Weird IVTC problem

[StephLG]

Hi,
I encounter a weird problem when trying to do IVTC with AVISynth and VirtualDub.
My source file is a 22Gb HuffYUV avi, captured from my "Star Wars" NTSC Laserdiscs using an Asus GeForce3 8200Ti video card and iu_vcr software. According to iu_vcr, there were no dropped frames during the 50min capture.

I wanted to try a new IVTC method (new to me), doing a manual IVTC in TMPGEnc, saving the project as a '.tpr' and then using the tprivtc.dll plugin for AVIsynth to perform ivtc according to information stored in the '.tpr' file.
The manual IVTC step in TMPGEnc (v2.52 Plus or something like that) worked very well; the '10100' pattern found could be applied to the whole movie. I checked the entire movie for a possible phase change but didn't find any. So, I saved the '.tpr' project and imported it in avisynth with the following script:

LoadPlugin("tprivtc.dll")
OpenDMLSource("capture.avi")
DoubleWeave()
Tprivtc("capture.tpr")

The result in VirtualDub (v1.4.10) is really strange; the resulting framerate is said to be 23.976, so IVTC should have worked well but half of the resulting frames are interlaced! Actualy, when I make a slow preview in VirtualDub, frame by frame, I can see there are 2 good frames, then 2 bad frames, then 2 good frames, the 2 bad, and so on...
I thought something was wrong with my '.tpr' file, but I also tried a more conventional IVTC method in avisynth, using script like:

OpenDMLSource("capture.avi")
DoubleWeave()
and each of the following pulldown patterns
Pulldown(0,2)
Pulldown(1,3)
Pulldown(2,4)
Pulldown(0,3)
Pulldown(1,4)

The result with the above script is the same. So, I guess the problem doesn't come from the tprivtc method but either from avisynth (v1.0b7a) or the avi file. I remember having successfuly performed IVTC with this same version of avisynth and the pulldown() method about a month ago so it must come from the source file. The weird part is that, when viewed in TMPGEnc, the source file looks perfect; I did a close watch, frame by frame, when trying to find the pattern '10100'.

What could it be? Anyone already had such a problem?

Thanks, Steph.

[morsafr]

Just try another method. I suggest to use Decomb 3.8 which gives good results.

Just load the plugin and add the following lines to your avs script :

Telecide(post=false)
Decimate(cycle=5)

Just tell me if there are still combed (interlaced) frames.

[bill_baroud]

vfapi don't work ?
(or perhaps you can't use vfapi...)

convert .tpr with the VFapi converter, if the Vfp support is installed in Tmpeg, this allow you to use Tmpeg project in Virtual Dub


else, i have see somewhere (but don't remember where ;_; )a command which allow you to import vfp directly in avs (but not the one you have used)

if i re-find it, i will post it here.

[StephLG]

Hi guys,

@morsafr
I've just tried the Decomb plugin; it works well (it's currently working as I write this). The only thing I wonder is if it will behave correctly if the phase changes during the movie part (I mean if the pattern changes) due to a dropped frame or just to the way the telecine was made. To be honest, I haven't read (not yet) the html help so there may be some parameters to use to avoid that kind of problems.
But, although Decomb is doing the job, that doesn't explain why the other two methods fail; as I wrote before, I had already successfuly used the Avisynth's pulldown function, but it was on a movie captured with a Pinnacle DV500 card under Premiere (using DV Codec since DV500 doesn't allow anything else, even for an S-Video analog capture).
Anyway, thanks for the Decomb idea.

@bill_baroud
I'm not familiar with vfapi; the only thing I wanted to get from the .tpr was the IVTC information to use with Tprivtc plugin. This way, I could be sure to have a perfect IVTC (if the pattern changes). If only it works :confused:

The next challenge will be to feed CCE 2.64 with the direct result of an Avisynth script; I know CCE 2.64 doesn't recognize avs files so maybe that's were vfapi comes (although I don't know how it works).
The VirtualDub step was just to see the result of the Avisynth script; once I know it works, I'd like to bypass VirtualDub because the whole raw capture will take about 60Gb (on a 80Gb HD) and I won't have enough space to make an intermediate HuffYUV file out of VirtualDub to give it to CCE.

Any tip will be welcome ;)

[bill_baroud]

Vfapi appears in windows like a read-only codec.

The vfapi converter create a pseudo .avi which could be opened in other application...


but i have never try with CCE, if i need mpeg encode (well, not very often ^^; -i should say never-) i stick with Tmpeg.

but if it's a codec, i don't see why it couldn't works with CCE.

i'm in exam right now, i haven't the time to check doc to find the avisynth command, sorry :/ .