[FFmpeg-devel] [PATCH 4/4] proresenc_kostya: properly account for alpha

[Christophe Gisquet]

Hi,

2014-08-12 1:56 GMT+02:00 Michael Niedermayer <michaelni at gmx.at>:
>> -                                  (2 + 2 * ctx->num_planes +
>> +                                  (2 + 2 * FFMIN(3, ctx->num_planes) +
>
> slice_hdr_size is  2 + 2 * (ctx->num_planes - 1);
>
> is this difference intended ?

I don't know why there was a difference between the original code and
slice_hdr_size. I don't know either why the +200 too (I expect it to
be an upper bound of the other stuff, like quant matrix and other
header info).

In any case, we may err on the safer side and use that.

Anyway, the hunk indeed is wrong as I misinterpreted the code.

> also ctx->num_planes is currently either 3 or 4 which would make the
> FFMIN a constant

Yeah, I'll drop it.

-- 
Christophe