Telecide/Decimate shortens length of material?

[sync]

I encoded a short (2:34) DV AVI sample with CCE that was frameserved with a script using the Telecide and Decimate filters. The result was one second shorter than the source and after muxing the audio sync was slightly off. I used pulldown.exe before muxing.

To rule out CCE being the cause I loaded the same script into VirtualDub and saved to AVI and again the result was one second shorter.

The material was captured from a TV show and is definately telecined.

Here's the script.

Loadplugin("D:\DVD2SVCD\Avisynth2 Plugins\InverseTelecine\decomb.dll")
Loadplugin("D:\avisynth2\plugins\dustv5.dll")
Loadplugin("D:\avisynth2\plugins\unfilter.dll")
AVISource("G:\SVCD\DoIt\Input\Source.avi")
ConvertToYUY2()
Telecide(post=false)
Decimate(cycle=5)
BicubicResize(480,480)
spacedust()
unfilter(20,20)

[Guest]

Serve the AVI into VirtualDub without Telecide/Decimate. What are the reported frame rate and frame count (use File Information)? Then serve it in again with Telecide/Decimate. What are the numbers this time?

[sync]

I'm using 1.5.2. I see "# of frames" and "Number of keyframes", which I assume is frame count. I don't see frame rate.

[trbarry]

And beware there is (IIRC) at least one version of DVD2AVI floating around out there that will say 30 fps material is really 30.00 fps instead of the more normal 29.97.

- Tom

[Guest]

Look again. It's right there on the first line. XXXfps.

What are the numbers, please?

[sync]

I guess fps is what you mean by frame rate.

without Telecide/Decimate:
frame rate: 29.970
frame count: 4621
audio frame: 4623

Telecide/Decimate:
frame rate: 23.976
frame count: 3696
audio frame: 3697

[Guest]

Originally posted by trbarry
And beware there is (IIRC) at least one version of DVD2AVI floating around out there that will say 30 fps material is really 30.00 fps instead of the more normal 29.97. Granted, but it doesn't matter. Decimate() does not alter the running time of the video (except for a very small amount that can result when the clip is not a multiple of the cycle size; but we're talking about hundredths of a second).

Also, he's starting with DV.

[Guest]

4621 / 29.97 = 154.1 secs

3696 / 23.976 = 154.1 secs

Thus something subsequent to Telecide/Decimate is the cause. Look to the codec.

(The numbers above are not exactly the same because your clip is not a multiple of the cycle size.)

Please don't throw around accusations without cause. You can mollify me by adding a question mark to your subject line.

[sync]

Originally posted by neuron2
Please don't throw around accusations without cause. You can mollify me by adding a question mark to your subject line.
It was meant to be a question, not an accusation. Sorry.

[Guest]

Originally posted by sync
I guess fps is what you mean by frame rate.Correct!

fps = frames per second = frame rate

[Guest]

Oh, sorry for getting bristly! My bad. :o

Let's try to solve this.

What codec were you encoding with? Settings? I'll try to duplicate it if you can give me that information.

[sync]

It's the Panasonic DV codec. Are you referring to codec settings? I didn't know there were any.:confused:

[sync]

I just tried it with Huffyuv and the time came out right.

[Guest]

That's very interesting, sync! Thanks for pointing it out.

I'm curious to know if you take the encoded AVI (the one that is too short by a second) and load it in VirtualDub, what are the numbers. Did it perhaps drop some frames, or did it set the frame rate wrong?

[sync]

I goofed when I said Panasonic codec. That's what I used to capture with. If I use either the Pansonic or Canopus codecs in VirtualDub I get the message "cannot start video compression: the source image format is not acceptable. (error code -2)" when I try to save.

So I saved uncompressed. I deleted the file and am running it again to get the numbers.

[Guest]

>the source image format is not acceptable

That's because the frame size has to be 720x480 for DV (and something else for PAL DV). Take out your BicubicResize().

Anyway I made a test with Panasonic DV just now and it came out exactly correct.

[sync]

Should I being using the defaults "Video - full stream processing", "Audio - direct stream copy"?

[Guest]

Originally posted by sync
Should I being using the defaults "Video - full stream processing", "Audio - direct stream copy"? Yes, that is fine. Or you can set Fast Recompress for video. Either way it should not matter.

[sync]

Is it possible to see how many frames are in an MPEG2?

[Guest]

Originally posted by sync
Is it possible to see how many frames are in an MPEG2? Sure, but you need some special tools.

Before we get into that, can you try it without pulldown.exe?

[Guest]

Just thought of something. When you make the MPEG2, are you setting it for 29.97 or 30.0 fps? If you set it for 30.0 fps, that will drop it to 153.8 seconds.

[sync]

I tried to create another avi using the script but I did something wrong. I need to get some sleep and clear my head.

Thanks a lot for your help!

[sync]

Originally posted by neuron2
Sure, but you need some special tools.

Before we get into that, can you try it without pulldown.exe?
What do you want me to try - muxing?

[Guest]

OK. Check out my last post. It's probably something to do with how you are making the MPEG2.

[Guest]

Originally posted by sync
What do you want me to try - muxing? Never mind. It was a silly idea.

[sync]

Originally posted by neuron2
Just thought of something. When you make the MPEG2, are you setting it for 29.97 or 30.0 fps? If you set it for 30.0 fps, that will drop it to 153.8 seconds.
I'm using 'MPEG tools' in TMPGEnc and all I did was specify the two inputs and the output file.

[sync]

The file I'm muxing is one second short. But perhaps this one second difference is due to rounding and the audio sync problem is caused by something else.

[Guest]

I'm ignorant of all this muxing/CCE stuff. I suggest that you take the problem to the CCE or TMPG forum now, because we have established that there is no Avisynth or filter problem.