Joined: 3/9/2004
Posts: 4588
Location: In his lab studying psychology to find new ways to torture TASers and forumers
This may not be indicative of anything, but I just spent some time grabbing some stats from YouTube.
I looked at some medium sized and large (labeled 720p) auto encodes there for H.264 in MP4 versus VP8 in WebM.
The results surprised me a bit.
For MSX, NES, SNES, CGB, and SGB, the VP8 ones were generally a bit smaller. In some cases, I found VP8 to be half or a third of the size of H.264 one.
For DMG, AGB, PSX, and N64, H.264 generally had the superior file size. In the case of DMG it was on average 30% larger, in other cases, 10-15%.
I didn't check quality in any of these cases, and this is not at all indicative of the results we'd get with manual encodes where we can tweak the settings. But if this is not just a case of YouTube's H.264 encoder settings for some games being lousy, then there may be something useful here that we can learn. Any encoders want to look more into this?
Warning: Opinions expressed by Nach or others in this post do not necessarily reflect the views, opinions, or position of Nach himself on the matter(s) being discussed therein.
Joined: 11/4/2007
Posts: 1772
Location: Australia, Victoria
I'll be interested when the VP8 encoders and decoders begin supporting YUV444/YV24 properly.
Otherwise, I don't foresee it being particularly useful, factoring the compatibility risks.
Joined: 3/9/2004
Posts: 4588
Location: In his lab studying psychology to find new ways to torture TASers and forumers
Flygon wrote:
Otherwise, I don't foresee it being particularly useful, factoring the compatibility risks.
If the quality turns out to be the same, if for some systems (games), the WebM file we produce can be half the size, you don't think that's useful?
Warning: Opinions expressed by Nach or others in this post do not necessarily reflect the views, opinions, or position of Nach himself on the matter(s) being discussed therein.
Joined: 11/4/2007
Posts: 1772
Location: Australia, Victoria
The point of encodes are generally to provide a high quality media playback file that's also widely compatible. Unless things have changed internally since I retired, this is the reason the H264/AAC for MP4 and H264/Vorbis for MKV combinations are so widely used.
I am reasonably sure Grunt will be able to provide a more up to date and accurate explanation.
Joined: 3/9/2004
Posts: 4588
Location: In his lab studying psychology to find new ways to torture TASers and forumers
Flygon wrote:
The point of encodes are generally to provide a high quality media playback file that's also widely compatible. Unless things have changed internally since I retired, this is the reason the H264/AAC for MP4 and H264/Vorbis for MKV combinations are so widely used.
The point of encodes are to provide high quality videos for our viewers to enjoy. We use H264/AAC for MP4 and H264/Vorbis for MKV because that's been giving us the best file size combination in recent years.
All the up to date video players can already play WebM, and last I checked, we like to be at the head of technology. We used Bit Torrent and H.264 before most people ever heard of these technologies or were in mainstream use.
Warning: Opinions expressed by Nach or others in this post do not necessarily reflect the views, opinions, or position of Nach himself on the matter(s) being discussed therein.
Joined: 11/4/2007
Posts: 1772
Location: Australia, Victoria
Again, poke the Dragon, not the Haystack. I'm just regurgitating what I was taught all that time back.
I'd imagine that if the staff team was going to use VP8, they'd trial by publishing two encodes at once or something. I don't claim to know how the site works anymore, I just claim to say that VP8 is a risker technology to use than the current tried and tested technology, particularly since the results seem quite variable. It is worth noting that the VP8 videos on YouTube for a given video mode do tend to have a worse video quality.
Joined: 3/9/2004
Posts: 4588
Location: In his lab studying psychology to find new ways to torture TASers and forumers
Flygon wrote:
I'd imagine that if the staff team was going to use VP8, they'd trial by publishing two encodes at once or something. I don't claim to know how the site works anymore, I just claim to say that VP8 is a risker technology to use than the current tried and tested technology, particularly since the results seem quite variable. It is worth noting that the VP8 videos on YouTube for a given video mode do tend to have a worse video quality.
We're not going to use VP8 on the site till extensive research is done.
This is just a call to do some testing to see where the technology stands.
Warning: Opinions expressed by Nach or others in this post do not necessarily reflect the views, opinions, or position of Nach himself on the matter(s) being discussed therein.
Joined: 11/4/2007
Posts: 1772
Location: Australia, Victoria
I understand.
I have actually attempted to test VP8 before due to the potential use of the YV24 colourspace, obviously this didn't pan out or I'd have been making such encodes already.
