Slashdot Mirror
H.264 and VP8 Compared
TheReal_sabret00the writes with a snippet from StreamingMedia.com: "VP8 is now free, but if the quality is substandard, who cares? Well, it turns out that the quality isn't substandard, so that's not an issue, but neither is it twice the quality of H.264 at half the bandwidth. See for yourself."
480x360 and some really old video, seriously? If you're going to compare then at least compare with HD resolution, and even then you should probably compare with all low, medium and high bitrates.
However, it looks like H.264 kicks VP8's butt with high motion video. Some of the VP8 pictures were quite blocky too.
MPEG LA, the group that formed a patent pool for H.264, does not protect their licensees against all patent infringement - but just against patent infringement suits by their licensors, and only then in the limited case of the specific case of patents included in the pool, and only then for limited times.
Q: Are all AVC essential patents included?
A: No assurance is or can be made that the License includes every essential patent. The purpose of the License is to offer a convenient licensing alternative to everyone on the same terms and to include as much essential intellectual property as possible for their convenience. Participation in the License is voluntary on the part of essential patent holders, however.
So you are in no way more protected by using the restricted H.264 license than you are by using the open VP8 license in the US. In most of the civilized world there's no such thing as software patents, so the only issue is which one of these is technically best.
And now MPEG LA is trying to form a patent pool for VP8. Will wonders never cease? Patents are broken. Let us hope that Monday SCOTUS rules that software patents are void in RE Bilski.
Help stamp out iliturcy.
Once again someone is comparing a codec to H264 using some small as hell resolution.
Welcome to 2010, if it's not encoded at 1080p nobody cares.
How do you know that some of the differences (lacking certain block sizes, for example) are precisely to avoid certain patents?
Just because it is similar does not mean that it infringes patents. Look at inventions in the past (like the first steam engine, avoiding a patent by Watt).
I've skimmed those patents. One of them is a patent on the concept of streaming compression. I do not believe it is possible to create a codec that doesn't infringe on a few, because getting a patent is very easy, even if the idea is obvious and prior art is widely known. Throw a few hundred of those bad patents together into a pool, and the cost of systematically invalidating every single one in court would be so great that it becomes cheaper to settle.
Unless you're going up against Google, for example. They might consider it worthwhile to spend the money to invalidate a few hundred patents. In which case MPEG-LA would risk losing its revenue stream. There's risk on both sides of this battle, and I can't see either party entering into it lightly.
Can You Say Linux? I Knew That You Could.
I think we need not worry about this in the long term. In the short term people are accustomed to the fact that their 9 month old phone doesn't have the latest technology - it's a phone, and they got it for cheap on contract. When their contract is up they go for the hot new stuff, which next week will include VP8 compatibility on Android phones and iPhones, which are both of the platforms that drive tech today. RIM will come around to whatever's hot because they don't want to lose share. As for the desktop, who needs it? Desks are not comfortable and they're not mobile. We move about now. We go where the work is, or we work wherever we happen to be.
Since I'm posting I might as well throw in some gems I've gleaned from the news. The ringtone hopes of phone vendors of being media content providers is pretty much dead. In the 2009 numbers online distribution has surpassed physical distribution for the first time (and we're not going back). Most audio is now bought online. One in four tracks purchased is bought through iTunes now. Amazon MP3 at about 1% is still in the top 10, but it's not going to be the wunderkind of media distribution once hoped.
If iTunes gets serious traction on video sales we're on our way to an iTunes culture. From my POV that would be unfortunate.
Help stamp out iliturcy.
The comparison seems to use sorenson squeeze (based on MainConcept if I am not mistaken).
I don't believe it can mach x264's capabilities and speed.
Using x264 for comparison would be much fairer.
The comparison is done by Jan Ozer. He's billed as a "video codec expert" but I don't think he has the technical expertise to, for example, make use of x264. His previous H.264 versus Theora comparison wasn't very impressive either. The x264 developers described Ozer's Theora versus H.264 comparison as "one of the worst articles they had ever seen".
How do you think the comparisons are staged, biased, and otherwise completely uninformative?
Because your post was nearly as uninformative as the patent smears...
The company had been around since the early 1990's. They were well aware of video patents, and monitored patent filings quite closely. Many of their features were adopted on the day that the statutory 1 year gap between publication of a method and possible patent filing expired. Much of the VP8 codec is actually prior art for the patents in the H.264 pool. On2 codecs have been used in Theora, Flash and Microsoft video products. If MPEG LA goes after them, it seems likely MPEG LA will lose more than they win - especially since all of us will be against them. Additionally, they'll be in court facing off with their patents against Google, and I hear Google has a few folks who know how to look stuff up like prior art. Heck, Google probably did this looking up before they decided to spend a hundred million dollars on buying the company just to give away its technology. It seems likely Google did look some stuff up before they decided to transcode their entire YouTube library to VP8. They're diligent like that.
