Changes between Initial Version and Version 1 of Ticket #7163, comment 20


Ignore:
Timestamp:
Apr 10, 2020, 3:55:03 PM (6 years ago)
Author:
Elliott

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #7163, comment 20

    initial v1  
    11Arri Meta Extract is not definitive proof.  Neither is any other metadata reading tool, because there is no "bit depth" metadata flag to read in a ProRes QT.  FFmpeg displays yuv422p10le not because it is "detected", rather it's the pixel format we choose to decode.
    22
    3 The patch for this ticket decodes 12-bit for ProRes 444 and XQ only.  ProRes 422 sources are still decoded as 10-bit.  I think the current behavior is mostly okay because 10-bit is ''typically'' adequate for ProRes 422 sources.  I do recognize that it's possible to have 12-bit precision in a 422 file, and I have verified this myself with test patterns, but I very much doubt that you could tell the difference with real footage from a camera because of the signal-to-noise ratio.
     3The patch for this ticket decodes 12-bit for ProRes 444 and XQ only.  ProRes 422 sources are still decoded as 10-bit.  I think the current behavior is mostly okay because 10-bit is ''typically'' adequate for ProRes 422 sources.  It may be possible to have 12-bit precision in a 422 file, but I very much doubt that you could tell the difference with real footage from a camera because of the signal-to-noise ratio.
    44
    55From my perspective the current behavior is good enough.  But it would be nice to have a flag for the user to choose 10 or 12 bit for all ProRes sources.