Deep Stream Smart Video Record: unfixable negative timestamp

It seems chromium-based browsers or apps are unable sometimes unable to play the videos created by StreamSmartRecord. The video is mp4 encoded with h264.
I say “sometimes” because it isn’t consistent.
If I re-encode it manually again with HandBrake with the same specifications it works.

error: MediaError
  code: 4
  message: "DEMUXER_ERROR_COULD_NOT_PARSE: FFmpegDemuxer: unfixable negative timestamp."

• Hardware Platform: GPU
• DeepStream Version: 6.0
• TensorRT Version: 8,0
• NVIDIA GPU Driver Version: 470.141
• Issue Type: Bug
• How to reproduce the issue?: Try to play an mp4 video created by Deep Stream Smart Video Record on chromium-based browsers or apps.

1 Like

could you share one this video? this issue can’t reproduce locally.

This is one of the videos:

[FRAME]
media_type=video
stream_index=0
key_frame=0
pts=-100
pts_time=-0.040000
pkt_dts=-213

it is because pts starts with negative value, need chromium to fix this bug, there is some similar bug, 731766 - chromium - An open-source project to help move the web forward. - Monorail

Yes, I am aware and so are they I hope.
1186621 - Chrome does not play some MP4 videos - chromium

But is there any alternative besides waiting for them to fix this or encoding it again ourselves?
As I said if I encode it myself with handbrake and it’s fine.
image

So this should be “fixable” from here. Doesn’t Smart Video Record have any options? Maybe something similar to ffmpeg’s avoid_negative_ts make_zero?

  1. smart record will use gstreamer plugin qtmux to generate mp4, after checking, there is no this kind of avoid_negative_ts setting。
  2. how often this issue happen? could you provide simplified code reproduce this issue?

It happens ~60% of the time.
We are just using the deepstream-test5 sample code.

There is no update from you for a period, assuming this is not an issue anymore.
Hence we are closing this topic. If need further support, please open a new one.
Thanks

test5_dec_infer-resnet_tracker_sgie_tiled_display_int8-sr.txt (8.5 KB)
After testing, I can’t reproduce this issue, did you modify the test5 code? could you share you configuration file?