Added optional start (dark green) and/or end (dark red) frames to signal the start/end of the test stream using the --start-end-indicators argument. Would it be useful to use a test pattern for the first frame instead or is a single color best for test automation reasons?
Added support for BT.2020 HDR output. If the source content is BT.2020, the script encodes the output using HEVC. (Note: ffmpeg ProRes output colorspace, primaries and transfer function signalling does not seem to be working as expected.)
Change:
Location of label and AV sync flash adjusted to provide enough space for a JS app to also display a timecode (top-centre) and QR code (right side of the screen).
Changes to address issues from Zack's review:
Added requirements.txt and updated the README with some more details (once we agree on the final version we can clean up/merge the older parts).
All scripts now require Python 3 instead of a mix of Python 2 and 3. Used 2to3 to translate the test sequence generator scripts to Python 3 and resolved a new related issue.
Now uses pathlib to avoid using OS-specific directory paths.
Remaining issues to look into:
Avoid generating QR codes, flashes and beeps into files and instead directly synthesize them in memory.
Adapt annotations when source content is HDR (to avoid overly bright graphics).
From Eurofins comments in issue #1 : "To make automated capture easy it may be necessary to silence spoken audio either side of the beeps". - This isn't done at present. If this is still a concern, perhaps the weighting of the original source audio vs the beeps can be adjusted to avoid any detection issues.
From Eurofins comments in issue #1 : "We would suggest having an audio tone at the start (for audio start up delay) and end, and at a consistent jump to point (e.g. 20s) to enable automated checking." - Can add a tone to accompany the start/end frames. Considering the irregular AV sync pattern repeats after a configurable duration, do we need to additionally define a specific jump point?
Comment in issue #1 : "With the audio need to make sure that every audio sample has a defined start and end tone (apply a different frequency than the beep - could also be a short jingle (identifying) and it is longer - like 1 seconds)." - The configurable irregular AV sync pattern should already address this, but to be confirmed.
Added the following functionality:
--start-end-indicators
argument. Would it be useful to use a test pattern for the first frame instead or is a single color best for test automation reasons?Change:
Changes to address issues from Zack's review:
Remaining issues to look into: