GoogleCodeExporter
commented
9 years ago Original comment by iss...@webmproject.org on 3 Nov 2011 at 9:20
- Added labels: ****
- Removed labels: ****
Closed GoogleCodeExporter closed 9 years ago
GoogleCodeExporter
commented
9 years ago Original comment by iss...@webmproject.org on 3 Nov 2011 at 9:20
GoogleCodeExporter
commented
9 years ago I am not quite sure what you are asking for here.
If it is just a new fixed Q mode then this is possible but not a priority.
It is possible to set the minimum and maximum Q to the same value, to force Q
to be constant.Original comment by paulwilk...@google.com on 4 Nov 2011 at 12:28
GoogleCodeExporter
commented
9 years ago There is no it doesn't approach. If will expose q макс = q mines gold shots
are destroyed, quality/stream becomes bad.
Advantage constant q in that that the codec doesn't spend resources for a scene
estimation - it works faster, besides in comparison with 2 passes a stream,
there can be a prize on quality, at least in VP7 advantage was.Original comment by mylo.m...@qip.ru on 5 Nov 2011 at 9:06
GoogleCodeExporter
commented
9 years ago I can add still that the mode --end-usage=2 is badly adjusted - so blocks
bright or with sharp details at the same quality dark blocks with soft details
very badly look remarkably, and. Make the amendment on brightness of the block
at least.
As in shades very rough gradients - all in squares.Original comment by mylo.m...@qip.ru on 5 Nov 2011 at 9:50
GoogleCodeExporter
commented
9 years ago [deleted comment][deleted comment]Has made the list of the most actual offers besides above stated:
1. To make fast first pass by picture reduction by an input on 50 % we will
tell 720p in 360p VP7 and early VP8 allowed such trick - an additive of speed
decent, quality especially doesn't influence.
2. To make before pass which will define scene borders, and different scenes to
distribute on different kernels of the processor (multiline first pass)
3. To harden the first pass that it wouldn't react to noise and small movement
is will allow to expose gold frames only during a great movement that very much
will increase compressibility (by 10-20 %)
4. The slow full frame provokes movement VP8 to exhibiting of gold frames
through 4 intermediate, as well as in VP7, however movement indemnification in
VP8 has considerably improved also the minimum exhibiting of gold frames it is
necessary to increase to a minimum 6-8 or about 1/2 seconds (the maximum time
psychological not perceptions of artifacts)
5. To correct a failure in quality on dark - - (red, brown, gray, dark blue).
The codec exposes the gold frame with normal detailed elaboration, falling in
quality if these colors in the frame then follows much
6. To make mode AQ if is to finish it to a normal state.
7. To make 10-bit a codec variant.
8. To add coding in color spaces I422 and I444.
9. To realize bilaterial gold frames: intermediate frames refer on being before
them and after them the gold frame (it is admissible н intermediate frames
standing to the gold frame refer to it and n intermediate frames standing after
the gold frame refer the same to it instead of as now the gold frame n
intermediate frames, the gold frame n intermediate frames - reduction in two
gold frames will lead to reduction bit rate at the same quality on 20-30 %).
10. To make at last already support on an input mkv, avi, mp4 and avs in
vpxenc, as in x264.
I not the programmer also can't make fork VP8 or help with a code with the
project, however all offers are based on long-term use of codecs of family VP.
If you plan to achieve considerable improvement by elimination of errors in my
opinion it already waste of time, the best compressibility and speed can
achieve only good tuning of the codec.Original comment by mylo.m...@qip.ru on 20 Nov 2011 at 3:18
Original issue reported on code.google.com by
mylo.m...@qip.ruon 3 Nov 2011 at 9:12