frames, not for the purpose of interlacing, but simply because the hardware and software aren't able to display everything on one frame. Such behaviors can also occur with some games for the Nintendo Entertainment Center and other such systems, but they're more common on the 2600.

\$\endgroup\$
7
  • \$\begingroup\$ Wasn't the non-interlaced signal simply the same frame written over the previous one? What I mean is, instead of 312.5 lines (PAL), there was 312 lines so that the 0.5 line didn't offset the electron beam to the odd frame. As you say, I believe many LCDs don't handle this correctly. \$\endgroup\$
    – carveone
    Jan 31 '14 at 12:36
  • \$\begingroup\$ The most common behavior I've seen from LCD sets when fed a non-interlaced signal is to arbitrarily decide one field is an "even" field and interpret the next as an odd field. If one were to on alternate fields show a message on the left and right sides of the screen (at the same Y position), either the left side would appear half a line above the right, or vice versa. On some sets, for some reason I can't really figure out, the fields would switch every few seconds. I doubt any of this behavior would be noticeable with any Apple //e software that wasn't designed solely to test it. \$\endgroup\$
    – supercat
    Jan 31 '14 at 15:52
  • \$\begingroup\$ a cheap chinese LCD i have does this; my video capture card does this too. AND I actually am using a NES. :D My //e monitor does not have this problem because CRT's handle interlacing well. Modern displays use special hardware, and I am wondering if I can make something quick and dirty to make it less visible, although maybe not totally gone. I do not own a //e; just the monitor. I had a commodore 64 and my friend sold me a //e monitor he had been given because he wanted the RGB kind and this was composite. \$\endgroup\$
    – Wyatt Ward
    Jan 31 '14 at 23:28
  • \$\begingroup\$ Also, people have been trying for years to mod a NES to use RGB video. The problem is that the PPU converts the signals it gets into composite internally and never exposes its input streams. So far the only solution has been to hunt down a rare Playchoice-10 arcade machine or a special console called the Famicom Titler, and swapping PPU chips. This is too expensive to be a viable option. \$\endgroup\$
    – Wyatt Ward
    Jan 31 '14 at 23:36
  • \$\begingroup\$ Many vintage machines generated composite video directly without generating any sort of RGB representation of the video. The issue isn't just one of what's exposed--an RGB signal might not even exist. If you're interested in gaming, I'd suggest acquiring some early-21st-century television set. If you can find some strong people to move them, some really nice sets can be had for practically nothing if not outright free. \$\endgroup\$
    – supercat
    Feb 1 '14 at 2:48

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.