diff mbox series

[FFmpeg-devel,v2] libavcodec: add tune_content option also for VP8.

Message ID 20240207160340.357523-1-darekm@google.com
State New
Headers show
Series [FFmpeg-devel,v2] libavcodec: add tune_content option also for VP8. | expand

Checks

Context Check Description
yinshiyou/make_loongarch64 success Make finished
yinshiyou/make_fate_loongarch64 success Make fate finished
andriy/make_x86 success Make finished
andriy/make_fate_x86 success Make fate finished

Commit Message

Dariusz Marcinkiewicz Feb. 7, 2024, 4:03 p.m. UTC
This exposes VP8E_SET_SCREEN_CONTENT_MODE option from libvpx.

Changes since v1:
- Put the new param initialzation in the right place,
- Account for cases when the encoder's output is queued up.

Co-authored-by: Erik Språng <sprang@webrtc.org>
Signed-off-by: Dariusz Marcinkiewicz <darekm@google.com>
---
 doc/encoders.texi      |  7 ++++--
 libavcodec/libvpxenc.c | 56 ++++++++++++++++++++++++++++++++++++++----
 2 files changed, 56 insertions(+), 7 deletions(-)

Comments

James Zern Feb. 8, 2024, 6:13 a.m. UTC | #1
Hi,

On Wed, Feb 7, 2024 at 8:04 AM Dariusz Marcinkiewicz via ffmpeg-devel
<ffmpeg-devel@ffmpeg.org> wrote:
>
> This exposes VP8E_SET_SCREEN_CONTENT_MODE option from libvpx.
>

For the subject use '(libavcodec|avcodec|lavc)/libvpxenc: ...' to
better document what is being changed. See the history of the file for
examples.

> Changes since v1:
> - Put the new param initialzation in the right place,
> - Account for cases when the encoder's output is queued up.
>

This is better left as a comment in the thread. It will lose its
meaning when committed.

> Co-authored-by: Erik Språng <sprang@webrtc.org>
> Signed-off-by: Dariusz Marcinkiewicz <darekm@google.com>
> ---
>  doc/encoders.texi      |  7 ++++--
>  libavcodec/libvpxenc.c | 56 ++++++++++++++++++++++++++++++++++++++----

Please bump LIBAVCODEC_VERSION_MICRO in libavcodec/version.h.

>  2 files changed, 56 insertions(+), 7 deletions(-)
>
> diff --git a/doc/encoders.texi b/doc/encoders.texi
> index c9fe6d6143..2a9b38f62a 100644
> --- a/doc/encoders.texi
> +++ b/doc/encoders.texi
> @@ -2186,6 +2186,11 @@ Enable error resiliency features.
>  Increase sharpness at the expense of lower PSNR.
>  The valid range is [0, 7].
>
> +@item tune-content

Please match the vpxenc command line. They're separate, vp8 has
--screen-content-mode.
Dariusz Marcinkiewicz Feb. 8, 2024, 10 p.m. UTC | #2
Hi.

On Thu, Feb 8, 2024 at 7:13 AM James Zern <jzern@google.com> wrote:
>
> Hi,
>
> On Wed, Feb 7, 2024 at 8:04 AM Dariusz Marcinkiewicz via ffmpeg-devel
> <ffmpeg-devel@ffmpeg.org> wrote:
> >
> > This exposes VP8E_SET_SCREEN_CONTENT_MODE option from libvpx.
> >
>
> For the subject use '(libavcodec|avcodec|lavc)/libvpxenc: ...' to
> better document what is being changed. See the history of the file for
> examples.


Sent out v3 of the patch addressing this and remaining comments.
diff mbox series

Patch

diff --git a/doc/encoders.texi b/doc/encoders.texi
index c9fe6d6143..2a9b38f62a 100644
--- a/doc/encoders.texi
+++ b/doc/encoders.texi
@@ -2186,6 +2186,11 @@  Enable error resiliency features.
 Increase sharpness at the expense of lower PSNR.
 The valid range is [0, 7].
 
