View Full Version : VP8 (Webm) Vs Flash(H264)
zerowalker
7th August 2012, 20:16
When i read about compares about the 2, all say that H264 is the shit if you know what i mean.
I myself Love H264 with x264, itīs Amazing!
But when i watch Youtube and compare Webm with Mp4, the Webm is better by quite a bit, havenīt done a Major test, but did compare a bit on 2 videos, and it was always better in Webm.
Can anyone tell me why that is?
Is Vp8 better at youtubes settings(bitrate limit)?
Or am i just blind?
sneaker_ger
7th August 2012, 20:22
It comes down to the settings (incl. bitrate), but in every fair comparison you can find x264 beats the VP8 encoder effortlessly.
zerowalker
7th August 2012, 20:23
Can you direct me to some comparasion (Pictures or Clips)?
Cause if x264 beats it without problems why is it worse on Youtube(though they donīt use x264, it should be pretty close right)?
Dark Shikari
7th August 2012, 20:32
Not all of Youtube's video modes use the same bitrate or comparable settings. If you want to do a comparison, you should do it in a controlled situation, not pull random videos off the internet.
zerowalker
7th August 2012, 20:40
Oh, i thought they used the same on letīs say 1080p etc, (the videos were My own videos, so i had the Raw original to compare to).
But i guess they use higher settings for some reason, though the size was the same i guess they use some pretty fast h264 settings to get by?
I have wanted to do that, but i canīt get the encoder to work when i made a build.
Can you link the latest encoder, or build it for me if you have the possibility?
Thanks:)
pandy
8th August 2012, 12:30
Google promote WebM as replacement for H.264 - they explicitly refuse to support H.264 even if H.264 is better, even if it better for customer.
smok3
8th August 2012, 14:08
Oh, i thought they used the same on letīs say 1080p etc, (the videos were My own videos, so i had the Raw original to compare to).
But i guess they use higher settings for some reason, though the size was the same i guess they use some pretty fast h264 settings to get by?
I have wanted to do that, but i canīt get the encoder to work when i made a build.
Can you link the latest encoder, or build it for me if you have the possibility?
Thanks:)
Can you change topic to "comparing youtoobe encodes" or something more appropriate (such generalization is annoying)? Thanks :)
Atak_Snajpera
8th August 2012, 14:15
Google promote WebM as replacement for H.264 - they explicitly refuse to support H.264 even if H.264 is better, even if it better for customer.
At least we will get Opus audio codec (royality free and open source) instead of AAC. Latest listening tests show that at 64kbps opus beats Apple/Fraunhofer HE-AAC ! Also turns out that Vorbis will be dead soon ;)
pandy
8th August 2012, 16:30
At least we will get Opus audio codec (royality free and open source) instead of AAC. Latest listening tests show that at 64kbps opus beats Apple/Fraunhofer HE-AAC ! Also turns out that Vorbis will be dead soon ;)
Yes but... - new codecs are good for PC (CPU vendors) and some mobile devices but hey are bad for industry with HW decoder implementations - even if HW is primitive and specialized semi programmable DSP still it can be less efficient and some times can't be reprogrammed at all - they introduce new codecs faster than market can adopt them - this is insane.
SeeMoreDigital
8th August 2012, 18:56
Well... given H.262 was adopted for std-def TV broadcasts. H.264 was adopted for high-def TV broadcasts. What's the betting H.265 will be adopted for ultra high-def TV broadcasts :D
Keiyakusha
8th August 2012, 20:30
they adopted subset of the h264 standard for high-def stuff. Dunno why. Maybe because otherwise it will be too expensive to produce hardware that can decode it and they will have hard time filling all space on blurays? If that's so, by adopting h265 they'll lose more than win, money-wise. cause it will be more cpu-hungry and they will have to sell 1BD where now they sell 2. Or if they'll adopt subset again, it will be no better :D
EDIT: however they may adopt subset and force everyone to buy new hardware to get the same stuff (from user's point of view)... that will be no good =)
zerowalker
12th August 2012, 19:11
Okay but one last thing, will VP8 be able to Beat/Compare to X264 for itīs use across the internet?
