Interlacing, Deinterlacing, and Everything in Between | Hacker News Reader

Interlacing, Deinterlacing, and Everything in Between

nicole.express

55 pointsMindGods23 comments

Technology Connections has a good video on how CRT’s work: https://youtu.be/l4UgZBs7ZGo

Captain Disillusion has a good video on interlacing specifically:

https://www.youtube.com/watch?v=5eu_KjKsnpM

> We hit a strange truth about analog monochrome television: the vertical resolution is strictly limited: it’s the number of lines you draw. But the horizontal resolution, the resolution within each line, is basically infinite; it’s a purely analog signal. The only thing that holds it back is how quickly the electronics in your television can keep up.

No, it's strictly limited by the bandwidth of the television signal. You could have infinitely responsive circuitry but you'd still have horizontal resolution limited by the bandwidth of the video portion of the broadcast TV signal.

About all you can say is that analog TV does not have pixels.

Anyway, here's a video (part of a series) on how analog TV works:

https://www.youtube.com/watch?v=l4UgZBs7ZGo

Full playlist:

https://www.youtube.com/playlist?list=PLv0jwu7G_DFUGEfwEl0uW...

>About all you can say is that analog TV does not have pixels.

Color analog TVs have discrete patches of different phosphors.

Those aren't really pixels though: the signal can (and does) vary within them.

What does industry do about interlacing?

I understand many TV shows were recorded and mastered on video with interlacing. When they re-release the TV shows for example for Netflix - what are they currently doing?

Last I heard, netflix has studios send them progressive video and lets them deal with de-interlacing.

AFAIK, for enthusiasts, the best deinterlacer is "QTGMC" http://avisynth.nl/index.php/QTGMC

For stuff shot like post 2015, I don't think anyone is interlacing anymore. Everything is progressive.

https://netflixtechblog.com/high-quality-video-encoding-at-s...

TV broadcasts are often still interlaced in some places

I sometimes find newly released deinterlacing failed production video on YouTube. I wonder how is it failed, and why the production recorded the original video on interlace.

Ideally, they deliver it to Netflix in the original form, properly marked as interlaced (or not) or for the small handful of shows filmed in 24p, as original films, marked as such (rather than presented with a 3:2 pulldown as originally broadcast).

In practice (based on experience as a viewer), things are kind of messy, and sometimes the markings must be off, or someone did a really poor deinterlace, and there's visible combing all over, ugh.

Netflix, or my streaming player, or my display can all deinterlace, so there's no need for the source to do it.

From what I've read, Netflix has been investing quite a bit towards automatically detecting these sorts of issues and rejecting video that's been poorly handled by the studios. AFAIK, they reject interlaced content.

From a compression standpoint, netflix wants everything to be progressive. Interlaced media (funnily) requires a lot more bandwidth to transmit with modern codecs. I say funnily because interlacing was a technique to save bandwidth for analog transmissions. It gives the illusion of a higher framerate. Unfortunately, modern video codecs work by tracking object movement from one frame to the next. Interlacing really messes that up requiring more bits spent for the same amount of quality.

Some of the older codecs have interlacing built into the standard (mpeg2 and h264). Newer ones have banned it all together (VP9, AV1) due to the codec complexity it introduces for very little benefit.

https://netflixtechblog.com/high-quality-video-encoding-at-s...

> It gives the illusion of a higher framerate.

I mean, it doesn't really give the illusion of it. It is higher frame rate when recorded from a camera[1], as most TV was. The fields are independent moments in time (and you can very clearly see this in freeze-frames of interlaced motion, that's part of what combing is).

What it gives the illusion of is greater vertical resolution.

[1] video game systems that used interlacing, like the n64, playstation, and up until the hdmi era, did not usually do this afaik, opting to instead drop the frame rate to 30fps with 2 fields per frame, so they really were in many ways the worst of all worlds.

> up until the HDMI era

I think that's wrong: didn't PS2-era consoles use interlacing to get higher resolution? You're definitely right for PS1 and earlier though.

I think what parent is saying is that camera-captured interlaced video was displaying scanlines from different images in each field, whereas game consoles using interlacing displayed the even/odd lines of the same image in two fields.

Thus parent's argument is that we can consider camera-captured interlaced video to be 60 (or 50 for PAL) fps despite being interlaced, but games using interlacing were 30 fps. It makes sense to me.

It was a choice of the game. There were definitely truly 60FPS interlaced games during the PS2 generation.

> video game systems that used interlacing, like the n64, playstation, and up until the hdmi era, did not usually do this afaik, opting to instead drop the frame rate to 30fps with 2 fields per frame, so they really were in many ways the worst of all worlds.

480i Nintendo consoles (N64, GCN, WII) all had many 60FPS games. The 30FPS limit was almost always because of the relatively underpowered hardware. Some N64 games even ran at 20FPS!

I wouldn't call the Wii a 480i console; it can run in 480p, 480i, or 240p, depending on system settings and the game itself (I don't think 240p is used much outside of a hidden mode in virtual console, but I could be wrong). I think the gamecube can too, but it requires the digital out port, and I think that was rarely used. N64 is 480i or 240p depending on the game.

Dont know about Netflix but I believe the BBC archive uses Martin Weston's adaptive 3-field deinterlacer. I think it's out of patent now and is available in ffmpeg and similar...

The BBC has an easier job than NTSC based outfits because filmed material will have been captured at 25 frames per second and converted to 50 fields per second. They are less likely to have problems like 3:2 interlacing breaking in an edit which used to be quite common in NTSC.

> Capacitors were still called condensers.

Funny that in German, a capacitor is still a Kondensator to this day.

The French word has kept the root too : "condensateur" !

Same in Japanese, コンデンサ.

Great pun in the title.

Reply on news.ycombinator.com