[FFmpeg-devel] [PATCH] vp9: ignore reference segmentation map if error_resilience flag is set.

[Hendrik Leppkes]

On Wed, Aug 13, 2014 at 12:11 AM, Ronald S. Bultje <rsbultje at gmail.com> wrote:
> Fixes ffvp9_fails_where_libvpx.succeeds.webm from ticket 3849.
> ---
>  libavcodec/vp9.c | 26 +++++++++++++++-----------
>  1 file changed, 15 insertions(+), 11 deletions(-)
>
> diff --git a/libavcodec/vp9.c b/libavcodec/vp9.c
> index 56975bd..186aec1 100644
> --- a/libavcodec/vp9.c
> +++ b/libavcodec/vp9.c
> @@ -278,7 +278,7 @@ static int vp9_alloc_frame(AVCodecContext *ctx, VP9Frame *f)
>
>      // retain segmentation map if it doesn't update
>      if (s->segmentation.enabled && !s->segmentation.update_map &&
> -        !s->intraonly && !s->keyframe) {
> +        !s->intraonly && !s->keyframe && !s->errorres) {
>          memcpy(f->segmentation_map, s->frames[LAST_FRAME].segmentation_map, sz);
>      }
>
> @@ -1344,16 +1344,20 @@ static void decode_mode(AVCodecContext *ctx)
>                  vp56_rac_get_prob_branchy(&s->c,
>                      s->prob.segpred[s->above_segpred_ctx[col] +
>                                      s->left_segpred_ctx[row7]]))) {
> -        int pred = 8, x;
> -        uint8_t *refsegmap = s->frames[LAST_FRAME].segmentation_map;
> -
> -        if (!s->last_uses_2pass)
> -            ff_thread_await_progress(&s->frames[LAST_FRAME].tf, row >> 3, 0);
> -        for (y = 0; y < h4; y++)
> -            for (x = 0; x < w4; x++)
> -                pred = FFMIN(pred, refsegmap[(y + row) * 8 * s->sb_cols + x + col]);
> -        av_assert1(pred < 8);
> -        b->seg_id = pred;
> +        if (!s->errorres) {
> +            int pred = 8, x;
> +            uint8_t *refsegmap = s->frames[LAST_FRAME].segmentation_map;
> +
> +            if (!s->last_uses_2pass)
> +                ff_thread_await_progress(&s->frames[LAST_FRAME].tf, row >> 3, 0);
> +            for (y = 0; y < h4; y++)
> +                for (x = 0; x < w4; x++)
> +                    pred = FFMIN(pred, refsegmap[(y + row) * 8 * s->sb_cols + x + col]);
> +            av_assert1(pred < 8);
> +            b->seg_id = pred;
> +        } else {
> +            b->seg_id = 0;
> +        }
>
>          memset(&s->above_segpred_ctx[col], 1, w4);
>          memset(&s->left_segpred_ctx[row7], 1, h4);
> --
> 1.8.5.5
>

Might it be useful to add some of those special samples to fate?

- Hendrik