HD-DVD Will Win the Format War!

[RobertR1]

And of the H.264 vs MPEG-2 things, I doubt that Sony decided to use MPEG-2 just to avoid the license fee. Remember that MPEG-2 also requires license fee and it's not cheaper than H.264. I think it's just the reasons they mentioned about: MPEG-2 encoders are more mature and more efficient than H.264 encoders.

Mpeg2 is used by Sony because Sony are the majority patent holder in the Mpeg2 licensing scheme. It's also the reason that it was mainly pushed for by Sony in the BR camp and actually the ONLY codec in the spec before the studios complained and h.264(avc) and VC-1 were adopted.

VC-1 is much more advanced in it's early stages than Mpeg2 is at it's end of life. Mpeg2 cannot improve to create a transparent to the master pic at Hi Def, at VC-1 bitrates because it needs too high of a bandwidth/bit rate to produce the same picture. This is another reason why the BW specs on Blu Ray were higher as it was clear that high BW would be needed if they were to use Mpeg2 throughout. The extra space 50GB follows the same reasoning as using higher BW codec stream means more data on disc.

The higher bandwidth and disc space (neither possible currently to due to lack or 50GB ROM media) only became an advantage once the other two codecs were apoted. Before that, both were critical just to keep pace using Mpeg2.

 

[Geo]

Ahhh, I didn't realize that Sony had a long-term financial interest in mpeg-2.

That's bad. I was cuffing them around in my mind for not dealing with a software thing well in advance so it wouldn't be an issue. Thinking to myself, well maybe they didn't have a great deal of choice on the hardware end --these things happen-- but they should have dealt with the software end.

Now I have to wonder if they intend to switch at all based on knowing the internal bean counters will be wailing against it.

 

[sir doris]

I think there is also a political angle here. VC-1 is mature, MS have put a lot of work into it and it shows with the quality of the US HD-DVD transfers. By contrast AVC is not as advanced and this is demonstrated by the quality of the Jananese HD-DVD transfers (which are not as good as the US HD-DVDs on the same hardware) which are using AVC.

So Sony had a choice:
1) use MPEG2 which is mature and they have a lot of expertese in optomising.
2) use AVC which isn't mature, is new to the engineers (may take longer to master) and may not end up looking much better than MPEG2 (if at all?).
3) use VC-1 and loose face using a codec developed by one of it's major rivals.

 

[pcchen]

Actually you can try VC-1 encoders by yourself. Microsoft has already released a beta VC-1 encoder DMO which can be used on Windows XP (with Movie Maker 2 or directly through command line).

Compared to x264 and Nero Digital, I think the VC-1 encoder by MS is quite good, but it's not better than x264. Of course MS's VC-1 encoder is quite complete, supporting interlaced encoding for example.

I'm not very familiar with VC-1 as I have not seen its specification. But in my understanding VC-1 does not have as many features as H.264 does. Of course, quality wise it's all about the implmentations of encoders. I'd say that H.264 has the potential to be more effective than VC-1, but right now VC-1's encoder is slightly better than H.264 encoders.

I have not seen the results from hardware H.264 encoders. Sony just released two AVCHD camcorders which uses H.264 to encode HDTV video (15Mbps 1440x1080i). Maybe when the samples are released we can see how the hardware encoders in these camcorders perform.

 

[Geo]

http://www.msnbc.msn.com/id/14060550/

BRUSSELS - European Commission antitrust officials are probing the licensing strategies of two rival new generation DVD developers, HD DVD and Blu-ray Disc, the EU executive said on Thursday.

I wonder what that's about? I wouldn't mind that leading to public info on what the associated fees and such are. Or do we know that already?

 

[RobertR1]

No idea Geo. The patent licensing codec info can be found on:

http://www.mpegla.com/index1.cfm

But if this is a hardware licensing issue, that information will likely never be made public.

 

[Bobbler]

http://www.msnbc.msn.com/id/14060550/

I wonder what that's about? I wouldn't mind that leading to public info on what the associated fees and such are. Or do we know that already?

Is this typical? Neither product has launched in EU yet, so I would imagine the EU commission would want some information on the licensing beforehand in something that is potentially a rather largely licensed thing. It doesn't seem like they are raiding these guy's offices or anything, just requesting information to preemptively make sure it's all legit beforehand. Not sure why this is news worthy really, but it seems it's being reported all over.

 

[Moloch]

In encoding time, of course
Serious coders (although I have no idea how many DVD are encoded this way) tunes the encoding parameter for many important frames (there are MPEG-2 encoders which let you specify encoding parameters for each frame), so encoding speed is quite important for these coders. H.264 encoders are still much slower than MPEG-2 encoders.

Good, you're not insane afterall
I doubt encoding speed is all that important since they most likely have insane computers to encode with

 

[quest55720]

I think it will come down to if blue ray group can get thier players to mass market price with in 6 months of when HD-DVD players get there. I still have doubts about the quality of the PS3 blueray player. The makers of stand alone players would be pretty pissed if the PS3 had the same quality for 400-500 dollars less. I can see PS3 being used by gamers 18+ who are single as a blue ray player. I can not see a house of 4 really able to use it as a player with out it being a headace.

