Posted: 2/4/2019 11:45 PM
Post subject: vp9?
Quote
feos
Site Admin, Skilled player (1236)
Joined: 4/17/2010
Posts: 11274
Location: RU
NES TAS of 2011
https://en.wikipedia.org/wiki/VP9 https://blogs.gnome.org/rbultje/2015/09/28/vp9-encodingdecoding-performance-vs-hevch-264/
  • Next-gen codecs provide 50% bitrate improvements over x264, but are 10-20x as slow at the top settings required to accomplish such results.
  • Normalized for CPU usage, libvpx already has some selling points when compared to x264; x265 is still too slow to be useful in most practical scenarios except in very high-end scenarios.
  • ffvp9 is an incredibly awesome decoder that outperforms all other decoders.
Dunno when we will have time for testing. All this talk about 30-50% savings on bitrate (at the cost of tremendous speed loss) isn't important for us, since we don't host or stream video, but if we can increase encoding speed while maintaining the same quality, it'd be very nice. Also haven't checked if it supports timecodes. But it does support 10-12 bit 4:4:4 and lossless.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Posted: 2/4/2019 11:54 PM
Quote
pirate_sephiroth
Active player (308)
Joined: 2/28/2006
Posts: 2275
Location: Milky Way -> Earth -> Brazil
New systems TAS of 2009
it's also not hardware-accelerated on older CPUs and GPUs.
"Genuine self-esteem, however, consists not of causeless feelings, but of certain knowledge about yourself. It rests on the conviction that you — by your choices, effort and actions — have made yourself into the kind of person able to deal with reality. It is the conviction — based on the evidence of your own volitional functioning — that you are fundamentally able to succeed in life and, therefore, are deserving of that success." - Onkar Ghate
Bisqwit wrote:
Drama, too long, didn't read, lol.
Posted: 2/5/2019 5:27 PM
(Edited: 2/6/2019 3:32 PM)
Post subject: Re: vp9?
Quote
creaothceann
He/Him
Editor
Joined: 4/7/2005
Posts: 1874
Location: Germany
feos wrote:
haven't checked if it supports timecodes
Wouldn't that be up to the container?
Posted: 2/6/2019 4:10 AM
Post subject: Re: vp9?
Quote
TheCoreyBurton
He/Him
Joined: 10/14/2013
Posts: 335
Location: Australia
creaothceann wrote:
Wouldn't that be up the container?
I believe so, and MKV should have no issues with VP9.
I'm not as active as I once was, but I can be reached here if I should be needed.
Posted: 2/6/2019 1:42 PM
Quote
HappyLee
Expert player (2454)
Joined: 12/23/2007
Posts: 822
NES TASer of 2012
Nobody seems to have tested x265 either, after I brought that up 3 years ago. HEVC is way popular now than VP9, and I've been using x265 for my The Simpsons encodes, and the result is much better than x264.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
Posted: 2/6/2019 3:05 PM
Quote
feos
Site Admin, Skilled player (1236)
Joined: 4/17/2010
Posts: 11274
Location: RU
NES TAS of 2011
HappyLee wrote:
Nobody seems to have tested x265 either, after I brought that up 3 years ago. HEVC is way popular now than VP9, and I've been using x265 for my The Simpsons encodes, and the result is much better than x264.
See my link, HEVC seems to be way heavier to decode. But since you're already using it, would you have time for some tests? The thing we're looking for is faster encoding speed while maintaining the same quality (whatever bitrate it'd be), or smaller file size while maintaining the same encoding speed. From what I've read, x265 isn't so advantageous for what we do (primary 10bit encode only; 512kb has to be paranoidally compatible, and YT doesn't care how we encode, we only aim for speed with it)
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Posted: 2/6/2019 7:28 PM
Quote
HappyLee
Expert player (2454)
Joined: 12/23/2007
Posts: 822
NES TASer of 2012
I think VP9 and HEVC both have the potential to replace the 10bit encode, but they are both overrated to me. Lots of times you see "saving about 20–50% bitrate compared to H.264", but most of the time it's done by killing details along with noise and making the video smoother. I find their largest advantage compared to AVC is that it's much harder to find blocks, and edges are usually smoother, making it more suitable for anime encoding, but in terms of TAS encoding, which we don't use deblock and smoothing, I'm not sure if they have real advantage at all. YouTube now uses VP9 (and probably the only one using VP9) for many of their videos. The VP9 encode is about 25% smaller than AVC encode, but the quality is slightly worse as many details are smoothed. I'll run some small scale tests of x265 tomorrow. I'd like to see some larger scale tests of both x265 and VP9 vs. x264 by encoders.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
Posted: 2/7/2019 3:01 PM
Quote
HappyLee
Expert player (2454)
Joined: 12/23/2007
Posts: 822
NES TASer of 2012
According to my small scale test (SMB TAS, crf 20), x265 takes about twice the encoding time, but produces about 50% larger file than x264. Also there's many settings we use for x264 encoding that x265 doesn't have. In short, x265 probably sucks for TAS encoding. I haven't tested VP9 because I'm not familiar with the encoder and settings at all.
Recent projects: SMB warpless TAS (2018), SMB warpless walkathon (2019), SMB something never done before (2019), Extra Mario Bros. (best ending) (2020).
Posted: 2/9/2019 7:06 AM
Quote
YaLTeR
He/Him
Joined: 12/2/2011
Posts: 129
Location: Moscow, Russia
Aren't the CRF values different between x264 and x265?
Pushing Half-Life to its limits.
Posted: 2/10/2019 3:39 PM
Quote
feos
Site Admin, Skilled player (1236)
Joined: 4/17/2010
Posts: 11274
Location: RU
NES TAS of 2011
Brief testing shows that reducing the quality down to what we use for primary runs twice as fast as x264, but produces a file that's 1.5 times bigger. Lack of detailed docs didn't help. I mostly used this command on SMW "game end glitch": 
ffmpeg -i smw.avs -b:v 0 -crf 40 -g 600 -row-mt 1 -c:v vp9 -an smw.webm
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.
Posted: 2/10/2019 6:28 PM
Quote
creaothceann
He/Him
Editor
Joined: 4/7/2005
Posts: 1874
Location: Germany
feos wrote:
Brief testing shows that reducing the quality down to what we use for primary runs twice as fast as x264, but produces a file that's 1.5 times bigger.
So theoretically it should be slightly faster to get an encode that is the same size and quality as x264...
Posted: 2/10/2019 6:55 PM
Quote
feos
Site Admin, Skilled player (1236)
Joined: 4/17/2010
Posts: 11274
Location: RU
NES TAS of 2011
Can't be proven without whoring the hell out of vp9 arguments. The main difference from intended vp9 use is that corporations have encoding farms and want to save on traffic. So I do believe they spend more cpu-power on crushing the filesize than we can afford. And they do save money through this, while we gain absolutely nothing either way. This is several times more true for AV1, which saves 30% more filesize than vp9, but takes years to encode.
Warning: When making decisions, I try to collect as much data as possible before actually deciding. I try to abstract away and see the principles behind real world events and people's opinions. I try to generalize them and turn into something clear and reusable. I hate depending on unpredictable and having to make lottery guesses. Any problem can be solved by systems thinking and acting.