Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Matej_K_
Beginner
28 Views

MPEG decoder outputs non-monotonous timestamps

Stream is progressive NTSC telecine (29.97fps, 24 frames per second) 

Timestamps send to decoder (bitstream->Timestamp)
0:00:00.000000000
0:00:00.083000000
0:00:00.050049999
0:00:00.217000000
0:00:00.133049999

0:00:00.166416665
0:00:00.334000000
0:00:00.250366666
0:00:00.300416665
0:00:00.467000000
0:00:00.384049999
0:00:00.417416665

Timestamps from decoder:

0:00:00.000000000
0:00:00.083000000
0:00:00.116366666
0:00:00.217000000
0:00:00.133044444
0:00:00.050044444
0:00:00.334000000
0:00:00.250355555
0:00:00.166411111
0:00:00.467000000
0:00:00.384044444
0:00:00.300411111

The input timestamp are correct (when ordered by temporal sequence number they are monotonous). Samples with 50ms length are RFF. I have no idea what the decoder tries to accomplish here. Could anyone shed some light into this?

(this has been tested with software decoder; I can't currently test hardware decoder)

0 Kudos
2 Replies
28 Views

Hello there - Can you give more information about your system and the set-up? Please use the following post for the information that is useful for us. https://software.intel.com/en-us/forums/topic/531083

Matej_K_
Beginner
28 Views

Attached is output from system analyzer. This is in Parallels on MacBook Air 2012. Since it's using software decoder I don't think it matters. Both 32 and 64bit versions behave the same. I can provide the file in question on request (PM). I'm currently unable to test this with hardware decoder. I'm not sure if it makes any difference for the decoder, but the file is muxed in matroska so DTSs are absent.

Also when I try to decode the file using directshow filter from the sample the filter just crashes (I didn't have time to try to debug this yet). I think you should be able to reproduce this with the sample as well.

Reply