On Wed, Mar 26, 2014 at 02:15:00PM -0700, Jason Garrett-Glaser wrote:
> + /* preserve the golden frame */
> + if (s->framep[VP56_FRAME_GOLDEN] == prev_frame) {
> + s->framep[VP56_FRAME_GOLDEN] = vp8_find_free_buffer(s);
> + if ((ret = vp8_alloc_frame(s,
> s->framep[VP56_FRAME_GOLDEN], 1)) < 0)
> + return ret;
> + copy_frame(s->framep[VP56_
> FRAME_GOLDEN]->tf.f,
> + prev_frame->tf.f, s->mb_width * 16,
> + s->mb_height * 16);
> + }
> +
> + for (j = 0; j < s->mb_height * 16; j++)
> + for (i = 0; i < s->mb_width * 16; i++) {
> + uint8_t y = dst[j*linesize + i];
> + dst[j*linesize + i] = av_clip_uint8(y + ((y *
> beta) >> 8) + alpha);
> + }
>
> Wouldn't it make more sense to combine the copy and weight into a
> single operation? This hits the same memory twice.
>
> It also seems more natural for this to be split out into a DSP
> function of some sort.
This was fixed >month ago.
Vittorio which of my FFmpeg patches did you base this off??
See http://lists.ffmpeg.org/pipermail/ffmpeg-devel/2014-February/154626.html
>> Other changes in PATCHv3:
>> - combine fade operation with gold frame preservation
>> - optimise inter_predict_dc usage
>> - described vp7_calculate_mb_offset
>> - dont bother zeroing block/block_dc as done by idct now
>> Thanks to Ronald and Jason for their thorough reviews. Cheers,
-- Peter
(A907 E02F A6E5 0CD2 34CD 20D2 6760 79C5 AC40 DD6B)
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