And so having done the math, MPEG LA is investigating creating a patent pool to support VP8. If you can't beat 'em, join 'em. It seems unlikely they'll find success in this, but they will try.
Help stamp out iliturcy.
Surely this point is moot?
Yes, google will be working on a new codec... So will the MPEG-LA.
The point is *now* we have Ogg Theora, VP8 and h264. Theora seems to be out of the race, given it's inferior quality to VP8 and lack of hardware decoding support, while being in no way superior to VP8.
Now the question is, which of h264 and VP8 is better. For now, the answer seems to be h264, by a whisker.
Yes, but it's not – as has been pointed out already, VP8 uses *more* power to decode, not less.
While you're doing that "see for yourself" stuff, take a close look at the data on how the files were encoded. I mean a really close look; put on your scientist hat and pay close attention. See for yourself that the test was staged to support the view that they're espousing.
Let's pretend my scientist hat is in the wash right now and not quite dry yet. Would you care to share what makes you believe that? Because I don't call a one kilobit difference in bitrate "staging".
Unless perhaps you mean that equal bitrate doesn't necessarily mean equal quality due to different compression algorithms? That would be true, but it would be irrelevant in this case since I'm fairly sure the purpose of the test is to see which one can deliver the best quality within a certain bandwidth limit.
Seriously, do enlighten us instead of implying we are gobbling the data and aren't "true scientists" if we don't come to the same conclusion. Nice syllogistic fallacy.
For the other big non-desktop market, TVs/STBs - that's going to be the biggest resistance to anything non-H.264 (and I say this working in that industry...) The cable and satellite industry just spent a huge amount of money converting all of their broadcast systems and set-tops to H.264, and they don't like doing that very often. Additionally, the chips that go into all TVs and Blu-Ray players all support H.264 (and MPEG2, etc) but not much else - which means the current VOD services that run on them like Netflix, Vudu, CinemaNow, YouTube, etc are all H.264 based.
All of the major video providers on the Web - even the porn ones - are now migrating to HTML5 and seem ready to offer their content on whatever Codec you have handy. They're not choosy. They don't have a dog in this fight. They're all about getting eyeballs on their content so they can sell ads against that, or sell access to their content. They really don't care. If you have flash, they'll give you flash. If you have VP8 or H.264, they'll give you that. They can afford to transcode and store three copies, or transcode on demand. We are after all paying them to serve us up the video we desire.
So have you heard about Google TV? It seems Google is about ready to offer TV over IP. That's going to hose up the business model for most cable TV providers. It's disruptive. If Google delivers TV well I'll be moving to IP only on my Comcast cable connection (right now I have the triple play), and I imagine I'm not alone there. I've got 50Mbps down, and that's more than fine both for my web surfing and to drive all of my TVs with video. It will probably put the brakes on the one hour a year of local programming I usually watch, but I won't miss it.
Help stamp out iliturcy.
None of the comparisons are of exactly the same frame. 3 of the 6 images have different times in the corner.
I suspect the writer selected frames so H.264 won, but gave VP8 one win at the end to not seem biased.
Also his 'standard SD encoding test file that I've been using for years' would also be a source of suspicion. It is possible that his source file is already in a format that encodes better into H.264 than it does into VP8. And has already been mentioned here the resolution of the source is quite low for todays HD broadband world...
He used Sorenson Media to encode the files. In all probability they may be just better at setting some of the encoding parameters in the codec they have had longer...
And that was just a quick look. There are possibly other flaws that I haven't noticed yet.
[The Universe] has gone offline.
Add India and Pakistan to China, and they're most of the civlized world. They're actually more than half of all the people in the world. None of them care about your list (Yes, I know Chinese patents are on your list - even the Chinese don't care about them - China has differring views on intellectual property that are difficult to describe here but can be summarized as: meh).
We forget sometimes in the US that our entire country is not as old as a decent British country house, nor a Taiwan temple, nor even a Chinese family land lease. Hell, the US is not even as old as most decent books. We are not most people and we're never going to be. Our inflated estimate of our importance is the cause of much misunderstanding in the wider world. The sooner we let it go the better.
We've got some decent insight on human interaction to share, but others may be rightfully suspicious of new ideas when they have a system that's similar that is proven to work over a span of 5,000 years. To those folk a quarter millenium is still just a "noble experiment", and frankly looking at what we're doing with it, we might not make it to a half millennium so who are we to say they're not civilized?
