Open acdha opened 10 years ago
Looks like this was fixed in 0.6 – there's still no overt indication of decode failures on the list but the the video parsing completes now and the decoding failure is incredibly obvious in the graphs.
Thanks for testing. At some point perhaps we should make the decoding log accessible, so I'm going to change to ticket to that.
Great, thanks – even a simple status badge would be handy for our content team triaging incoming deliveries from partners
http://chris.improbable.org/experiments/browser/video/waters_of_destiny.mp4 is malformed. It plays back despite this using the native Windows or OS X APIs but most other software chokes (see http://chris.improbable.org/experiments/browser/video/waters_of_destiny.html).
I was hoping that QCTools could be used to reliably catch problems like this as early as possible in the process. When processing that video, QCTools 0.5 hangs on frame 9797:
This may be an argument for a hard timeout as it's been chewing 120% CPU for around 20 minutes on the same frame. I've saved a sample here: https://gist.github.com/0703fc6af3623de1d265
The good news is that despite this, the glitch shows up rather prominently in the graphs so it's easy to discover – the decoding first fails at frame 1337 and since it isn't quietly ignored like e.g QuickTime the lack of thumbnails and the huge diff make it easy to localize: