Snow, Dirac, Tarkin, and the future of video codecs

[sysKin]

XviD doesn't even scratch the surface of MPEG-4 P2. Object-based/arbitrary shape coding, one of the flagships of MPEG-4, hasn't been implemented anywhere yet.

Well, because they are useless to what we're doing. In theory you could cut an AMV using mp4 rather than video editing tools, but find me a person crazy enough to do that, even if there is software.
You can do MUCH more than all this using macromedia's flash, but you don't see it replacing "normal square" video compression.
XviD is a video codec, all the sprite/mesh/shape thingies are NOT related to video as such, they are rather editing tools.

In short, don't write off XviD just yet. AVC may be better for the kind of thing we're doing, but it's really a whole different animal.

I'm rather "writing off" future XviD development, not usage. XviD is the best video codec of the generation and nothing will change that, let it live forever.

[Tab.]

XviD is a video codec, all the sprite/mesh/shape thingies are NOT related to video as such, they are rather editing tools.

Well, if nothing else, from the video coding point of view I would expect them to offer some powerful ways to manipulate quality among more or less significant parts of a scene. I guess the benefits aren't worth the costs of implementation, though. Wasn't Gruel talking about implementing mesh-based ME awhile ago?

The comment about writing off XviD was more directed at z3r0, of course

About MP4, it's an unfortunate and probably unnecessarily complicated situation. So far the only decently working splitter that supports AVC that I've come across is Nero's, and that's not free. For the rest of MPEG-4, 3ivX's will probably remain king, unless someone ports lavf like I've been hoping.
I wouldn't highly advise switching to AVC just yet, if only for the fact that I can get just as decent results with XviD and ffdshow postprocessing. In the future, when there's a definite splitter and decoder to link to, then we'll be able to successfully phase it in to the masses. Until then, it's not worth the hassle IMO.

As far as decoding anything goes, I can't stress the greatness of ffdshow enough. It's the most useful thing for a noob because they need install only that, and they're set for decoding just about anything they'll be downloading, ever. It's the most useful thing for a competent user, because of all the options. IMO mplayer postprocessing is the best there is. In addition to those, it's fast.

In fact, I think it may have been Tab.

For the record, though I agree wholeheartedly with that statement, I can't lay claim to it.

I think it might make sense for a mod to split this thread into the AVC one from z3r0's initial comment on :O

[Zero1]

I found this poll on www.xvid.org a while ago, I thought the results were very interesting.

Code: Select all

higher speed                    9.31 % (1141)
better quality                 18.63 % (2284)
more advanced features          1.22 % (150)
full encoding application       5.08 % (623)
own audio codec                 2.59 % (317)
MPEG-4 AVC (H.264)             20.34 % (2494)
MPEG-4 main profile             2.57 % (315)
proprietary extensions          0.21 % (26)
wavelet compression            10.27 % (1259)
streaming support               1.72 % (211)
parallel encoding              15.24 % (1868)
nude pix of the developers     12.82 % (1571)

Notice how a similar number of people who wanted better quality also voted for H264, to some extent voting for H264 is sort of like voting for better quality and vice versa. I wonder what percentage of those who voted better quality didn't know what H264 is and/or what it would do for them.

So I was wondering...

Is it possible to add H264 features such as those seen in the vfw version of x264 to the v1.1 of XviD, or would XviD require an entire re-write?

Since XviD in it's current state is still good competiton for H264 codecs, it just goes to show how tweaked it is, so if it's possible to add H264 features to XviD it would make sense (to me at least) to carry on the development of XviD as a H264 codec. You could probably have a switch in the options to enable and disable the H264 features (and probably have the switch auto assign a new fourcc code as opposed to XVID).

Or even, what about the possibility of partnering with the x264 dev guys?

Perhaps I'm barking up the wrong tree, but XviD with x264 (obviously not all the features, I understand some are not possible with AVI) features sure would be awesome.

As I mentioned before, moving to a new container and 2 new codecs at the same time will be daunting for most people with the software in its current states.

A H264 XviD would be an awesome way of transitioning eventually to full H264 (in mp4).

Both of us decided it's pointless to continue - so unless you'll find some new developers, or new interesting ideas for me to explore - yeah, development will stop.

Excuse how this sounds, but have you just had enough of programming XviD, or is integrating H264 into XviD worth looking into?

Lets say XviD devolopment completely stops, are there any other projects (not limited to video) you wish to pursue, or when XviD development stops, so does everything?



Well compliments on the 1.1 beta, it's definately impressive, and I wish you all the best in the future.

[Tab.]

z3r0, you'll like this post (incidentally, made by sysKin) from a year ago:
http://forum.doom9.org/showthread.php?s ... post430696

[sysKin]

z3r0, you'll like this post (incidentally, made by sysKin) from a year ago:
http://forum.doom9.org/showthread.php?s ... post430696

Ah no, I seriously changed my mind since then. Especially since we already have x264.

It is not possible to "port" h264 features into an mpeg-4 encoder. They are completely different at too many levels. Mpeg 1/2/4(asp) were mostly the same thing, but AVC is *not*.

As for partnering wirh x264 guys, I personally decoded to BECOME one of x264 guys. Easier & simpler