Deinterlacing video

[darryl]

Hey, I don't know if this is standard procedure for video, but I'm shooting with an olde Digital 8 Sony camera and dumping into an old version of iMovie for quick editing.

I've found that when I export as H.264 QuickTime files, that turning *on* deinterlacing results in much better results here on SmugMug.

Similarly, after recently editing and burning a CD, I decided to rip out and transcode the chapters using the great Linux tool, dvd::rip. Again, got much better results after turning deinterlacing on.

Otherwise you get lots of ugly motion lines whereever there's much movement.

*Shrug*. I don't really understand how this all works -- I think it has something to do with old tube televisions -- but it works for me. :-}

[i_worship_the_King]

The basic idea of interlacing was to cut down on how much data needed to be processed and displayed in one chunk. Basically it takes your screen/monitor and breaks it down into a series of lines and labels them ABABABABABAB... until there are no more lines. At any one 'pass' it would render either all the A's or all the B's. This process happens so fast that all we see is a single image.

Using the magic of compression, however, warrants using a 'deinterlaced' frame - one that shows both A&B data all at once. The reason is that instead of recording all the data the codec is only saving the changes to that data between frames. For instance if you have a 1 second clip of something static in the frame (say a photo, for instance) instead of writing 30 frames of data the codec writes 1 frame of data and tells it to repeat 30 times, saving 29 frames of space. Obviously that is a very simple example, but it should explain why almost anything in the digital age using a compression codec needs to be deinterlaced (otherwise the changes between the A&B frames would be tremendous, and consequently do not match up as well.)

For more info why not check out wikipedia?:D

[DavidTO]

Interlacing had to do with old TVs and the fact that there was a single cathode ray scanning the tube. Tied to the cycles in the current, it would scan the tube 60x per second. They broke each frame into to bits as stated above, ABABAB, and half the lines scanned in the first pass, the other half scanned in the second pass. This is an artifact that has less and less to do with anything these days.

When transferring film to tape (24fps to 30fps) they use the 3:2 pulldown method, which would end up with video frames that were 1/2 one film frame and 1/2 the other. I deal with this constantly in my work.

[darryl]

Thanks for the detailed info. I guess my point is, if your converted video has a bunch of lines in it when you load it up to SmugMug, try turning on deinterlace when you convert it.

I didn't see any mention of this in the various posts here, and thought it might be helpful.