IVTC - Best Method?

[wrawlind]

I have 3GB .avi that was telecined from 24 fps film into a 29.97 fps NTSC .avi. I want to do a high quality IVTC so that I can get back to the original 24 fps. What is the best and quickest way to do this?

The easiest method I've found so far is to go into the frame rate menu in VirtualDub and use the "Reconstruct from fields - adaptive" option. Unfortunately, I'm not 100% clear on the differences between "Direct Stream Copy," "Fast/Normal Recompress" and "Full Processing Mode." In "Full Processing Mode," VDub turns my 3GB file into a 20GB file! Is this the video mode I should be using? Also, I've read that VDub's IVTC is not the best quality for longer video files (mine's about 15 minutes long). Will there be a noticeable difference between this method vs another?

I also briefly toyed with Gordian Knot. Unfortunately, I've never used the program before and couldn't get it to read my .avi file.

Anyone know of any guides that can walk me through this relatively simple procedure?

Thanks in advance.
-R

[jagabo]

VirtualDub's Reconstruct From Fields Adaptive option isn't very good.

There are lots of automated IVTC plugins for AVISynth if you're willing to learn how to use it. These give pretty good results.

My favorite IVTC method for sources that aren't too badly mangled is VirtualDub's Reconstruct From Fields Manual method. If there aren't too many breaks in the IVTC pattern this will give you very good results. You have to encode in sections though -- where sections are demarcated by the pattern breaks.

VirtualDub Video processing modes:

Direct Stream Copy simply copies the compressed data from the source to the destination. No filtering, no IVTC.

Fast Recompress decompress the source to whatever colorspace the decompressor outputs (typically YUY2 or YV12) then sends that to the compressor. No filtering, no IVTC.

Normal Recompress decompresses to RGB but doesn't allow filtering. No filtering, no IVTC.

Full Processing decomrpesses to RGB, allows filtering and IVTC.

For all except Direct Stream Copy, you have to specify the output video compression codec or you will get a very large uncompressed video file.

[manono]

The easiest method I've found so far is to go into the frame rate menu in VirtualDub and use the "Reconstruct from fields - adaptive" option.

Neither the best nor the quickest. You'll have to learn some AviSynth (as jagabo says) in order to fulfill your requirements.

I also briefly toyed with Gordian Knot. Unfortunately, I've never used the program before and couldn't get it to read my .avi file.

Just what kind of an AVI is it? If you can open it in AutoGK, GKnot's little brother, it uses AviSynth and one of its IVTCs.

[wrawlind]

jagabo--
Thanks for spelling out the differences between the various video processing modes in VirtualDub. Yeah, I was trying to avoid having to learn AVISynth because I'm in a bit of a hurry and don't really have the time. I know there are some sample IVTC scripts out there so maybe I'll tweak a few of those and see what happens.

manono--
My .avi is compressed with SonyDigitalVideo. Looks like AutoGK is able to read it. The only IVTC option I could find was in the "hidden options" menu--"Force Normal IVTC." So all I need to do is click this option and run the job, right? It's running as I type.

Thanks for your help, guys.

-R

[jagabo]

Lately I've been using TFM() from the TIVTC package:

http://web.missouri.edu/~kes25c/

Usage can be a simple as:

AVISource("input.avi")
TFM(order=0) #0=BFF, 1=TFF
TDecimate()

[manono]

Hi-

So all I need to do is click this option and run the job, right? It's running as I type.

To tell you the truth, I'm not entirely sure if it analyzes AVIs the way it does DVD/MPEG sources. Ordinarily it'll run a full analysis to determine the kind of video, and if it needs it, will IVTC. So you usually don't have to force an IVTC as it'll figure it out by itself. But forcing it in the Hidden Options will accomplish the same thing, as long as you're sure it needs the IVTC. It uses the Decomb IVTC. For regular old stuff, it's fine. For difficult sources I'm with jagabo in believing TIVTC to be a bit better. Its built in deinterlacer is certainly much better. But that's if you're fairly experienced using AviSynth and are going to encode manually. Decomb is still the most commonly used AviSynth IVTC, at least partly because AutoGK uses it.

[wrawlind]

jagabo--
Your 3-line script worked like a charm. Quick and easy. I compared frames from the output of TFM to those of VDub's IVTC and TFM is clearly superior.

Thanks, manono. Thanks jagabo.

-R

[SLK001]

To tell you the truth, I'm not entirely sure if it analyzes AVIs the way it does DVD/MPEG sources.

Since it is being frameserved via AviSynth, they should both be handled identically - both as a psuedo AVI file. AFAIK, the end-using program cannot regress back into the source format.