Or do you think itīs a failed project?
And about X265, i read that itīs supposed to have up to 50% less files for the same quality as X264, is this true or overestimated?
Thanks:)
Atak_Snajpera
12th August 2012, 19:36
regarding vp8
dark shikari says noooo ...
regarding future x265
in theory yes ...
zerowalker
12th August 2012, 20:05
so vp8 is not the "shit" if i say so?
Whatīs the problem with it?
I mean, itīs Open Source and all, isnīt it what we have been waiting for?
J_Darnley
12th August 2012, 20:11
Please understand that there is a difference between a video format (H.264 and VP8) and a video encoder (x264 and libvpx). If I remember Dark Shikari's writings about VP8 correctly, VP8 could be a decent enough competitor to H.264. The encoders are a different story. As far as I know, libvpx does not come close to x264 despite some "recent" improvements to libvpx. The xvp8 spin-off is still a work in progress.
As for the upcoming video standard widely known as H.265 or HEVC, it is widely said to be "50% better" than H.264 but there don't seem to be any encoders ready for a comparison.
zerowalker
12th August 2012, 20:17
Yes of course, but i donīt understand the thing with the different Encoders, i mean, from my perspective, H264 is H264 no matter what encoder, but this isnīt true right?
Even if the bitstream and all is following the standard, different Encoders do better or worse, and x264 is The Best(i think?) H264 encoder out there?
But that is the lack from libvpx (Googles vp8 encoder?) compared to X264(i understand that x264 has been massively improved by the community over the years.).
Atak_Snajpera
12th August 2012, 20:25
basicly vp8 standard does NOT have as advanced algorithms as h.264 specification.
vp8 is like mp3 where h264 is like aac or opus in audio compression
vivan
12th August 2012, 21:12
Whatīs the problem with it?It has only one advantage (that 99% people doesn't care about) and a lot of disadvantages.
- x264 is much better that any vp8 encoder
- ... and even if there was encoder as good as x264 - it still would be worse.
- zero hardware support. While any modern device supports h/w AVC decoding.
- even Flash doesn't support it, so you will have to use ugly and buggy html5 player.
So, who needs it? Only Google.
I mean, itīs Open Source and allOpen Source? Standart? What? Both H.264 and vp8 specifications are open - so anybody can write encoder/decoder.
we have been waiting for?who, and what they were waiting for?
basicly vp8 standard does have as advanced algorithms as h.264 specification.something have changed since this (http://x264dev.multimedia.cx/archives/377)?
VP8, as a spec, should be a bit better than H.264 Baseline Profile and VC-1. Its not even close to competitive with H.264 Main or High Profile. If Google is willing to revise the spec, this can probably be improved.
zerowalker
12th August 2012, 21:43
Oh, i thought that the H264 Standard wasnīt royalty free?
So even if we use X264 to encode to h264, we should still have to pay the royalty fee for commercial use, or something like that?
And with the HTML5, i though it was going to be, The Thing of The Year!!, as i mean, itīs not a plugin ad itīs out of the box, like playing from the PC itself.
But that doesnīt seem to be the case, as itīs as you say, very buggy and on Opera (my favorite browser) itīs worse of all i think, so i canīt use it even if i want to.
vivan
12th August 2012, 22:01
For web H.264 is royalty free. And not for web - only starting from 100,000 units...
I use Opera too, but even in Chrome HTML5 player is awful.
zerowalker
12th August 2012, 23:04
Oh, well that doesnīt matter to us then;D
Oh really?
I though Chrome would be, well good atleast;S
Itīs sad as i had good hopes for HTML5 and stuff, as i am not a fan of flashplayer;S
Atak_Snajpera
13th August 2012, 10:35
@vivan
obviously it was a typo from my side.
i ment "does not have" :)
schweinsz
13th August 2012, 11:01
And about X265, i read that itīs supposed to have up to 50% less files for the same quality as X264, is this true or overestimated?