+@item tune-content
+Set content type.
+For VP8: default (0), screen (1), screen with aggressive rate control (2).
+For VP9: default (0), screen (1), film (2).
+
 @item ts-parameters
 Sets the temporal scalability configuration using a :-separated list of
 key=value pairs. For example, to specify temporal scalability parameters
@@ -2268,8 +2273,6 @@  colorspaces:
 @end table
 @item row-mt @var{boolean}
 Enable row based multi-threading.
-@item tune-content
-Set content type: default (0), screen (1), film (2).
 @item corpus-complexity
 Corpus VBR mode is a variant of standard VBR where the complexity distribution
 midpoint is passed in rather than calculated for a specific clip or chunk.
diff --git a/libavcodec/libvpxenc.c b/libavcodec/libvpxenc.c
index 80988a2608..c73c92d49b 100644
--- a/libavcodec/libvpxenc.c
+++ b/libavcodec/libvpxenc.c
@@ -164,6 +164,7 @@  static const char *const ctlidstr[] = {
     [VP8E_SET_MAX_INTRA_BITRATE_PCT] = "VP8E_SET_MAX_INTRA_BITRATE_PCT",
     [VP8E_SET_SHARPNESS]               = "VP8E_SET_SHARPNESS",
     [VP8E_SET_TEMPORAL_LAYER_ID]       = "VP8E_SET_TEMPORAL_LAYER_ID",
+    [VP8E_SET_SCREEN_CONTENT_MODE]     = "VP8E_SET_SCREEN_CONTENT_MODE",
 #if CONFIG_LIBVPX_VP9_ENCODER
     [VP9E_SET_LOSSLESS]                = "VP9E_SET_LOSSLESS",
     [VP9E_SET_TILE_COLUMNS]            = "VP9E_SET_TILE_COLUMNS",
@@ -1262,6 +1263,16 @@  static av_cold int vpx_init(AVCodecContext *avctx,
 #endif
     }
 #endif
+#ifdef VPX_CTRL_VP8E_SET_SCREEN_CONTENT_MODE
+    if (avctx->codec_id == AV_CODEC_ID_VP8 && ctx->tune_content >= 0) {
+      if (ctx->tune_content == 2 && ctx->is_alpha) {
+        av_log(avctx, AV_LOG_ERROR,
+               "Transparency encoding with screen mode with aggressive rate control not supported\n");
+        return AVERROR(EINVAL);
+      }
+      codecctl_int(avctx, VP8E_SET_SCREEN_CONTENT_MODE, ctx->tune_content);
+    }
+#endif
 
     av_log(avctx, AV_LOG_DEBUG, "Using deadline: %d\n", ctx->deadline);
 
@@ -1379,14 +1390,15 @@  static int storeframe(AVCodecContext *avctx, struct FrameListData *cx_frame,
  * @return AVERROR(ENOMEM) on coded frame queue data allocation error
  */
 static int queue_frames(AVCodecContext *avctx, struct vpx_codec_ctx *encoder,
-                        struct FrameListData **frame_list, AVPacket *pkt_out)
+                        struct FrameListData **frame_list, AVPacket *pkt_out,
+                        int queue_only, int *frame_enc)
 {
     VPxContext *ctx = avctx->priv_data;
     const struct vpx_codec_cx_pkt *pkt;
     const void *iter = NULL;
     int size = 0;
 
-    if (!ctx->is_alpha && *frame_list) {
+    if (!queue_only && *frame_list) {
         struct FrameListData *cx_frame = *frame_list;
         /* return the leading frame if we've already begun queueing */
         size = storeframe(avctx, cx_frame, NULL, pkt_out);
@@ -1401,7 +1413,7 @@  static int queue_frames(AVCodecContext *avctx, struct vpx_codec_ctx *encoder,
     while (pkt = vpx_codec_get_cx_data(encoder, &iter)) {
         switch (pkt->kind) {
         case VPX_CODEC_CX_FRAME_PKT:
-            if (!ctx->is_alpha && !size) {
+            if (!queue_only && !size) {
                 struct FrameListData cx_frame;
 
                 /* avoid storing the frame when the list is empty and we haven't yet
@@ -1411,6 +1423,8 @@  static int queue_frames(AVCodecContext *avctx, struct vpx_codec_ctx *encoder,
                 size = storeframe(avctx, &cx_frame, NULL, pkt_out);
                 if (size < 0)
                     return size;
+                if (size > 0)
+                    *frame_enc = 1;
             } else {
                 struct FrameListData *cx_frame = av_malloc(sizeof(*cx_frame));
 
