Skipping commercials: Real frame numbers?

[erik]

Does GBPVR use the real frame numbers of the played back recording to determine where to skip using the frame numbers in the .txt file from SA or Comskip or are these frame numbers from the .txt file first converted to time codes using the fps and are these timecodes then being used during playback to know where to skip?

[sub]

or are these frame numbers from the .txt file first converted to time codes using the fps and are these timecodes then being used during playback to know where to skip?

Correct. It doesnt use frame numbers internally. Frame numbers are real pain in the arse to deal with DirectShow. I have to determine the FPS for the file, the do the math to work out position in time within the video.

[erik]

OK, so I assume we both have a problem with files that switch between different fps like -stattik- describes, sigh, I hoped I could shift the problem to you
If you can teach me how you determine the real fps of such a file I hope to copy it in Comskip so the Comskip output for those files will work in GBPVR

[Jere_Jones]

What do ya'll think about updating the comskip format to use timestamps? Maybe something simple like floating point decimals?

There would need to be some sort of header like:

Version 2
------------------------
0.0 30.32342321
145.30231202 323.32032
etc....

I'm sure it would be easier for the users to read and understand as well.

Jere

[erik]

Would it make life simpler for sub? I don't mind.

On the other side it will create confusion with the users because the old frame number format must be available too because someone may update SA/Comskip and stay with their old reliable GBVPR (like me, over 5 months I have not found one single problem that forces me to upgrade)

So adding a additional format seems the only solution, but then there are already time based formats available.

Jere,

Did you ever experience the problem of changing fps in one file?

[-stattik-]

OK, so I assume we both have a problem with files that switch between different fps like -stattik- describes, sigh, I hoped I could shift the problem to you
If you can teach me how you determine the real fps of such a file I hope to copy it in Comskip so the Comskip output for those files will work in GBPVR

I can't find any evidence that ATSC files are supposed to have frame rate switches like we've seen. It doesn't make a lot of sense. I'm doing some more testing so I'll have more info soon.

[erik]

I checked in another player and there is no fps change but only a switch between interlaced and progressive (which I also do not understand)

If I force comskip to assume a fps of 25 the debugwindow functions perfectly.
Other tools also fail to position correctly (like mpeg2cut2)

[Jere_Jones]

Did you ever experience the problem of changing fps in one file?

I have seen errors in the fps field that gave the impression of a change, but the next header had a correct entry.

If the recording was created by a Hauppauge card, then it is possible for the error to run for several headers. Of course other cards may experience the same thing, but I have only seen it happen on the Hauppauge cards.

By and large the fps shouldn't change. In theory interlaced/progressive can change from frame to frame depending on the encoder's judgment of which is more efficient. I haven't seen it happen yet though.

Jere

[sub]

It would certainly make life easier to have a file format with times instead of frames. I'd prefer a different filename than just a header change. How about videofilename.ad?

[Jere_Jones]

It would certainly make life easier to have a file format with times instead of frames. I'd prefer a different filename than just a header change. How about videofilename.ad?

How about *.ads? That will keep it a traditional 3 character extention.

Also since Sub, Erik and I (who would seem to be all the major players) are in on this conversation, are there any other changes to the format that would be useful?

Jere