New libvpx version#82
Conversation
This invalidates a set of configurations because legal values for VP9 "cpu-used" have changed, and require an update to the fixed-q patch for VP8. This touches a number of verification tools. NOTE this commit: Verification tool work showed that the way "verify_scores" works is not good. Another CL intended in order to fix that problem.
This commit adds extensive reporting on what changed when a score changes. It also fixes a bug where the old score would be used for the new encoding.
The aq-mode and end-usage parameters have been recommended for RT usage, so adding them.
alvestrand
force-pushed
the
new-libvpx-version
branch
from
575a322 to
d80b00f
Compare
|
I think it's ready for review now. |
| @@ -139,8 +139,19 @@ static vpx_codec_err_t validate_config(vpx_codec_alg_priv_t *ctx, | ||
| RANGE_CHECK(cfg, g_timebase.den, 1, 1000000000); | ||
| RANGE_CHECK(cfg, g_timebase.num, 1, cfg->g_timebase.den); | ||
| RANGE_CHECK(cfg, g_timebase.num, 1, 1000000000); |
There was a problem hiding this comment.
All the changes in vp8_fixed_q.patch are made by:
- Applying the previous version of the patch
- Finding out were the patch failed to apply
- Adjusting the code until the non-applying part worked again
So this shouldn't be a change that I wrote, it should be a change from 1.3 to 1.6 in the file being patched. But I need to read the file really carefully to be sure that's the case!
| self.extension = 'webm' | ||
| self.option_set = encoder.OptionSet( | ||
| encoder.IntegerOption('cpu-used', 0, 16), | ||
| encoder.IntegerOption('cpu-used', 0, 8), |
There was a problem hiding this comment.
What does this cpu-used change mean? The description in http://www.webmproject.org/docs/encoder-parameters/ seems a bit cryptic to me. Which is fancy words for that I don't understand it. :-)
There was a problem hiding this comment.
They used to use the range 9..16 for experimental speedups. Now I think they use -8..-1 for the same purpose. I've cut off the upper numbers (which will invalidate the parameter sets that use those numbers).
| #git checkout v1.3.0 | ||
| # Check out the Oct 20 2014 version of libvpx. | ||
| git checkout 9c98fb2bab6125a0614576bf7635981163b1cc79 | ||
| # git checkout 9c98fb2bab6125a0614576bf7635981163b1cc79 |
There was a problem hiding this comment.
How about cleaning up these comments and commented out lines?
| @@ -71,15 +74,30 @@ def VerifyOneTarget(codec_names, rate, videofile, old_scores, score): | |||
| result, | |||
| old_score if old_score else float('-inf'), | |||
| new_score if new_score else float('-inf')) | |||
There was a problem hiding this comment.
Again reminder to consider making this function testable.
3 participants
This PR updates the libvpx binaries to version 1.6.0.
It doesn't address the problem of worsening performance, which is tracked in #86.