Thanks:)
No, the HEVC project is only about 30% better than the H.264 on tools. There is no H.265 or X265, the official name is HEVC.
You can read the latest comparison in the document JCTVC-J0236 (http://phenix.int-evry.fr/jct/doc_end_user/documents/10_Stockholm/wg11/JCTVC-J0236-v5.zip).
Keiyakusha
13th August 2012, 16:50
HEVC was supposed to be able to give room for 50% improvement (or so I heard). referense HEVC encoder maybe beats reference AVC encoder but so far x264 beats both.
BTW for me speed also matters. Is it possible for some HEVC encoder to deliver 50% less filesize for the same quality at the same encoding speed as x264? I hardly believe even in 30% number... I don't need these 30% if it will take me week of continious 100% cpu load to encode a movie.
nevcairiel
13th August 2012, 17:05
The point of HEVC is to increase compression at the expense of computational complexity, which means it'll rather be slower then faster.
schweinsz
13th August 2012, 18:36
HEVC was supposed to be able to give room for 50% improvement (or so I heard). referense HEVC encoder maybe beats reference AVC encoder but so far x264 beats both.
BTW for me speed also matters. Is it possible for some HEVC encoder to deliver 50% less filesize for the same quality at the same encoding speed as x264? I hardly believe even in 30% number... I don't need these 30% if it will take me week of continious 100% cpu load to encode a movie.
You can read the JCTVC-J0236, the HEVC is with 22.1% better than the H.264 for all intra coding (IIII coding), 33.7% better for random access coding (the most efficient coding, hierarchical B coding), 36.7% better for the low delay coding (IBBBBBB coding).
There is no the highly optimized HEVC encoder currently. But it is easy to say that the HEVC decoding complexity is comparable to the H.264 decoder.
Keiyakusha
13th August 2012, 19:11
You can read the JCTVC-J0236, the HEVC is with 22.1% better than the H.264 for all intra coding (IIII coding), 33.7% better for random access coding (the most efficient coding, hierarchical B coding), 36.7% better for the low delay coding (IBBBBBB coding).
There is no the highly optimized HEVC encoder currently. But it is easy to say that the HEVC decoding complexity is comparable to the H.264 decoder.
Yes I understand that it may be better by xx%, but what price we should pay for it?
lets say with x264 i'm encoding source at 5fps with target filesize 500Mb and recieve some quality X
can some hypothetical highly optimized HEVC encoder encode the same source with the same speed (5fps or faster) and give result with the same filesize (500Mb or smaller) and quality X+30%?
If no, this is not much of an improvement. Sounds like something similar to x264 --preset uberplacebo ^__^
schweinsz
13th August 2012, 19:33
Yes I understand that it may be better by xx%, but what price we should pay for it?
lets say with x264 i'm encoding source at 5fps with target filesize 500Mb and recieve some quality X
can some hypothetical highly optimized HEVC encoder encode the same source with the same speed (5fps or faster) and give result with the same filesize (500Mb or smaller) and quality X+30%?
If no, this is not much of an improvement. Sounds like something similar to x264 --preset uberplacebo ^__^
Assuming that the future HEVC encoder is same optimized with the x264, you can expect the 30% better quality with the same filesize but I believe that you must pay by more computation load. In other word, you get the 30% gain by more computation load.
Besides, I don't believe that the x264 placebo is 30% better than the x264 slow.
Keiyakusha
13th August 2012, 19:49
Besides, I don't believe that the x264 placebo is 30% better than the x264 slow.
Of course no. What I said is far from real-world example. I just hoped that after all HEVC brings some new efficient algoritms. Having like 10-20% improvement without any additional computation load is probably a minimum that I can call real improvement that worth establishing the whole new standard. So far looks like this is just moneymaking machine for them, nothing more.
