Comskip Support Forum

Weird bug I've confirmed with 2 files now. Around frame 45 comskip seems to add 3 frames. In the editor, when going frame-by-frame through the file at around frame 40 it does a weird little skip. Then, if you move backward it seems it has inserted 3 frames at around frame 45 (they don't show on the first scroll forward).

The result is that all edits are off by 3 frames (which is how I discovered this - I was getting a "flash" of a commercial that had been edited out, on stations where there wasn't much of a black interlude)

I've looked at the same videos in Avidemux (which I use as my editor), and in VirtualDub, and the weird three frames are not there.

I am recording using ATI TV Wonder 650, GB-PVR, 9000 Kbits/sec variable bitrate, Windows XP, and I process all my files (demux) using ProjectX, remux with MPlex.

When you set
verbose=10
in comskip.ini
Do you see any
video PTS jumped
messages?

No. Just the following, from the mpg file I described:

Audio PTS jumped 1 frames at frame 6, repairing timeline
Audio PTS jumped 1 frames at frame 22, repairing timeline

I tried a quick record without ProjectX/MPlex processing, 3 of the above-type errors, same 3 frame skip

Could you set
output_timing=1
and rerun comskip on the original file and mail me the zipped large generated _timing.csv file?

Do you care whether the file has been processed by ProjectX and MPlex? If so, I will have to make a new recording, and process that for you.

I don't care about the history of the file as long as you process a file that is causing the problem.

This bug appears to be in all my recordings (10+ checked so far). It is always visible between frames 40 and 50 somewhere (although I have seen weirdness before frame 40, like all frames seemed greyed out).The end result is always 3 frames extra added by Comskip, so my cut points are always exactly 3 frames too far.

I check the file you send and there is nothing wrong with the recording you are processing.
it must be in the seek algorithm in Comskip that switches around frame 45 to a different seek method. The only problem is that there is no evidence of this in the detailed log file you send.
I can send you the upload details for my ftp server so you can upload a small (100MByte) recording that shows the problem.