[MPlayer-dev-eng] [PATCH] Fix H.264 SPS parsing in case of scaling list present

[Loren Merritt]

On Wed, 13 May 2009, Marco Munderloh wrote:

>> Then what does the patch fix? If you don't parse the lists, then you
>> only fix the case where the flag for "code the flags for individual
>> lists" is true, but all of the individual flags are false.
>
> The problem is, that, in case of a scaling list present, the current
> version forgets to skip the 8 individual bits so that all remaining
> flags in the sps are parsed wrong by 8 bits resulting in, for example,
> the fps parameter to be wrong so it has to be manually added with -fps
> to play the file.

In the case of a scaling list present, the current version forgets to 
check the 8 individual bits to see how many scaling lists are present, 
and skip their respective sizes (which are variable, so the lists 
themselves also have to be parsed).

Here's some test files for you:
http://akuvian.org/stuff/spscqm0.264 has no cqm, which of course works.
http://akuvian.org/stuff/spscqm1.264 has cqm flags present but all 0. Your 
patch fixes this.
http://akuvian.org/stuff/spscqm2.264 has some cqms. This is still borked.

--Loren Merritt