WebM Codec

[hamedhemmati]

Is webM codec going to be added to the list of codecs to choose from? The VP8 quality looks really good.

[s55]

Decode support will likely come for "free" (that is, little or no development work) when ffmpeg adds it and HB updates. (which is infrequently)

Encode support will require a fair bit of work and as it stands, there doesn't appear to be any developer interest in it. (In other words, there are currently no plans to add it, (that may / may not change) so it'd require a patch from someone up to the task.

As for the quality, I personally was really unimpressed. It looks very soft / blurry in comparison to x264 although it's certainly an improvement over Theroa.

[hamedhemmati]

I thought Google has already made the patches available. This is the page where they have the patches for FFmpeg.
http://code.google.com/p/webm/downloads/list

The sample video On2 has put out has the smoothing on I want to actually have the chance to play with it and see the quality and size at HD level not low quality video.

[s55]

Yes patches for ffmpeg are available. That doesn't change anything.

[mduell]

The VP8 quality looks really good.

Huh? Neither the standard (such as it is) nor the encoder (ugh) look "really good."

[tlindgren]

The VP8 quality looks really good.

Huh? Neither the standard (such as it is) nor the encoder (ugh) look "really good."

Right, some Theora developers has latched on to the fairly decent PSNR value compared to x264 Baseline (if we ignore the slowness). Unfortunately as he has show several times before the PSNR value is often useless and sometimes actively bad because it rewards blurry encodes! From other comments and snapshots VP8 doesn't do well in actual quality and thus likely in SSIM.

There's an older article by him why looking at PSNR too much during development can be very bad for your encoder with examples from various h.264 techniques and encoders that's well worth reading first. Based on VP3 and VP8 it seems like On2 really likes PSNR, lately Theora has been repairing some of the worst results of this in what was originally VP3.

The "real" test is actual visual quality, unfortunately that's much harder to measure, though often SSIM gives you a fairly decent estimate (certainly light years ahead of just using PSNR). But there's a reason why he's already involved in trying to arrange a test using actual visual quality, it is the golden standard.

Best line in the article...

While the C code isn’t half bad, the assembly is clearly written by retarded monkeys.

In comment 56 he gives his best guess on where it is now and where it could become in the future, comparing it to current x264.

Hard to say without testing it, and it takes a long time to encode with VP8 on its best encoding mode. If I had to make a guess, I’d say 15-25% for theoretical future as-good-as-x264 vp8 and 40-60% for current vp8. It depends heavily on the content though — in a case where x264’s psy opts help more, vp8 will need more extra bits to compensate.

[dynaflash]

Ultimately it comes down to some developer wanting to write it into hb. good or bad.

I would suggest a case in point. Right now HB would benefit from moving from our current faac encoder for aac audio to ffmegs aac encoder. There are some "issues" with the license of faac. However the sheer inferiority of ffaac to faac makes that all but unworkable at this point in time.

Philosophical merit imo, does not replace technical merit all other things being equal. Otoh this is just my .02 .

[creamyhorror]

If the goal is quality, x264 is the way to go for the foreseeable future. HB isn't particularly directed toward the creation of web video anyway (which is the major point of WebM/opensourced VP8).

If patent issues concerning VP8 are settled (either in court or out) and a capable team starts working on bringing the encoder to a competitive state, then I think VP8 encoding support will naturally become of more interest to everyone. It'll no doubt be easier to bring the encoder into Handbrake at that stage, too.

[towolf]

Another case in point, Handbrake could with good conscience be recommended as the one-stop shop to encoding web videos, if it would encode both H.264 and webm (possibly in auto-batched pairs).

[towolf]

HB isn't particularly directed toward the creation of web video anyway.

It isn’t? But it also doesn’t seem to discourage creating videos for the web either.

[mduell]

Another case in point, Handbrake could with good conscience be recommended as the one-stop shop to encoding web videos, if it would encode both H.264 and webm (possibly in auto-batched pairs).

Who says HB wants to be that? Certainly not the project leadership.

HB isn't particularly directed toward the creation of web video anyway.

It isn’t? But it also doesn’t seem to discourage creating videos for the web either.

It doesn't support the freetard's previous propsal, theora/vorbis in ogm. But it doesn't discourage you from producing h.264 in mp4 for the web, with the progressive streaming option.

[towolf]

Another case in point, Handbrake could with good conscience be recommended as the one-stop shop to encoding web videos, if it would encode both H.264 and webm (possibly in auto-batched pairs).

Who says HB wants to be that? Certainly not the project leadership.

Are you some kind of spokesperson for the project? How come I can’t find you in AUTHORS?

Anyway, increasing mindshare and use cases certainly can’t hurt.

HB isn't particularly directed toward the creation of web video anyway.

It isn’t? But it also doesn’t seem to discourage creating videos for the web either.

It doesn't support the freetard's previous propsal, theora/vorbis in ogm. But it doesn't discourage you from producing h.264 in mp4 for the web, with the progressive streaming option.