If the blue ray group can't get the cost of thier players under controll they will lose. I was pretty gun ho about waiting for a winner then buy. Now soon as someone gets a player under 250 USD that up scales as well as the current toshiba player I am in and will take my chances. I will not buy that many movies till players get near 150 USD that way I can have them on all my TVs in the house.

 

[pcchen]

Good, you're not insane afterall
I doubt encoding speed is all that important since they most likely have insane computers to encode with

Well, IMHO most coders don't have "insane computers" to encode

 

[Moloch]

Well, IMHO most coders don't have "insane computers" to encode

Have you heard what the studios use for encoding

 

[pcchen]

Have you heard what the studios use for encoding

Let's put it this way: basically there are two choices, using speical hardwares or using softwares. Special hardwares are expensive and they don't necessarily encode best because they are somewhat less flexible. Softwares are slow. Most encoding softwares (even the "professional" ones) don't support clusters, or only with very large granuity (i.e. you encode different "clips" on different nodes, rather than using all nodes to encode one clip). So all you can do is to use a multi-CPU box. Since most encoding softwares are now x86 based (or PPC based, but as Mac is now x86 based too...) the upper limit of the number of CPUs is, well, quite low

Of course this may change, especially when CELL based encoders are out (and of course, CELL itself).

 

[Moloch]

Let's put it this way: basically there are two choices, using speical hardwares or using softwares. Special hardwares are expensive and they don't necessarily encode best because they are somewhat less flexible. Softwares are slow. Most encoding softwares (even the "professional" ones) don't support clusters, or only with very large granuity (i.e. you encode different "clips" on different nodes, rather than using all nodes to encode one clip). So all you can do is to use a multi-CPU box. Since most encoding softwares are now x86 based (or PPC based, but as Mac is now x86 based too...) the upper limit of the number of CPUs is, well, quite low

Of course this may change, especially when CELL based encoders are out (and of course, CELL itself).

Well do most even support 4 way SMP?

 

[pcchen]

Well do most even support 4 way SMP?

To my understanding, most pro encoders do support unlimited number of CPUs, however I don't know the efficiency of more than 4 CPUs. In my experiences 2 CPUs almost always give you twice performance, ditto to 4 CPUs. Even dual core CPUs do give you twice performance in most cases. So, basically the best you can get is a 4-way dual-core SMP machine, which may give you 8x the performance of a single core desktop machine.

Note that current fast desktop computers (especially the dual core ones) can encode high quality SD MPEG-2 videos faster than real time. Good H.264 encoders are much slower (more than 10 times as slow), though.

 

[Moloch]

To my understanding, most pro encoders do support unlimited number of CPUs, however I don't know the efficiency of more than 4 CPUs. In my experiences 2 CPUs almost always give you twice performance, ditto to 4 CPUs. Even dual core CPUs do give you twice performance in most cases. So, basically the best you can get is a 4-way dual-core SMP machine, which may give you 8x the performance of a single core desktop machine.

Note that current fast desktop computers (especially the dual core ones) can encode high quality SD MPEG-2 videos faster than real time. Good H.264 encoders are much slower (more than 10 times as slow), though.

I'm not sure I like h.264 since it's harder to decode, and now you say it's much slower to encode.

 

[willardjuice]

I'm not sure I like h.264 since it's harder to decode, and now you say it's much slower to encode.

Anyone know how hard it is to encode/decode VC-1 vs mpeg2 and mpeg4?

 

[Moloch]

Anyone know how hard it is to encode/decode VC-1 vs mpeg2 and mpeg4?

It's supposedly harder than vc-1 IIRC.

 

[ERP]

Anyone know how hard it is to encode/decode VC-1 vs mpeg2 and mpeg4?

H264 is Mpeg4 technically (Part 10 of the spec), although what people usually refer to is Mpeg4 part 2, which is mostly the same as Mpeg 2 but with some restrictions relaxed.

The simple answer is it's a lot more expensive than any of the other options.

Where H264 extends Mpeg2/4 style coding to include things like references accross frames, variable sized macroblocks and an additional processing step on the DCT values. VC1 as I understand it, is quite different, but I'm basing that on comments I've heard about some of the properties of the codec since I've never seen a description of the algorithm.

 

[MistaPi]

Mpeg2 is used by Sony because Sony are the majority patent holder in the Mpeg2 licensing scheme. It's also the reason that it was mainly pushed for by Sony in the BR camp and actually the ONLY codec in the spec before the studios complained and h.264(avc) and VC-1 were adopted.

So for a layman's explanation, why are the studios using Mpeg2 with Blueray when they can use VC-1 and get better quailty and more materials on the disc?

 

[London Geezer]

So for a layman's explanation, why are the studios using Mpeg2 with Blueray when they can use VC-1 and get better quailty and more materials on the disc?

A lot of movies are already stored in HD resolutions in MPEG2, and that includes Sony movies as much as other studios. It is therefore cheaper to release these movies in MPEG2 as there is little to no re-encoding needed. Converting these movies to another format would take time and money, and obviously the bulk of initial releases are all about releasing early to get some cash pronto, not about quality at all.

Don't ask me why this is happening with Bluray and not HDDVD cause that will be a mystery for a while, execpt for the fact that the initial Bluray movies were Sony's movies, and we all know why Sony wants to push MPEG2 more than other studios.