Help stamp out iliturcy.
Patent reform:
make the PTO legally responsible for the validity of the patent.
If the patent is attacked in court and ruled invalid, the PTO will cover legal fees, and return submission and upkeep costs plus interest.
Nothing else will make them do their job right.
Now the question is, which of h264 and VP8 is better. For now, the answer seems to be h264, by a whisker.
The other question of license fees seems to have slipped your mind. Not only that, but what kind of terms and conditions you must agree too to get that license is also another very important question.
The Grey Goo disaster happened 3 billion years ago. This rock is covered in self replicating machines!
Who in their right mind would compare codec quality by encoding screenshots in a lossy image format??? To add insult to injury, the GIF image files in the article have a .jpg extension.
This article, and the person who wrote it, are worse than useless.
The immediate problem with this article is that it uses the Sorenson encoder rather than the state-of-the-art x264 encoder for h.264. If x264 was used, the h.264 encodes would demonstrate higher quality and the quality conclusions would most likely be more in favor of h.264. Since x264 is both the best h.264 encoder, and FOSS, it is the ultimate benchmark for any new video codec implementations, and that should be used. The point of VP8 however is that it is now the best free video codec, replacing Theora in that category (which is still being improved and will probably remain relevant in some niche scenarios still). The quality of VP8 is likely not going to surpass h.264, even with open source tinkering, but it will still revolutionize the web through html5 video, it will achieve widespread software support in a matter of a few months, and your devices will pick up support in a year or so (the next generation hardware). VP8 is free, and good enough to be in the ballpark of h.264 even if it is not as good. And that is a huge win.
https://dalgamotor.wordpress.com/ - Elektronik beyinlere ozgurluk asisi (Turkish)
Why exactly? Apple's previous track record with content is not only getting the music industry to agree to sell content on the internet, but to (eventually) do it without DRM. Even if they weren't the very first to offer it, they made it a profitable and viable model that broke the industry's mindset that was firmly "why would people pay when they can just get it off a p2p service".
In their history they tend to go for (if patented) at least open codecs and formats (with the exception of Sorenson in Quicktime, but it didn't last long).
How do you know that some of the differences (lacking certain block sizes, for example) are precisely to avoid certain patents?
Technically, the H.264 format has zero patents. The patents are all related to working on it and ways to generate it. That, however, is a non-trivial problem and reinventing the wheel is so costly and technically difficult even without infringing, that it's not really an option. If you start by the time the standard is adopted you won't have a product until it's obsolete. So the practical method is to simply license what you need and build on existing work. The $100k or whatever the licensing will cost doesn't buy a whole lot of engineering effort. Startups burn through $10-$20M in this field, for a team of 10-15 to work for a couple of years, usually on a limited production or consumption side component to fit in the chain. TO the rest, cost of licensing is really not a significant problem to anyone out there -- other than open source projects, obviously.
They probably have all the material ready. After all, let's say you wanted to look at the claims of a couple thousand patents. You then wanted to look for prior art out there. What would you need? Well first you'd need access to a massive repository of information for all over the web going back years. You'd then need to be able to mine this information for specific pieces of data quickly and efficiently... Wait a sec, that is PRECISELY what Google does. They are the uncontested kings of data crawling and mining. They can search like nobody else, and are extremely good at pattern matching and so on.
As such it is reasonable to assume that they have done their homework, and that they could hit back hard with prior art and the like in the event of a suit. They've said as much as well. They say: "We have done a pretty thorough analysis of VP8 and On2 Technologies (VP8's developer) prior to the acquisition and since then, and we are very confident with the technology and that's why we're open sourcing."
Sounds to me like they've researched all this and said "Yep, we can win this one if it comes down to a fight."
Reminds me of the guy who patented the wheel, er.. I mean a "circular transportation facilitation device": http://news.bbc.co.uk/2/hi/asia-pacific/1418165.stm
Seriously, the whole patent system has to be scrapped and recreated from scratch. It's terminally broken, far beyond any hope of salvation by reformation.
The H.264 video on this comparison site is encoded using Baseline Profile, which is really only for low-power mobile devices, and is not representative of what H.264 is actually capable of. Switching to Main or High Profile gives us CABAC coder (10-20% improvement), bi-directional frames (20-40% improvement), adaptive 8x8 DCT (3-5% improvement). A MP or HP H.264 will blow VP8 out of the water every time. The fact that H.264 manages to look better in most cases despite being encoded using Baseline Profile (and Sorenson Squeeze, which doesn’t seem to have ever been compared to other H.264 encoders and probably is not as good as x264) is a pretty damning assessment of how good VP8 actually is—that is to say, not very.
[insert witty comment here]