[FFmpeg-devel] [PATCH 4/6] avfilter/vf_nlmeans: avoid if () to help paralellization
Paul B Mahol
onemda at gmail.com
Sun Oct 24 23:25:00 EEST 2021
Signed-off-by: Paul B Mahol <onemda at gmail.com>
---
libavfilter/vf_nlmeans.c | 13 ++++++-------
1 file changed, 6 insertions(+), 7 deletions(-)
diff --git a/libavfilter/vf_nlmeans.c b/libavfilter/vf_nlmeans.c
index d5a71291af..af165c861c 100644
--- a/libavfilter/vf_nlmeans.c
+++ b/libavfilter/vf_nlmeans.c
@@ -332,6 +332,7 @@ struct thread_data {
static int nlmeans_slice(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs)
{
NLMeansContext *s = ctx->priv;
+ const uint32_t max_meaningful_diff = s->max_meaningful_diff;
const struct thread_data *td = arg;
const ptrdiff_t src_linesize = td->src_linesize;
const int process_h = td->endy - td->starty;
@@ -383,13 +384,11 @@ static int nlmeans_slice(AVFilterContext *ctx, void *arg, int jobnr, int nb_jobs
const uint32_t b = ii[x + dist_b];
const uint32_t d = ii[x + dist_d];
const uint32_t e = ii[x + dist_e];
- const uint32_t patch_diff_sq = e - d - b + a;
+ const uint32_t patch_diff_sq = FFMIN(e - d - b + a, max_meaningful_diff);
+ const float weight = weight_lut[patch_diff_sq]; // exp(-patch_diff_sq * s->pdiff_scale)
- if (patch_diff_sq < s->max_meaningful_diff) {
- const float weight = weight_lut[patch_diff_sq]; // exp(-patch_diff_sq * s->pdiff_scale)
- wa[x].total_weight += weight;
- wa[x].sum += weight * src[x];
- }
+ wa[x].total_weight += weight;
+ wa[x].sum += weight * src[x];
}
ii += s->ii_lz_32;
}
@@ -506,7 +505,7 @@ static av_cold int init(AVFilterContext *ctx)
s->pdiff_scale = 1. / (h * h);
s->max_meaningful_diff = log(255.) / s->pdiff_scale;
- s->weight_lut = av_calloc(s->max_meaningful_diff, sizeof(*s->weight_lut));
+ s->weight_lut = av_calloc(s->max_meaningful_diff + 1, sizeof(*s->weight_lut));
if (!s->weight_lut)
return AVERROR(ENOMEM);
for (int i = 0; i < s->max_meaningful_diff; i++)
--
2.33.0
More information about the ffmpeg-devel
mailing list