Unanswered questions on processing and compressing DV-AVI

[sd_smoker]

The only thing I did on the source in Full Processing Mode was demux the audio. Would that have converted to RGB? Other than that the source is untouched since I captured it off the camera using WinDV so why would it show as RGB24?

Edit: I just installed the Cedocida codec and now when I run AviSource().info in AVISynth on the source file it says that it's in YUV colorspace. I'm clearly not understanding how this works. Nothing has changed with the source but depending on what DV codec you have installed the info function gives different results. What gives?

[jagabo]

If you are using Panasonic DV Codec and AvisSource(): Panasonic's DV decoder always converts to RGB with luma expansion. Use Cedocida instead and allow YUY2 output.

At what point does this conversion occur?

Pansonic DV Codec converts to RGB before it hands frames to AviSynth or VirtualDub.

My understanding was that VDub converted to RGB when VDub is run in Full Processing Mode.

If VirtualDub receives YUV video it will convert it to RGB when you enable Full Procession mode or Normal Recompress mode. In Fast Recompress mode it will pass the YUV to the encoder. But if you use Panasonic DV codec VirtualDub will not receive YUV video, it will receive RBG. Even working in Direct Stream Copy or Fast Recompress modes will result in RGB output.

Does it convert to RGB when all you do in VDub is open an AVISynth script?

If you're using Panasonic DV codec, the decoder is converting to RGB before AviSynth gets the video. Then AviSynth will pass the RGB video to VirtualDub.

Stop using Panasonic DV Codec and use Cedocida instead. Be sure Cidocida has YUY2 output mode enabled. Then VirtualDub or AviSynth will receive YUY2 video from the decoder.

What you're missing is the fact that many video decoders can output to several different formats. YUY2, YV12, RGB, etc. VFW and Directshow negotiate for the best color format supported by the video decoder and the receiving application. But Panasonic DV Codec only outputs RGB. There is no way for a program to get YUV data from it. An application might say "I prefer YUY2 but I'll also accept RGB." VFW or Directshow will first ask the decoder for YUY2. If the decoder says "no", it will fall back on the applications second choice, in this case RGB.

[2Bdecided]

You don't need to deinterlace for colour correction.

You do normally need to deinterlace for noise reduction. It depends on the denoiser (some are "interlace aware"). That "separate fields and process them separately" method you linked to is pretty hopeless for some denoising (it depends on the noise) - better to bob or smart bob, and then re-interlace later.

Cheers,
David.

[sd_smoker]

jagabo: Great info, thanks for taking the time to explain. I've installed the Cedocida codec so this shouldn't be an issue any longer.

2B: Thanks.

Quick question about AVISynth. When you load a script it obviously executes AviSource() immediately because the first frame of the video shows up in VDub, but does script execution always stop at that point until you "Save as AVI"?

[gadgetguy]

The entire script is run for each frame presented. Even if some filters only apply to certain segments, it still runs the the entire script to determine that. This is why exceptionally long scripts, even with only simple filtering, will take longer to execute.

[sd_smoker]

Ah, okay. Is VDub still able to display the input and output separately so you can see before and after or does that only work when filtering in VDub?

[gadgetguy]

Vdub sees what Avisynth sends it (after processing) as the input. The output window only shows after additional processing in Vdub.

[jagabo]

Note that AviSynth sends frames one at a time to VirtualDub -- when VirtualDub requests them. When you first open a script VirtualDub asks for the first frame. That first frame passes through AviSynth's processing and is sent to VirtualDub (depending on the script AviSynth may actually read many frames from the source file before it sends one to VirtualDub). When you scrub through the video VirtualDub will request the corresponding frames, one at a time. AviSynth will read the requested frame, process it, and pass it to VirtualDub. When you get around to encoding VirtualDub will request the frames one at a time, in sequence, until all frames have been read.

[sd_smoker]

Thanks. I think I'm going to start another thread asking for people to share AVISynth scripts specifically for dealing with DV camcorder video (as opposed to broadcast, VHS, etc. captures). This specific info seems really hard to find. I'm still open to doing all processing in VDub if someone has something that works really well, though.

[sd_smoker]

Let me ask you guys this. If I decide to keep my videos interlaced is Xvid a better option than DivX? Seems like I read somewhere that DivX should only be used for dienterlaced video.

[jagabo]

Xvid and Divx both support interlaced video now. But interlaced video isn't real convenient with them. You have to manually tell a player to deinterlace when viewing.