Creative Communities of the World Forums

The peer to peer support community for media production professionals.

Activity Forums Compression Techniques Dynamic lossless image Seq intermediate

  • Dynamic lossless image Seq intermediate

    Posted by Manuel Weber on October 5, 2014 at 12:30 pm

    Dear Cow

    I’m optimizing an ffmpeg cluster for DCP creation and I have a question on a format.

    I’m doing an output to image sequence for later DCI j2c encoding.
    The most robust way is to do RGB 16bit tiff output. But sometimes this is a bit of an overkill.

    >> Is there a fast image format that takes into account, what quality the original movie image has?

    Meaning that the image has to be lossless, BUT better just save what the original contains, instead of just blowing it up to 16bit in any case…

    I know jpeg2000 has a lossless specification, and does a good analysis on the original, but it’s terribly slow while encoding. PNG seems to be fast and dynamic, but it’s limited to 8bit…

    Thanks to any suggestions!

    Best
    M

    Manuel Weber replied 11 years, 7 months ago 2 Members · 3 Replies
  • 3 Replies
  • Manuel Weber

    October 8, 2014 at 6:04 pm

    I guess it’s fair to give the answer myself, as far as it’s possible

    There is a JPEG variant: LJPEG and JPEG-LS
    Both are lossless, the LS variant is considered to be faster.

    I had 14 fps to TIFF, now 110 fps to JPEG-LS
    (decode is minor here)

    The downside is that both seem not very widely supported in apps.

    Imagemagick doesn’t know it in ffmpeg’s “jls” container.
    ffmpeg wouldn’t know IM’s “DCM” container.

    Cheers
    Manu

  • Reuben Martin

    October 13, 2014 at 4:37 am

    PNG isn’t limited to 8bit.

  • Manuel Weber

    October 13, 2014 at 11:35 am

    Thanks for pulling me back to the original assumption.
    I read and tested it wrong, i guess.

    True, PNG can do 8bit and 16bit in colour!

    M

We use anonymous cookies to give you the best experience we can.
Our Privacy policy | GDPR Policy