Results

This page shows the video encoding results available on this particular site.

No-limit results

This is the result of comparing encodings based on PSNR and file size being below a certain bitrate, with no restrictions on tools applied, and no penalty applied for moving bits around within the bitstream.

The first column gives the baseline codec; the top row gives the codec it is compared to, and the number gives the change in bitrate between the two for the same quality metric.

Positive numbers indicate an increase in filesize (left codec is better); negative numbers indicate a decrease in filesize (top codec is better).

PLACEHOLDER VP8 x264 H263
VP8   -22% -71%
x264 +30%   -64%
H263 +262% +203%  

Tuned results listing

This is the same results as above, but this time, we add the ability to tune the parameters at each data point.

This will show how far it’s possible to squeeze the performance of each codec - “you can’t do better than this”.

In the graphs linked below, each codec is represented by a dotted line showing “this is the best we can do at this target bitrate”, and by a solid line showing the result of running with the single “best” configuration for all bitrates.

PLACEHOLDER VP8 x264 H263
VP8   -22% -71%
x264 +30%   -64%
H263 +262% +203%  

### Fixed QP results

This table shows the overall numeric results based on PSNR, computed according to the BD-PSNR method. Click on a percentage to go to the page that shows the detailed results, and results based on other metrics.

PLACEHOLDER VP8 2-Pass x264 2-Pass HEVC-HM
VP8 2-pass   +10% -12%
x264 2-Pass -8%   +17%
HEVC-HM -6% +7%  

Note that the table is not symmetric; if A beats B by 12% when A is the baseline, A will beat B by 8% if B is the baseline. This follows from the way in which the percentages are computed. (FIXME: Check numbers)

Fixed bitrate results

This table shows the overall numeric results based on PSNR and bitrate overrun. The score for any clip is reduced by 0.1 for each second of accumulated delay, divided by the length of the clip.

PLACEHOLDER VP8 2-Pass x264 2-Pass HEVC-HM
VP8 2-pass   +10% -12%
x264 2-Pass -8%   +17%
HEVC-HM -6% +7%  

Interactive Results

This table shows the overall numeric results based on PSNR and bitrate overrun. The score for any clip is reduced by 1 for each second of accumulated delay, divided by the length of the clip.

In addition, to qualify for this ranking, an encode must be performed using less CPU time (single CPU) than the length of the clip, and it must have no forward looking elements - that is, a decoded frame N must be exactly the same no matter what the content of the original frame N+1 and later were.

PLACEHOLDER VP8-RT X264-RT HEVC-HM
VP8   +10% DNQ
X264 -8%   DNQ
HEVC DNQ DNQ