[XviD-devel] Has this bug been fixed yet?

[Edouard Gomez]

Edouard Gomez (ed.gomez at free.fr) wrote:
> Grayscale mode  works exactly like  complete color mode, except  that in
> the last encoding  stage (once we have MVs), we  just mark the chromatic
> blocks as being not coded. Now the part i'm not sure, is how MPEG4 treat
> this  non coded  blocks.   If they  are  still considered  as being  the
> difference coding,  then they are  just telling the decoder:  "hey, this
> block  is  the  same as  the  one  in  the  frame before"  (with  motion
> displacement applied). 
> 
> I think  that is  why your  first totally black  images have  still some
> remanent purple/dark  blue blocks, because that were  their color before
> the sudden black change.

Ok, i'm not sure of what i say so please tell me if i'm wrong:

In an IFrame, if we want only Luminance information we should initialize
the DC value to 128 (U/V_OFFSET for the zero value) not 0 ?

In a  PFrame like  it's predicted  frame, we should  just say  the block
ignored. Btw we set the DC value is actually set to 0, which is probably
safe.

In bframes, there is not even greyscale handling ?! 

If someone  is kind enough to  clarify things until i  go bed, otherwise
i'll have a  look at the standard (so boring)  and report my conclusions
tomorrow. 

--
Edouard Gomez