Bug intercale and file size

[GreatLord]

Hi
I am using the lates Koepi build of xvid (XviD-05082002-1).
The file size I did specfiy was 117MB
With the interlace off in xvid the file was 266MB (I did specfiy max at 117MB file size)

With the interlace on in xvid the file was 234MB (I did specfiy max at 117MB file size)

The interlace on in xvid encoder gave only block picture,
On x-card (hardware mpeg4 decoder) or ffshow. Is that a bug sor should the xvid intralece on produes only block picture so you can not watch it ??.

[cult]

imho interlaced should be used only when capturing.
Also ffdshow isnt able to play xvid when this option is enabled

[GreatLord]

Okey
Why can I not play it with the x-card (Sigma Designs X-card) it supoport interlace mpeg1/2/4 movies ??
Sigma Designs mp4 encoder does have interlace coding and it working with the x-card why does not Xvid interlace works ???

[-h]

Okey
Why can I not play it with the x-card (Sigma Designs X-card) it supoport interlace mpeg1/2/4 movies ??
Sigma Designs mp4 encoder does have interlace coding and it working with the x-card why does not Xvid interlace works ???

I wrote the interlacing code for XviD, so I can only assume that either:

- XviD's interlacing support is broken (it's possible I guess)
- Sigma aren't using "real" MPEG-4 interlacing, but a filter like DivX5, thus they don't really support MPEG-4 interlacing (same as ffdshow)
- Sigma supports MPEG-4 interlacing, however *their* implementation has a bug

I honestly don't know which one it will be. I'll have to test it next week. I'd hate to think there's a bug in the current interlacing code, however if there is, it would be a small matter to create a converter to "correct" the bitstream losslessly (and of course fix the bug itself).

-h

[GreatLord]

I have talk to Sigma Designs about this problem.
The have replay with follow answer. About X-Card and Xvid interlacing.

"Xvid encoder has a problem. It's not broken but they are not using standard interlacing approach. They need to fix it.
Our RMP4 encoder and the Xcard drivers are not buggy."

[-h]

"Xvid encoder has a problem. It's not broken but they are not using standard interlacing approach. They need to fix it.
Our RMP4 encoder and the Xcard drivers are not buggy."

Well that's interesting. Doesn't help much with fixing the problem though.. all XviD does for interlacing is decide whether to use the field-based DCT or not, there's none of the more complex field-based motion estimation (yet).

I will look into this, Monday at the latest. I would hate to think that everyone who's used the interlacing feature has been generating invalid streams.

-h

[GreatLord]

The intlace frames from xvid on the x-card
some are complet blocky, some are mix betwin 1/3 okey with blocks. the first frame display 100% correct on the x-card.


if you need more detail data how the picture lock I can take some photo ??


if you need any help to test the interlacing on the x-card I am willing to help you with that.

If you need some more info plz let me know or e-mail at magnusolsen@chello.se

I am really like xvid.

[-h]

The intlace frames from xvid on the x-card
some are complet blocky, some are mix betwin 1/3 okey with blocks. the first frame display 100% correct on the x-card.

That is *very* interesting - the first frame (which would be an I-frame) uses the same interlacing feature as the frames that would follow it.

if you need more detail data how the picture lock I can take some photo ?? if you need any help to test the interlacing on the x-card I am willing to help you with that.

Nope that's OK, I will test XviD against the ISO reference MPEG-4 implementation - if XviD interlacing works fine with the ISO code, then XviD is working properly (because we aim to be ISO-compliant).

-h