... freetards ...

You’re aware that HB’s capabilities largely rely on the work of what you could also call freetards? So what tards is Handbrake for instead?

[JohnAStebbins]

http://www.urbandictionary.com/define.php?term=Freetard

[towolf]

http://www.urbandictionary.com/define.php?term=Freetard

Oh. Thanks. Sounds like a misapplication of the term then.

OGM was never proposed for web use by anyone, that was OGG. OGM was a dubious hack by some DVDtards. Some webtards wanted OGG because Theora goes naturally into OGG and it has no royalties. So they probably were econotards rather than freetards. No?

Also funny that Google shelled out 120 million to become part of freetard club. Must be exclusive club.

Disclaimer: Personally I love H264 to (transparently) compress the hell out of my 22mbps Canon Ixus P&S videos.

[mduell]

Are you some kind of spokesperson for the project? How come I can’t find you in AUTHORS?

Absolutely not, but I can read the statements of the project leadership in the forums and IRC channel.

Anyway, increasing mindshare and use cases certainly can’t hurt.

It always comes at the expense of something else. Is this what the developers want to spend their time on? Not today.

... freetards ...

You’re aware that HB’s capabilities largely rely on the work of what you could also call freetards? So what tards is Handbrake for instead?

HandBrake certainly stands on the shoulders of giants. There's a huge gap between useful open source software and freetards.
Theora is freetard. x264 is incredible. Ogg is freetard. libmp4v2 and libavformat are great.

[towolf]

There's a huge gap between useful open source software and freetards.
Theora is freetard. x264 is incredible. Ogg is freetard. libmp4v2 and libavformat are great.

.... Oh I see.

Now comes a catch 22 for you. What about FAAC? Freetard or incredible?
And Vorbis, freetard or great?

[JohnAStebbins]

faac isn't freetard or incredible. In fact it's got copyright issues and is on the crappy side of mediocre.
vorbis is both freetard and incredible. It's the only good thing to come out of that lobbying group called xiph.

The point is, being free doesn't make it great. The freetards make a habit of calling something great because it is free and not based on it's real world merits.

[mduell]

There's a huge gap between useful open source software and freetards.
Theora is freetard. x264 is incredible. Ogg is freetard. libmp4v2 and libavformat are great.

.... Oh I see.

Now comes a catch 22 for you. What about FAAC? Freetard or incredible?
And Vorbis, freetard or great?

Faac: useful (read: least worst)
Vorbis: freetard

[towolf]

faac isn't freetard or incredible. In fact it's got copyright issues and is on the crappy side of mediocre.
vorbis is both freetard and incredible. It's the only good thing to come out of that lobbying group called xiph.

The point is, being free doesn't make it great. The freetards make a habit of calling something great because it is free and not based on it's real world merits.

I don’t care about dmuell, but I’m disappointed that you take a similar stance, being the GTK developer guy and obviously knowledgeable.

There are a number of great codecs under the Xiph roof that all have outstanding technical merits. Vorbis, FLAC, Speex, CELT are either successful or technically cutting edge. With Theora they took a nasty code drop over the wall and made something proper with it. Don’t forget it’s really old and limited. In this regard it is similar to the LAME project.

But in business (here: web) you have to evaluate the relative technical and legal/financial merits. I’ve already seen people bash VP8 now because initial encodes after the drop look somewhat poorer than x264. Maybe they should have a chat with Nero about how they find dealing with the respective tards on their side.

I would say that the existence of Theora (and its obvious limitations) led directly to the release of VP8. And I doubt anyone will say that will have bad consequences for the market.

[creamyhorror]

faac isn't freetard or incredible. In fact it's got copyright issues and is on the crappy side of mediocre.
vorbis is both freetard and incredible. It's the only good thing to come out of that lobbying group called xiph.

I see there's quite a bit of alignment between your opinion and some of the community on Doom9 (e.g. DS)

The point is, being free doesn't make it great. The freetards make a habit of calling something great because it is free and not based on it's real world merits.

Nevertheless, it's useful to have strident voices bashing the MPEG-LA for the threat of royalty extraction and to have corporate giants challenging it by championing a "freetard" standard. In that light I think they certainly play a useful role, especially if they succeed in making MPEG-LA relax its terms even further.

[JohnAStebbins]

I don't have a lot of love for mpeg-la either. I'd say I find mpeg-la and freetards equally distasteful. But quality matters to me. I don't care much who is backing a format. It wins on it's merits with regard to quality. And when freetards try to argue that VP8 and theora are competitive quality-wise, well its just silly and insulting.

[dynaflash]

... and as I said above, all of this becomes meaningless with regard to HandBrake unless some dev wants to do the work to put it in. The rest is pretty much semantics. MKV in my opinion had as much or more to offer and it took saintdev deciding to spend the time to write the code to put it into handbrake. Thats how it goes.

[s55]

and Locked. My original point still stands.