Intel Considers Hardware Acceleration For Google's WebM Format
CWmike writes "Intel is considering hardware-based acceleration for Google's new WebM video file format in its Atom-based TV chips if the format gains popularity, an Intel executive said on Thursday. Announced last Wednesday at Google I/O, WebM files will include video streams compressed with the open-source VP8 video codec, which was acquired by Google when it bought On2 Technologies in February. 'Just like we did with other codecs like MPEG2, H.264 and VC1, if VP8 establishes itself in the Smart TV space, we will add it to our [hardware] decoders,' said Wilfred Martis, a general manager at Intel's Digital Home Group."
WTF? Intel might add hardware support for this codec and you declare victory? Intel is a bit-player in hardware video decoders. H.264 is already everywhere. Also, I don't know where you get the idea that it's patent-free. You simply can't make a modern video codec without treading on someone's patent any more, and this is no exception. Remember MS proudly announced that VC1 was patent-free, too. It's all a form of corporate trolling.
Its not the first time that someone has had to build and incredibly similar version yet slightly worse, just to fill a civic need. On2 is doing what Tesla did when Edison prevented him from using his lightbulb design.
How long will it take to get popular? 1 year?
How long will it take to design the hardware implementation? 6 months?
How long will it take to get into production? 1 year?
How long will it take to get into a product that is on store shelves? 6 months?
This is too long. Intel may as well have said they aren't interested. 3 years from now there are going to be how many tens of millions of devices with hardware H.264 support and no way of upgrading to VP8 support? People aren't going to toss these things in the trash just so the can buy brand new devices that give them the exact same experience.
Given that VP8 is really just a minor modification of h.264, and Intel already supports h.264 decoding in hardware, what exactly has to be done to support VP8? Modify the driver to reload the proper DCT constants and other minor things. The only hardware stuff I can see is if the hardware is hardwired for h.264 in which case they need to rewire it to be a little more flexible. But given they support many codecs already with the same pieces, it should be already in place (a lot of other pieces get reused decoding VC-1, for example).
Surely all the h.264 blocks could be re-used as VP8? In which case Theora's practically dead because everything supports h.264 decoding already and can probably be trivially converted to support VP8 as well.
Heck, you probably can do the same with an h.264 encoder to have it spit out a VP8 bitstream...
google is an h.264/MPEG-LA licensee so they won't be affected if it infringes on that patent set.
They will be affected if such additional usage of patented items requires the payment of further license fees that they did not anticipate.
It is a miracle that curiosity survives formal education. - Einstein
I have used VP8, and I'd also say "VP8 doesn't suck." I will agree that On2's website for it is misleading. I will also agree that H.264 does a better job of compressing video than VP8. However, just because VP8 falls short of the best performing format available doesn't mean it sucks. VP8 still is a good format, and I would argue that it is the best patent-free format for web video, assuming Google is correct about its patent status.
What a ridiculous statement.
H.264 is massively entrenched. Which content do you think you're going to get in VP8? DirecTV already adopted H.264. Cable is stuck on MPEG-2 at the moment, but will definitely take whichever format allows them to use their limited bandwidth most efficiently (H.264). What about cable in Europe, DVB? Nope, that already went to H.264. Will pirates give up a little bandwidth to use a free CODEC? They're already pirating content, you think they care about licensing fees for CODECs? Blu-ray? H.264 (and VC-1). CD-DVD? AVS.
The bizarre part to me is that you got modded up for your comment. Who thought this comment added to the conversation?
I think VP8 will get some use, but victory over H.264 is extremely unlikely. It's just too little, too late to take the lead.
http://lkml.org/lkml/2005/8/20/95
For the devices using a SIMD engine and a DSP to do H.264 with instructions in firmware, it's just a firmware upgrade to support other codecs that are substantially similar. That maybe not be the approach everyone's using, but it seems a likely combination for some devices. Lots of ARM-based devices tend to use this, if I'm not mistaken. Cell phones and tablets would be a good place to not have to spend an extra dollar for licensing costs, too.
Exactly, see the VP8 technical analysis by Dark Shikari
The very example the OP refers to (DCT) is telling:
H.264 uses an extremely simplified “DCT” which is so un-DCT-like that it often referred to as the HCT (H.264 Cosine Transform) instead. This simplified transform [...] can be implemented entirely with adds, subtracts, and right shifts by 1.
VP8 uses an extremely, needlessly accurate version[..]