H.264 High Definition Decoder

[grumpy]

Anyone know more about this (http://www.dvdinsider.com/news/view.asp?ID=4981)?

[MfA]

No, but it seems to be a hardware product ... so it doesnt get me excited.

[trbarry]

Actually it sounds like JVC/H.264 is sort of a nifty powerful codec but fairly CPU intensive to decode. So an eventual cost effective hardware decoder card might be just what the doctor ordered.

- Tom

[superdump]

Yes, it may spur on codec development somewhat if there's hardware support for it already. Also, does this mean that the specs for h.264 have been finalised? I know it was in the yes/no stage just recently. If the hardware could ENcode too, that would be very handy. :)

[shlezman]

I wouldnt break my piggibank to buy a press release :D
I'm just wondering how someone can announce a product like a decoder, when there are no encoders around, just a thought....
Ohhh I've got an idea.

[superdump]

What's your idea? There are currently no encoders (as far as I know) that can compete with the mature mpeg-4 and other codecs of today. But at least, this time, the hardware companies are prepared. Or at least ti seems that way. Hopefully the software will catch up. :)

[shlezman]

Originally posted by superdump
What's your idea?
Just sounds like someone is trying to make waves in stormy seas, but I've always been skeptic :rolleyes:
It's not the first company to declare H.264 RT codec (UB-Video declared it a year ago, and that's all I've heared since). ASIC's arent available yet, and I really hope that in "early 2004" we'll see the chip working, but doubt it. And no sencible company will order such ASIC's without testing them.
Originally posted by superdump
There are currently no encoders (as far as I know)
That's true ! so why annoncing decoders, who can tell if it's any good ? I guess developing decoder first is the right way to solve the chicken-egg problem of codec development, but announcing it long before any encoder is available, is just useless.
As for MPEG-4/H.264 comparison, I think it's a useless for two reasons. First the MPEG4 is too mature to be compared with a very basic implementation of the H.264 draft, which is the only one available. Second the H.264 is much more advanced and without any doubt will beat the MPEG4, the question is will it be 20% or 50% (bitrate for the same quality).
Originally posted by superdump
But at least, this time, the hardware companies are prepared. Or at least ti seems that way.
TI is always ready (Equator too) and aimed at new growing markets, that's our luck (as consumers). Hopefully it will lead to the revolution we (I) really hope to see.

[superdump]

Originally posted by shlezmanOriginally posted by superdump
There are currently no encoders (as far as I know)
That's true ! so why annoncing decoders, who can tell if it's any good ? I guess developing decoder first is the right way to solve the chicken-egg problem of codec development, but announcing it long before any encoder is available, is just useless.

VSS have a h.264 codec and hdot264 have one too. I believe both are quite closely based on the reference codec at the moment however. You shouldn't have only quoted the start of my sentence, it could have lead to some misinterpretation, but never mind. :)

Originally posted by shlezman
As for MPEG-4/H.264 comparison, I think it's a useless for two reasons. First the MPEG4 is too mature to be compared with a very basic implementation of the H.264 draft, which is the only one available. Second the H.264 is much more advanced and without any doubt will beat the MPEG4, the question is will it be 20% or 50% (bitrate for the same quality).

H.264 will surely be better than the current mature MPEG4 codecs but I was saying that currently they are not. As for how much better, who knows?

I have seem some "h264" files in the ffmpeg cvs but I cannot get it to compile for windows. I will start a new thread on that though. :)

Originally posted by shlezmanOriginally posted by superdump
But at least, this time, the hardware companies are prepared. Or at least ti seems that way.
TI is always ready (Equator too) and aimed at new growing markets, that's our luck (as consumers). Hopefully it will lead to the revolution we (I) really hope to see.

TI? I meant "Or at least _it_ seems that way.". Lol. :)

Have fun.

[Tommy Carrot]

Originally posted by superdump


I have seem some "h264" files in the ffmpeg cvs but I cannot get it to compile for windows. I will start a new thread on that though. :)



This seems to be the most active and most promising h.264 development. They say the decoder is functional and the encoder will be as well quite soon. So hopefully ffvfw and ffdshow will soon get h.264 support.

hdot264 project seems to be completely dead (again), and VSS said not sure their encoder will be free (just the decoder). So the ffmpeg project stands alone.

[superdump]

Thanks for the info Tommy. I don't suppose you're any good with MingW are you? If so, could you have a look at my compiling FFMPEG for Windows thread?

Thanks :)

[dillee1]

Both hdot264 and vss h.264 seems dead now....:(

Anyway where is the sourcecode for hdot264? SF.net projects site is essentially empty.

[lilhobo]

maybe they got bought out :D

remember the Hollywood plus board???? havent we come a long way with mpeg2 encoding?

[Tommy Carrot]

Originally posted by superdump
Thanks for the info Tommy. I don't suppose you're any good with MingW are you? If so, could you have a look at my compiling FFMPEG for Windows thread?

Thanks :)

Sorry, i have no MingW experience. I'm just following the mailing lists. :)

[shlezman]

The closest thing to H264 is the VSS project which is completely based on the JM-Something, and as such will never get far.

The ffmpeg team has some h264 code which looks like a decoder, probably with very basic functionality (not even full baseline implementation), but as Tommy wrote, they are working on it and with the experience of this team, they can get it up and running hopefully soon.

As for a commercial application, all I see is press releases, and they dont compile on my VC6 :confused:

[spyder]

There was a lot of code recently posted on the hdot264 ML. It is not complete I think but it seems to be a good base. Just thought I'd let ya know ;)

[mf]

On the ffmpeg stuff; afaik ffmpeg was making a QuickTime decoder, when they found out QT is actually some weird implementation of h.264. That might be the base of what they have now, but as my information is very vague it comes without any guarantee :D.

[BoNz1]

Originally posted by mf
On the ffmpeg stuff; afaik ffmpeg was making a QuickTime decoder, when they found out QT is actually some weird implementation of h.264. That might be the base of what they have now, but as my information is very vague it comes without any guarantee :D.

I believe an anonymous developer committed the code to decode sorenson 3.5 video. And then they discovered that it had some significant similarities to h.264. But, I'm fairly certain though they are not using parts of the sorenson decoder to decode h.264, I believe they are writing their own encoder and decoder from the ground up. This is just speculation though since I've not heard this from any of the developers. As for building ffmpeg for windows, it seems the windows build is broken ATM, I can't make it build on MingW or Cygwin. However, it is very possible that I am building it incorrectly, but other people seem to be having the same problems...