[FFmpeg-cvslog] r24206 - trunk/libavcodec/vc1dec.c
David Conrad
lessen42
Mon Jul 12 17:11:33 CEST 2010
On Jul 12, 2010, at 9:23 AM, Benoit Fouet wrote:
> Hi,
>
> On Mon, 12 Jul 2010 00:52:51 +0200 (CEST) conrad wrote:
>> Author: conrad
>> Date: Mon Jul 12 00:52:51 2010
>> New Revision: 24206
>>
>> Log:
>> vc1: Fix ordering of loop filter for I/B frames
>>
>> All horizontal edges must be filtered before all vertical edges
>>
>
> Seems this changed the result of vc1 fate test:
> --- /home/hack/ffmpeg/x86/tests/ref/fate/vc1 2010-06-21 11:32:10.000000000 +0200
> +++ tests/data/fate/vc1 2010-07-12 15:22:04.000000000 +0200
> @@ -1,15 +1,15 @@
> -0, 0, 38016, 0xf1b25db5
> -0, 3600, 38016, 0xf1b25db5
> -0, 7200, 38016, 0xf1b25db5
> -0, 10800, 38016, 0x9ed7f0de
> -0, 14400, 38016, 0x5a0b2cf6
> -0, 18000, 38016, 0x5e3ed87b
> -0, 21600, 38016, 0xe2e3432d
> -0, 25200, 38016, 0xe2e3432d
> -0, 28800, 38016, 0x76854773
> -0, 32400, 38016, 0x76854773
> -0, 36000, 38016, 0xefbb2133
> -0, 39600, 38016, 0xefbb2133
> -0, 43200, 38016, 0xefbb2133
> -0, 46800, 38016, 0xefbb2133
> -0, 50400, 38016, 0xefbb2133
> +0, 0, 38016, 0xf4715db5
> +0, 3600, 38016, 0xf4715db5
> +0, 7200, 38016, 0xf4715db5
> +0, 10800, 38016, 0xf46af0e1
> +0, 14400, 38016, 0x96992cf1
> +0, 18000, 38016, 0xbaadd874
> +0, 21600, 38016, 0x751f4328
> +0, 25200, 38016, 0x751f4328
> +0, 28800, 38016, 0xf7294772
> +0, 32400, 38016, 0xf7294772
> +0, 36000, 38016, 0xf1d12133
> +0, 39600, 38016, 0xf1d12133
> +0, 43200, 38016, 0xf1d12133
> +0, 46800, 38016, 0xf1d12133
> +0, 50400, 38016, 0xf1d12133
> make: *** [fate-vc1] Error 1
>
> Am I the only one experiencing that?
Yes, this is expected and fate must be updated.
It's still not bitexact with the reference but it's a little closer.
More information about the ffmpeg-cvslog
mailing list