On 1/30/12, Ronald S. Bultje <[email protected]> wrote:
> Hi Paul,
>
> On Mon, Jan 30, 2012 at 8:12 AM, Paul B Mahol <[email protected]> wrote:
>> On 1/30/12, Ronald S. Bultje <[email protected]> wrote:
>>> Hi,
>>>
>>> On Mon, Jan 30, 2012 at 5:58 AM, Paul B Mahol <[email protected]> wrote:
>>>>
>>>> Signed-off-by: Paul B Mahol <[email protected]>
>>>> ---
>>>> libswscale/yuv2rgb.c | 4 ++--
>>>> 1 files changed, 2 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/libswscale/yuv2rgb.c b/libswscale/yuv2rgb.c
>>>> index 39c8b9c..1927cd5 100644
>>>> --- a/libswscale/yuv2rgb.c
>>>> +++ b/libswscale/yuv2rgb.c
>>>> @@ -529,9 +529,9 @@ SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c)
>>>> case PIX_FMT_RGB48BE:
>>>> case PIX_FMT_RGB48LE: return yuv2rgb_c_48;
>>>> case PIX_FMT_ARGB:
>>>> - case PIX_FMT_ABGR: if (CONFIG_SWSCALE_ALPHA && c->srcFormat
>>>> ==
>>>> PIX_FMT_YUVA420P) return yuva2argb_c;
>>>> + case PIX_FMT_ABGR: if (CONFIG_SWSCALE_ALPHA &&
>>>> isALPHA(c->srcFormat)) return yuva2argb_c;
>>>> case PIX_FMT_RGBA:
>>>> - case PIX_FMT_BGRA: return (CONFIG_SWSCALE_ALPHA &&
>>>> c->srcFormat
>>>> == PIX_FMT_YUVA420P) ? yuva2rgba_c : yuv2rgb_c_32;
>>>> + case PIX_FMT_BGRA: return (CONFIG_SWSCALE_ALPHA &&
>>>> isALPHA(c->srcFormat)) ? yuva2rgba_c : yuv2rgb_c_32;
>>>> case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
>>>> case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
>>>> case PIX_FMT_RGB565:
>>>
>>> That doesn't do anything, see this code in swscale_unscaled.c:
How that function have anything to do with above code?
Above code deal with unaccelerated colorspace conversion.
>>>
>>> /* yuv2bgr */
>>> if ((srcFormat == PIX_FMT_YUV420P || srcFormat == PIX_FMT_YUV422P ||
>>> srcFormat == PIX_FMT_YUVA420P) && isAnyRGB(dstFormat) &&
>>> !(flags & SWS_ACCURATE_RND) && !(dstH & 1)) {
>>> c->swScale = ff_yuv2rgb_get_func_ptr(c);
>>> }
>>
>> You lost me here. How can i see that something is really wrong?
>>
>> Command to reproduce issue you found would be nice.
>
> What I mean is that your patch won't make any difference. The
> isALPHA() check is intended to make the check more generic and have
> the condition be true for formats other than YUVA420P also, e.g. for
> YUVA444P. However, here, that won't be true, since the caller function
> ensures that srcFormat is one of YUV420P, YUVA420P or YUV422P, and
> thus the only case where the srcFormat has an alpha channel is where
> srcFormat == YUVA420P.
>
> In other words, the patch is unnecessary.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
