[FFmpeg-devel] [PATCH 3/7] fftools/ffmpeg_filter: auto-insert csp conversion filter

[Niklas Haas]

From: Niklas Haas <git at haasn.dev>

To convert between color ranges/matrices, if needed by the codec
properties. Swscale can't do this currently, so we fall back to using
zscale. We momentarily duplicate the mjpeg strictness logic to also
enfoce full range. This duplication will be cleaned up in the next
commit.

In theory this scaler can also handle chroma locations, but the
configuration for it is non-trivial due to the split between x/y
offsets, and no codec currently signals a chroma location requirement.
(Though this is subject to change in the future)
---
 fftools/ffmpeg_filter.c | 73 +++++++++++++++++++++++++++++++++++++++++
 1 file changed, 73 insertions(+)

diff --git a/fftools/ffmpeg_filter.c b/fftools/ffmpeg_filter.c
index c738fc3397..c1bedfda00 100644
--- a/fftools/ffmpeg_filter.c
+++ b/fftools/ffmpeg_filter.c
@@ -169,6 +169,8 @@ typedef struct OutputFilterPriv {
     int width, height;
     int sample_rate;
     AVChannelLayout ch_layout;
+    enum AVColorSpace csp;
+    enum AVColorRange color_range;
 
     // time base in which the output is sent to our downstream
     // does not need to match the filtersink's timebase
@@ -184,6 +186,8 @@ typedef struct OutputFilterPriv {
     const int *formats;
     const AVChannelLayout *ch_layouts;
     const int *sample_rates;
+    const enum AVColorSpace *csps;
+    const enum AVColorRange *color_ranges;
 
     AVRational enc_timebase;
     // offset for output timestamps, in AV_TIME_BASE_Q
@@ -368,6 +372,12 @@ DEF_CHOOSE_FORMAT(sample_fmts, enum AVSampleFormat, format, formats,
 DEF_CHOOSE_FORMAT(sample_rates, int, sample_rate, sample_rates, 0,
                   "%d", )
 
+DEF_CHOOSE_FORMAT(out_color_matrix, enum AVColorSpace, csp, csps,
+                  AVCOL_SPC_UNSPECIFIED, "%s", av_color_space_name);
+
+DEF_CHOOSE_FORMAT(out_range, enum AVColorRange, color_range, color_ranges,
+                  AVCOL_RANGE_UNSPECIFIED, "%s", av_color_range_name);
+
 static void choose_channel_layouts(OutputFilterPriv *ofp, AVBPrint *bprint)
 {
     if (av_channel_layout_check(&ofp->ch_layout)) {
@@ -594,6 +604,7 @@ static OutputFilter *ofilter_alloc(FilterGraph *fg)
     ofilter           = &ofp->ofilter;
     ofilter->graph    = fg;
     ofp->format       = -1;
+    ofp->csp          = AVCOL_SPC_UNSPECIFIED;
     ofilter->last_pts = AV_NOPTS_VALUE;
 
     return ofilter;
@@ -682,6 +693,37 @@ int ofilter_bind_ost(OutputFilter *ofilter, OutputStream *ost)
     case AVMEDIA_TYPE_VIDEO:
         ofp->width      = ost->enc_ctx->width;
         ofp->height     = ost->enc_ctx->height;
+        if (ost->enc_ctx->colorspace != AVCOL_SPC_UNSPECIFIED) {
+            ofp->csp = ost->enc_ctx->colorspace;
+        } else {
+            ofp->csps = c->csps;
+        }
+        if (ost->enc_ctx->color_range) {
+            ofp->color_range = ost->enc_ctx->color_range;
+        } else {
+            ofp->color_ranges = c->color_ranges;
+
+            // MJPEG encoder exports a full list of supported pixel formats,
+            // but the full-range ones are experimental-only.
+            // Restrict the auto-conversion list unless -strict experimental
+            // has been specified.
+            if (!strcmp(c->name, "mjpeg")) {
+                static const enum AVColorRange mjpeg_ranges[] =
+                    { AVCOL_RANGE_JPEG, AVCOL_RANGE_UNSPECIFIED };
+
+                const AVDictionaryEntry *strict = av_dict_get(ost->encoder_opts, "strict", NULL, 0);
+                int strict_val = ost->enc_ctx->strict_std_compliance;
+
+                if (strict) {
+                    const AVOption *o = av_opt_find(ost->enc_ctx, strict->key, NULL, 0, 0);
+                    av_assert0(o);
+                    av_opt_eval_int(ost->enc_ctx, o, strict->value, &strict_val);
+                }
+
+                if (strict_val > FF_COMPLIANCE_UNOFFICIAL)
+                    ofp->color_ranges = mjpeg_ranges;
+            }
+        }
         if (ost->enc_ctx->pix_fmt != AV_PIX_FMT_NONE) {
             ofp->format = ost->enc_ctx->pix_fmt;
         } else {
@@ -1149,6 +1191,7 @@ static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter,
     OutputFilterPriv *ofp = ofp_from_ofilter(ofilter);
     OutputStream *ost = ofilter->ost;
     OutputFile    *of = output_files[ost->file_index];
+    FilterGraphPriv *fgp = fgp_from_fg(fg);
     AVFilterContext *last_filter = out->filter_ctx;
     AVBPrint bprint;
     int pad_idx = out->pad_idx;
@@ -1209,6 +1252,36 @@ static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter,
         pad_idx     = 0;
     }
 
+    if (!fgp->disable_conversions) {
+        av_bprint_init(&bprint, 0, AV_BPRINT_SIZE_UNLIMITED);
+        choose_out_color_matrix(ofp,    &bprint);
+        choose_out_range(ofp,     &bprint);
+        if (bprint.len) {
+            const AVDictionaryEntry *e = NULL;
+            AVFilterContext *filter;
+
+            while ((e = av_dict_iterate(ost->sws_dict, e)))
+                av_bprintf(&bprint, "%s=%s:", e->key, e->value);
+
+            if (!av_bprint_is_complete(&bprint))
+                return AVERROR(ENOMEM);
+
+            snprintf(name, sizeof(name), "conv_out_%d_%d",
+                     ost->file_index, ost->index);
+            ret = avfilter_graph_create_filter(&filter,
+                                               avfilter_get_by_name("scale"),
+                                               name, bprint.str, NULL, fg->graph);
+            av_bprint_finalize(&bprint, NULL);
+            if (ret < 0)
+                return ret;
+            if ((ret = avfilter_link(last_filter, pad_idx, filter, 0)) < 0)
+                return ret;
+
+            last_filter = filter;
+            pad_idx = 0;
+        }
+    }
+
     snprintf(name, sizeof(name), "trim_out_%d_%d",
              ost->file_index, ost->index);
     ret = insert_trim(of->start_time, of->recording_time,
-- 
2.42.0