Seeking issues with GOP edited files

I am a new user… (due to Synology dropping their Video Server/Player), Infuse looks like a good option early trials go OK, with a couple of issues so far.

Bulk of the videos come from an Enigma2 based PVR where the ads have been cut out. The Engima2 cutter does a simple chop between GOPs leaving the PTS values alone, which appears to the be the root of the issues seen.

The first issue that has been encountered is that, on an ATV4, when a point at which a cut has been made, the playback is getting stuck in a loop of stepping back a fraction, then playing forward over the same few video frames. Bring up the jump bar, skip forward a fraction and it plays forward to the next cut. This has been seen on one video so far, so yet to determine if it is a general problem.

Second issue seen on the Mac is using the 10 second step forwards works, however the 10 second step back is also choking at a cut point by going forward to the point at which the step back was clicked. At a guess, the underlying code for both forward and backward is recalculating a new PTS and seeking forward for the next GOP. Works fine when going forward, however for backward, the calculated PTS lands in the cut out section it seeks forward and find the next playable point is close to where it was sent back (perhaps a seek backward to the previous GOP?)

It is likely that the player will handle file with gaps in the PTS sequences? Previoius players that have been tried has also choked on PTS sequence resets on unedited files. It looks like the broadcast hardware can arbitrarily restart the PTS sequence to zero in the middle of broadcast.

Any helpful comments/suggested workaround appreciated.

You might send a file to James to check out:
firecore.com/upload

Short segment that demonstrates the issue uploaded as suggested and a email sent to accompany it.

Thanks for the sample.

I’ve not had a chance to dig into the actual video, but attempting to seek this video results in position/skipping issues in all popular players (Infuse, IINA, VLC, etc…). This likely indicates the file was edited in a non-standard way.

Thanks for the comment. As I understand it, the Enigma2 “editor” simply removes complete GOPs without adjusting any other values (that is, PTS values remain at the values in the original recording). This is known to cause issues synchronizing Closed Caption presentation when the players track their own elaspsed time.

One issue that you might like to put into the back of the mind in case it crops up with others is that of the PCR being reset in the middle of a recording. It is known to create crashes with software that is not on the lookout for it. Why ? From investigation of the Transport Stream specification, the PTS is stored as a UInt64 value, thus systems do not respond well when calculating delta movements at the PTS level when the value unexpectedly goes back to zero. Think of seeking backward 10 seconds after a reset and you might calculate the PTS as being UInt64(10000 - 90_000 * 10), which, needless to say, does not end happily.

Still happy with Infuse and no real show stoppers.