Hi,

On Sat, Jan 7, 2012 at 11:56 AM, Mike Melanson <[email protected]> wrote:
> On 1/7/2012 11:50 AM, Ronald S. Bultje wrote:
>> On Wed, Dec 28, 2011 at 4:44 AM, Ungueranu Vlad Victor
>> <[email protected]>  wrote:
>>>
>>> The samples were in the fate suite, but no tests were made.
>>> kgv1.avi
>>
>> This test still gives valgrind errors, the file is broken. Another
>> person submitted the same test already earlier (Oana).
>
> There's another issue as well-- KGV1 outputs to a 16-bit RGB colorspace. I
> have the worst time getting the framecrc output to match on little- vs.
> big-endian systems. In fact, I ran into this again as I was trying to put
> together another set of FATE tests.
>
> Before a test goes in for KGV1, we need to validate it on PPC and x86, at
> least.

Agreed.

The file can be fixed by fixing the AVI stream header, BTW, I have
that fixed locally, just not rsynced up tot he server yet.

As for the 15/16bpp, it'll output to native endianness so you need to
force to -pix_fmt rgb555le to have the same output on BE machines. I
believe the conversions in swscale for this are lossless but may be
wrong. If it's not, poke me and I can fix that.

Ronald
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to