Posted on July 7, 2011 at 16:30 by Kevin Chung in Technology
In the current series of five blogposts we discuss a new concept/method to preserve the visual quality of video. The series consists of the following topics: Interlacing versus Progressive, Format war, Telecine, NTSC versus PAL and Television manufacturers. The central motto: “simplicity is brilliance”. This week part 3: telecine.
Previous posts in this series:
Part 3: Telecine
Film cameras capture pictures at a rate of 24 frames per second. To match the specifications of the NTSC standard telecine has been invented. The process of telecine yields 30 fps output from 24 fps source material. If you want to be specific 29.97 fps (rounded). I don’t like that number but that’s something for part 4, next week.
The application of telecine involves duplicating fields from odd frames. When frameserving you see a pattern of 3 progressive frames and 2 interlaced frames. This process is completely reversible by doing an Inverse Telecine.
In a previous blogpost I mentioned that if interlacing wasn’t invented at all then the current implementation of telecine would not exist either. Telecine is very prone for errors caused by people editing them. How many times haven’t I seen people encoding a telecine stream as progressive or deinterlacing a telecine stream? Or even worse: a “hybrid video” consisting of mixed film and video footage.
Motion
If you want to keep motion consistent, you would at best need to go through the entire video manually and decide where to apply IVTC or deinterlace and create a timecodes file. This would ultimately create VFR video which I also don’t like.
VFR is asking for problems. Because without the timecodes file you’re getting nowhere and even if you have it and intend to make it CFR then you’d need to insert duplicate frames or delete it, therefore hurting motion. In preserving the quality of video, motion is important, not only the visuals. It’s unacceptable to have motion that’s jerky.
DVD movies are the classic example where pulldown is applied. In most of the cases a hard pulldown is applied as opposed to soft pulldown. I’d prefer the latter because that leaves frames progressive which is more redundant as I mentioned in part 1 of this series. For soft pulldown flags are used to tell the DVD player when to duplicate fields during playback.
This takes a lot less storage because they are just instructions. Therefore you also gain more headroom to use a higher bitrate to encode your 24 fps progressive footage instead of 30 fps. Or to be compliant to NTSC specifications 23,976 fps and 29,970 fps.
There is a dozen more pulldown techniques which I won’t cover here.
Frame rates
Can you see how ridiculously tedious this is? It seems telecine is even worse than interlacing. So I’ll refer back my earlier plea: keep it progressive! Like some TVs do today it’s better to have your screens refresh rate matched to film video at 24 fps, which is 24 Hz or a multiple of it.
Frame rate conversions should be avoided at all cost to keep motion consistent. Having two widely used standards, NTSC and PAL, with specifications that are somewhat incompatible with each other is a huge loss. Especially for file exchange, like I mentioned in part 2. If telecine is also sent to oblivion, that is one huge headache less and it makes working with video less error prone.
As you can slowly see when I make examples, fundamental blocks of video affect each other greatly. Can you see the connections I keep drawing to previous posts? They relate to each other. And the key to success in highlighting issues and redrawing the concept? Simplicity.
Because simplicity is brilliance!