Audionut
13th August 2012, 23:29
Comparing HEVC with x264 atm is stupid. HEVC is only new, give it some time to be optimized.
x264 was by no means fast when it first hit the scene either. It's had 7+ years to optimize speed/quality.
http://forum.doom9.org/showthread.php?t=141352
A better comparison would be a build of x264 from around 2005/2006 with a current HEVC build. They would both have about the same development hours.
edit: it feels like not long ago, that everyone was complaining about the slow speed of x264 vs xvid.
pandy
15th August 2012, 11:50
so vp8 is not the "shit" if i say so?
Whatīs the problem with it?
H.264 is far more flexible and cover wider area of the potential use scenarios, also efficient H.264 HW decoders (low power consumption) are vital for portable devices - VP8 still is mostly implemented as a software not hardware thus it will reduce area potential uses to non-portable mostly.
I mean, itīs Open Source and all, isnīt it what we have been waiting for?
It have only open source encoder/decoder and available some documentation however H.264 have also open source encoder/decoder available and H.264 is much better documented... VP8 don't have even draft of standard comparable in terms quality of description to this available for H.264.
However VP8 is NOT open source as it can be (should be) - it was pointed at the beginning that VP8 is closed codec.
Also no royalties it is only claim of Google however real situation around VP8 is not so clear.
iwod
15th August 2012, 15:44
Assuming that the future HEVC encoder is same optimized with the x264, you can expect the 30% better quality with the same filesize but I believe that you must pay by more computation load. In other word, you get the 30% gain by more computation load.
Besides, I don't believe that the x264 placebo is 30% better than the x264 slow.
So by 50% they were referring not to H.264 but to MPEG 4 or what?
It sounds strange that internal documents shows 30% improvement while the PR went for 50%.
And it has been shown that encoder and tuning matters MUCH more then spec. So unless there is another x264 team alike works on HEVC, any HEVC wont be outperforming x264 anytime soon. Which is rather sad.
The good thing about HEVC though is that we can get rid of the H.264 profile silliness. The difference from Baseline to High is far too great. And we can finally have Apple to adopt a spec that offers potential higher quality video with dedicated Hardware decoder.
schweinsz
15th August 2012, 18:32
So by 50% they were referring not to H.264 but to MPEG 4 or what?
It sounds strange that internal documents shows 30% improvement while the PR went for 50%.
You can refer to the references of the document JCTVC-J0236, there are some old comparisons. There is the results for the IBBBBB comfiguration that shows about 50% gain.
mp3dom
15th August 2012, 20:48
And it has been shown that encoder and tuning matters MUCH more then spec.
Infact from a 'finished' HEVC encoder you can probably expect even higher ratio of improvements. The standard is quite new btw, you can't compare a new reference encoder with another encoder with years of development.
zerowalker
16th August 2012, 08:18
Is there any HEVC encoder released?
Would be fun to play around with it:)
sneaker_ger
16th August 2012, 21:55
http://hevc.hhi.fraunhofer.de/ (reference software, don't know where to get a binary)
But x264 will probably beat it in speed and quality.
schweinsz
16th August 2012, 22:26
http://hevc.hhi.fraunhofer.de/ (reference software, don't know where to get a binary)
But x264 will probably beat it in speed and quality.
You can open it using the vc2008 and build it (select the x64 release version).
Someone reported that the HEVC reference software can beat the x264 on quality.
sneaker_ger
16th August 2012, 22:29
Someone reported that the HEVC reference software can beat the x264 on quality.
That'd be nice. At what speed?
schweinsz
16th August 2012, 23:22
That'd be nice. At what speed?
Very slow, need more than 1 minute to encode one 1920x1080 frame (only one thread).
iwod
19th August 2012, 07:54
Very slow, need more than 1 minute to encode one 1920x1080 frame (only one thread).
Well if that is true then that is at least a very good start. I still remember H.264 reference wasn't better then xvid.
I tend to think speed is much easier to obtain ( at first ) then quality.
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.