2010-05-20, 21:38 | Link #21 |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
I could imagine a specialized algorithm for each non-homeomorphic triangle smaller than an 8x8 square and bigger than 4x4... That's be around ~100 different combinations.
I'd have to look more closely at the various DCT algorithms that exist and whether that'd be reasonable computationally.
__________________
|
2010-05-21, 00:51 | Link #22 |
( ゚∀゚)アハハ八八ノヽノヽノヽノ \ / \
Join Date: Apr 2010
Age: 36
|
it'd be nice if we could somehow detect all the lines and shapes in each frame and redraw them using vectors so that they could be scaled on the decoder side, without (dramatically) effecting quality.
somehow mastering each frame...err tracing it and vectorizing it, then having each frame be a svg, but with temporal reference so you could have accelerated decoding somehow... it wouldnt really be b-frames would it? what were the others? I and P? i need sleep, i dont even know if this made sense. i'll assume it didnt. |
2010-05-21, 07:49 | Link #23 |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
It would take some serious research, I think, but I could imagine having a collection of 8x8 transforms, one for each non-homeomorphic triangle that fits inside an 8x8 square... About 8*64/2=~256 different mostly sparse integer matrix transforms that act like the DCT similar to the fixed integer transform in h.264.
You'd fix a side and populate a sparse matrix then apply the appropriate transform. That way the process could be as computationally efficient and perfectly reversible as usual. Or perhaps it's not that bad and you could use the 24ish possible combinations embedded in a 4x4 matrix to algorithmically generate the 8x8 and 16x16 transforms for larger triangles instead of forcibly splitting them up smaller. I'm not confident the math would work out so easily for that, though.
__________________
Last edited by Quarkboy; 2010-05-21 at 08:23. |
2010-08-29, 07:21 | Link #30 | |
Far out, man!
Fansubber
Join Date: Jul 2008
Age: 40
|
Quote:
|
|
2010-08-29, 08:04 | Link #31 | |
x264 Developer
Join Date: Feb 2008
|
Quote:
Of course, there's a special irony to complaining about "free" when dealing with fansubs, which are technically illegal to begin with. |
|
2010-08-29, 09:17 | Link #32 | |
Far out, man!
Fansubber
Join Date: Jul 2008
Age: 40
|
Quote:
Also, I thought we were on the subject for using H.264 as 'standard' format for HTML5 <video>, not fansubs. Everybody knows that fansubbing violates the Berne convention in pretty much every case. |
|
2010-08-30, 04:59 | Link #34 |
Far out, man!
Fansubber
Join Date: Jul 2008
Age: 40
|
But Google gives free patent licenses. MPEG-LA doesn't. As for the indefinite extension, the summary document doesn't reflect that change yet. Does that also cover free patent licensing grants for other methods of distributing H.264 content than the internet?
|
2010-08-30, 08:56 | Link #35 | |
Pioneer in Fansub 2.0
Join Date: Aug 2007
|
Quote:
And the answer to that is obviously no.
__________________
|
|
2011-01-07, 16:31 | Link #36 |
Junior Member
Join Date: Jan 2011
|
I don't think the choice of video codec should be based solely on how well it performs currently. You do need to think about how users will be watching the videos. Fansubs are a type of videos, and often a part of the web too. The development of HTML5 should have a weight in the choice of codecs by fansubbers.
As you might know, Firefox doesn't support h264 in HTML5 video, so it can only rely on Flash to play h264 videos. This would not be a problem if Flash always worked well, but haven't you had experiences of buggy, sluggish video due to problems in Flash or the various Flash video players? I understand that the situation is better in Windows than in Linux or Mac, and that not all Flash players are buggy, and a non-Flash video player isn't necessary bug-free. Still, with HTML5 you only need a working browser, instead of dealing with potential bugs in one of many Flash players chosen by the webmaster. Speaking of webmasters, the code needed to embed HTML5 videos is cleaner than for Flash too, and you don't need to host a Flash movie in addition to the video. Almost all videos I watch online are either on YouTube, or are animes. YouTube already converts most new uploads to vp8, so if fansubbers start using vp8 too, Firefox users like me would be able to significantly reduce the need for Flash. There are "advantages" with Flash videos too: you can embed ads and prevent the user from downloading the video. But come on, are they really advantages for users? Are we more likely to support a website because its contents are easy to get, or because it forces us to watch/click ads and stay on that site? Firefox users have requested it to support h264 video, but Mozilla does not think it can legally do so without paying MPEG-LA. There's little evidence for the contrary either, because while Google Chrome and Safari can play h264 videos without Flash, Google and Apple are licensees of MPEG-LA. Fansubbers and fansub watchers might be already skirting the law, but isn't it still better to avoid potential lawsuits from one more company? Last edited by hagabaka; 2011-01-08 at 05:17. |
2011-01-08, 02:30 | Link #37 | |
Excessively jovial fellow
Join Date: Dec 2005
Location: ISDB-T
Age: 37
|
Quote:
You seem to be under the impression that it's a good idea to reuse fansubs as-they-are for web streaming. This is a retarded idea for the following reasons: 1) HTML5 doesn't support ASS subtitles and WebSRT or whatever they are calling it these days is a bad joke, or at least it would be if it was something more than vaporware, which it isn't. 2) >3mbit/s extremely VBR video streams are extremely unsuited to a streaming environment. You'd have to reencode anyway to get a file that is useable as a stream to most people. 3) Your web browser is not a good media player. No, really, if you think it is, get the fuck out. I like (ironically) your herp-derp concern trolling regarding lawsuits too, but it's even dumber than the rest of your post. Even if the actual argument (that it'd be a good idea to try to avoid software patent infringements when encoding warez videos) had any kind of merit at all (it doesn't, you're retarded), it would be sorta invalidated by the fact that the responsibility for getting a patent license with the MPEG-LA lies with the entity distributing your media player and/or H264 encoder, not with either a) the person encoding the video or b) the person watching the video. If you're going to play internet lawyer and hurf a durf about patents, at least read up on the MPEG-LA licensing terms first. It's not like they're hard to find. Edit: I also find it funny how you think Firefox is the One Browser to Rule Them All. Safari and IE do not support VP8.
__________________
|
|
2011-01-08, 06:13 | Link #38 |
Translator, Producer
Join Date: Nov 2003
Location: Tokyo, Japan
Age: 44
|
Just as an on-topic aside:
Has there ever been a fansub released in VP8? I recall there being a bunch of fansubs over the years that were ended up in odd and/or dying formats... I had some .asf's of Utena (actually those wer VHS captures), and of course there were .rm files back in the day (still going strong in china, it seems). There was one group that insisted on releasing in VCD compatible mpeg-1 files... Was there ever a group that released in Dirac or Snow?
__________________
|
Tags |
codecs, google, h.264, mpegla, patents, vp8 |
Thread Tools | |
|
|