Log in

View Full Version : Hey developers, what about the bandwith restriction?


leonhard
3rd April 2004, 07:39
Actually I am asking if I am to stupid or if there is really this bandwith restriction in XVID. If I am not simply to blind to see how to overcome this restriction, I at least don 't see the point not to allow to fill up a DVD for Masterquality. I am regulary working with 50fps progressiv in best as possible relution and sharpness, and thinking of HDTV to come ....

Perhaps I can get some understanding? Or may I give some motivation?

communist
3rd April 2004, 09:17
What bandwidth restriction? I have no clue what you are talking about :confused:

leonhard
3rd April 2004, 09:35
did you ever go over 10 Mbits per seconds ? Thats the limit. With other words, if you have a footage of 15 min for example, you are not able to fill a DVD.

sysKin
3rd April 2004, 09:55
Of course not, no dvd reader is ever going to read a dvd in 15 minutes.

There is no bandwidth restriction in XviD.

leonhard
3rd April 2004, 10:09
well, of course I am not using theese masters for every day usage on a standalone. It is for computer usage only. And I actually tried to fill up a DVD, using the codec version of last christmas. I explicitly set the the size to DVD size an got a bit more than half the space filled. Trying to specyfy a bitrate, at 10 Mbit was the edge. If you can do, please tell me the recipe.
Olaf

kilg0r3
3rd April 2004, 11:06
First of all, I don't know if I really understood what you ment to say. But anyway.

If you just want to increase filesize and quality, experimenting with custom matrices like the 6of9 or semi-insane could be your way. Although this might intorduce playback problems when using FFDShow. For these matrices are not spec compliant.

If your problem is that your dvd-rom or burner does not play dvds with such a high bitrate, it might be due to their transferrate limit.

sysKin
3rd April 2004, 12:02
Originally posted by kilg0r3
For these matrices are not spec compliant.They are :)

Anyway, leonhard, I think you're looking for a lossless codec.

RadicalEd
3rd April 2004, 12:57
The best you could get would be constant quant 1, an mpeg intra matrix of all 8's, and no motion search. That'll give you essentially lossless compression, save for the slight roundoff error of storing the DCT coefficients as integers. Or, for even more size inflation (beyond uncompressed), as well as some really cool artifacts, use a matrix of all 1's.

kilg0r3
3rd April 2004, 13:08
Originally posted by sysKin
They are :)Anyway, leonhard, I think you're looking for a lossless codec.
I thought there was reason why ffdshow had/s these probs witht 6of9. [?]

Soulhunter
3rd April 2004, 19:59
Originally posted by kilg0r3
I thought there was reason why ffdshow had/s these probs witht 6of9. [?] AFAIK, This happens with all matrices their inter-matrix doesn't start with a 16, but its ffdshow's fault... ;)


Bye

leonhard
3rd April 2004, 20:25
well I so far don't have a big knowledge about matrixes, and as far as all the other things like Qpel and so on are concerned. I was try to use up for 4.2 Gbit for film of 50 fps, 720x576, 15min and I got only half the space used. Doing it for the second time, I got the same. As I was told that the bandwith isn't constrained in XVID, there must be another spot to find an error. This Encoding was totally independed of dvd, except, after all, it should fit on a DVD.

But you are right, I was looking to something nearly lossless, and I was thinking to replace the usage of lossless codecs, because i liked this bitrate calculting feature.

As I said it was Koepis Christmas build. Perhaps there is not a constrain in XVID but a malfunctioning bitrate calculator? May I overead this somewhere? As I said, may be some other unkown thing just somewhere .... . If somebody wants perhaps to run a try, I would be interesting in the result. For longer Videos prepared for DVD size, erverthing always went fine for me, for shorter one for CD Size as well. So perhaps somebody already has some experience?

Olaf

leonhard
3rd April 2004, 20:27
Of course using VBR, 2-pass

kilg0r3
3rd April 2004, 21:15
Well for a first attempt you could

1 install the latest koepi build
2 call up encoder configuration
3 hit the upper most 'more'- Button in the Profiles line
4 select MPEG-Custom in the Quant Type Menu
5 Hit 'Edit Matrix' button
6 Enter an 8 in each box
7 Select 'unrestricted' in the Profiles Menu
8 I usaully leave Qpel and GMC swithced off
9 Report back .. :)

BTW If you want to use Bframes then go Max 2, QuRatio:1, QuOffset:1. Don _not_ use Bframes at the same quant as your P-Frames.

Cheers kilg0r3

PS: You may _edit_ your posts instead of reposting

Soulhunter
3rd April 2004, 22:29
Originally posted by kilg0r3
Don _not_ use Bframes at the same quant as your P-Frames.Why... :confused:

kilg0r3
4th April 2004, 00:40
Because, at least in my experience, they have been larger than the respective P frames.

Soulhunter
4th April 2004, 01:12
Originally posted by kilg0r3
Because, at least in my experience, they have been larger than the respective P frames. AFAIK they should be smaller... :confused:


Example with X-MAN 2 (DVD/R2)
____________________________________________________________________

XviD: Version 1 RC2
Fixed Quantizer 2 | (VHQ1)
I-frame interval = 250
Matrix = Soulhunters (v2)

Size = 4.42GB
____________________________________________________________________

XviD: Version 1 RC2
Fixed Quantizer 2 | (VHQ1)
I-frame interval = 250
BVOP's = 1 / 1 / 0
Matrix = Soulhunters (v2)

Size = 4.21GB
____________________________________________________________________

Bye

Joe Fenton
4th April 2004, 02:50
Originally posted by leonhard
I was try to use up for 4.2 Gbit for film of 50 fps, 720x576, 15min and I got only half the space used.

720x576 at 50 FPS for 15 min gives a raw YV12 size of 28 GBytes. Fitting this exactly on a DVD means you want to do a compression of no more than 6 to 1. Any more than that and it will not fill the DVD. Get rid of any option that improves compression. No B-Frames, limit your coefficients, don't do motion compensation, etc.