DGIndex "film" vs VOB->VDubMP2 filtering

[BitBasher]

When I use DGIndex to IVTC NTSC video (29.97fps to 23.976fps) it's pretty much a set-it-and-forget-it feature. I never have to worry about lots of tweaking options.

If the source video is genuine film material encoded to ugly NTSC interlace, DGIndex does a TERRIFIC JOB of converting it to "film" frame rates, all nicely de-interlaced.

Now, lets say I want to feed a VOB file directly into VirtualDubMP2 (fccHandler's baby). What would be the best filter(s) to use in VDMP2 that would give me the equivalent results that DGIndex does?

I'm wondering if I need 2 filters, one to IVTC and another to decimate?

I'm hoping I can get the same excellent results that DGIndex provides, but I'm wondering if DGIndex has "more information" to work with than a VDub filter does - that is, can DGIndex detect and "fix" film material better than any VDub filter can?

I've played with some of the VDub and AVIsynth IVTC filters before, but I get lost with all the options and tweaky settings. If I have good film source material, what VDUB filters would be the best to IVTC/decimate (can one filter do this) and what would be good default settings?

Thanks for any info!!!
Bit.

[Mug Funky]

Vdubmod (not sure about vdubmp2) ignores pulldown flags.

this means no need for IVTC on pure film sources, and also means wildly out-of-sync audio on hybrid sources.

try avisynth with telecide in the above case (or in case of hard-telecine, where the 3:2 pattern is encoded as-is, rather than the encoder intelligently IVTC'ing on the fly).

[BitBasher]

Hmm, I need to understand MORE stuff. My question above is actually meaningless (for what I was intending to do), as to do IVTC/decimation in VDUB means that I would need to use Full Processing Mode.

I'm not really familiar with "color spaces". I understand that VOB files are generally in a YUV format, which is NOT RGB like my computer display, and that color space conversions are not always perfect.

I read (somewhere on the DivX site) that they recommend using the DivX compressor's cropping and scaling instead of VDUB filters because (they claim) that if I use VDUB's fast recompress option, then no color space conversion is done, and supposedly this can produce better results.

So, if I feed a VOB through DGIndex (using the forced film option, which works well for me), then if I use an AVIsynth script to feed VDUB frames, ala MPEG2Source("file.d2v"), and if I use DivX's cropping/scaling and VDUB's fast recompress, then is the video stream "suffering" from color space conversions?

I guess a more real-world question would be, does it matter? Will I notice a color space conversion, or is this a "purist" issue.

Lots of stuff in video to learn, so little time! :P
Bit.

[drcl]

you wouldnt notice the difference

the reason full processing mode isn't recommended i think is because its slower due to the conversion. I would direct you toward avisynth for processing. it works in yv12,yuy2 or rgb (a specific filter may have colorspace requirements)

as i understand it. FILM is stored on the disk as progressive 23.976fps. then the flags signal the software/hardware player to duplicate fields to acheive 29.97fps. So in realty a pure film source has no interlacing.

[Mug Funky]

though it's got more learning curve, IVTC in avisynth is the best way to do this (in my opinion of course :)).

colourspace conversions are noticable in flat areas - you'll get exaggerated blocks, flickering patches of noise, off-colour areas (drifting toward green and purple) and banding on subtle gradients (how many movies have scenes with a blue sky?). it's definitely best to avoid them.

once avisynth is set up nicely it'll actually be quicker that way, too. DGindex's avisynth template tool helps - it'll generate a script for you, so all you need do is install avisynth and a couple of tools (DGdecode you'll no doubt already have, DeComb is easy to find) and you'll be making nice encodes :)