EL-Benjo

Q: Question regarding VBR (and hardware encoding)

Does Compressor 4.2.2 use variable bit-rate when I am exporting an MPEG-4 file with multi-pass on and the data-rate set to custom 10,000kbps?

There is no VBR or automatic setting in the MPEG-4 settings, so I was thinking maybe the custom data-rate I input is used as an average and selecting multi-pass makes it use VBR. I do want VBR to be enabled. Someone enlighten me

 

PS, does multi-pass use hardware acceleration? And does allowing segmenting disable hardware acceleration?

 

My specs:

iMac 5K Late 2015

Intel Core i7 4GHz Quad Core Skylake

32GBs RAM

AMD 4GB GPU R9 M395X

iMac with Retina 5K display, OS X El Capitan (10.11.5), 4GHz Quad, 32GB RAM, 4GB GPU.

Posted on Jul 5, 2016 5:13 AM

Close

Q: Question regarding VBR (and hardware encoding)

  • All replies
  • Helpful answers

  • by Russ H,Solvedanswer

    Russ H Russ H Jul 5, 2016 6:41 AM in response to EL-Benjo
    Level 7 (21,760 points)
    Quicktime
    Jul 5, 2016 6:41 AM in response to EL-Benjo

    The specified bit rate is an average and it's VBR.

     

    Hardware acceleration is for single pass only.

     

    Russ

  • by EL-Benjo,

    EL-Benjo EL-Benjo Jul 5, 2016 6:47 AM in response to Russ H
    Level 1 (12 points)
    Video
    Jul 5, 2016 6:47 AM in response to Russ H

    G'day, thats what I'v been hearing! I agree with you on the first point about VBR, but I'm a little confused about hardware acceleration because I ran a few tests. I encoded a video with multi-pass on and one instance, so theoretically hardware encoding should NOT kick in. I got 14 min encoding time. I then went up to 2 instances, and multi-pass was still on, but the encoding slowed down and too 18 mins. Repeating the test replicated these results. Upon doing some research, I discovered that in Apple's compressor path notes for one of the updates they specified that hardware acceleration now works with multi-pass. Some were speculating that it might be using the GPU. I haven't done any further testing to verify this, but according to my results it appears multi-pass with one instance is faster than multi-pass with 2 instances. However at 6 instances it begins to be faster than single instance. This suggests that hardware acceleration does support multi-pass, but I've read in multiple other places that it doesn't as well.

  • by Russ H,Helpful

    Russ H Russ H Jul 5, 2016 8:28 AM in response to EL-Benjo
    Level 7 (21,760 points)
    Quicktime
    Jul 5, 2016 8:28 AM in response to EL-Benjo

    Thanks for sharing the testing.

     

    I looked up the release notes for 4.2 and now see the entry that the update provided support for multi-pass hardware acceleration. Although I can't test myself until later this afternoon, it appears I was misinformed  – following others' obsolete advice. (The 4.2 user guide is completely silent on this topic.) Thanks for enlightening me!

     

    Russ

  • by EL-Benjo,

    EL-Benjo EL-Benjo Jul 5, 2016 8:34 AM in response to Russ H
    Level 1 (12 points)
    Video
    Jul 5, 2016 8:34 AM in response to Russ H

    If you get around to testing, please do share your results.

    Thanks