@@ -1430,6 +1444,8 @@  static int queue_frames(AVCodecContext *avctx, struct vpx_codec_ctx *encoder,
                     return AVERROR(ENOMEM);
                 }
                 memcpy(cx_frame->buf, pkt->data.frame.buf, pkt->data.frame.sz);
+                if (pkt->data.frame.sz > 0)
+                  *frame_enc = 1;
                 coded_frame_add(frame_list, cx_frame);
             }
             break;
@@ -1693,6 +1709,7 @@  static int vpx_encode(AVCodecContext *avctx, AVPacket *pkt,
     vpx_svc_layer_id_t layer_id;
     int layer_id_valid = 0;
     unsigned long duration = 0;
+    int frame_enc = 0;
 
     if (avctx->qmax >= 0 && enccfg->rc_max_quantizer != avctx->qmax) {
         struct vpx_codec_enc_cfg cfg = *enccfg;
@@ -1856,9 +1873,32 @@  FF_ENABLE_DEPRECATION_WARNINGS
         }
     }
 
-    coded_size = queue_frames(avctx, &ctx->encoder, &ctx->coded_frame_list, pkt);
+    coded_size = queue_frames(avctx, &ctx->encoder, &ctx->coded_frame_list,
+                              pkt, ctx->is_alpha, &frame_enc);
+    if (avctx->codec_id == AV_CODEC_ID_VP8 && frame_enc == 0 &&
+        ctx->tune_content == 2 && frame) {
+        // VP8 tuned for screen content with aggresive rate control - returned
+        // OK status code but produced no output, this indicates frame was
+        // rolled back due to bitrate overshoot - try to encode it again.
+        av_log(avctx, AV_LOG_VERBOSE,
+               "Attempting to reencode dropped VP8 screencast frame.\n");
+        res = vpx_codec_encode(&ctx->encoder, rawimg, timestamp,
+                               duration, flags, ctx->deadline);
+        if (res != VPX_CODEC_OK) {
+            log_encoder_error(avctx, "Error encoding frame");
+            return AVERROR_INVALIDDATA;
+        }
+        if (!coded_size)
+          coded_size = queue_frames(avctx, &ctx->encoder, &ctx->coded_frame_list,
+                                    pkt, /*queue_only*/0, &frame_enc);
+        else
+          queue_frames(avctx, &ctx->encoder, &ctx->coded_frame_list, pkt,
+                       /*queue_only*/1, &frame_enc);
+    }
+
     if (ctx->is_alpha) {
-        queue_frames(avctx, &ctx->encoder_alpha, &ctx->alpha_coded_frame_list, NULL);
+        queue_frames(avctx, &ctx->encoder_alpha, &ctx->alpha_coded_frame_list,
+                     NULL, /*queue_only*/1, &frame_enc);
 
         if (ctx->coded_frame_list && ctx->alpha_coded_frame_list) {
             struct FrameListData *cx_frame = ctx->coded_frame_list;
@@ -1946,6 +1986,12 @@  static const AVOption vp8_options[] = {
     { "auto-alt-ref",    "Enable use of alternate reference "
                          "frames (2-pass only)",                        OFFSET(auto_alt_ref),    AV_OPT_TYPE_INT, {.i64 = -1}, -1,  2, VE},
     { "cpu-used",        "Quality/Speed ratio modifier",                OFFSET(cpu_used),        AV_OPT_TYPE_INT, {.i64 = 1}, -16, 16, VE},
+#ifdef VPX_CTRL_VP8E_SET_SCREEN_CONTENT_MODE
+    { "tune-content",     "Tune content type",                          OFFSET(tune_content),    AV_OPT_TYPE_INT, {.i64 = -1}, -1,  2, VE, "tune_content"},
+    { "default",          "Regular video content",                      0,                       AV_OPT_TYPE_CONST, {.i64 = 0}, 0,  0, VE, "tune_content"},
+    { "screen",           "Screen content mode on",                     0,                       AV_OPT_TYPE_CONST, {.i64 = 1}, 0,  0, VE, "tune_content"},
+    { "screen-rate-control", "Screen content mode with aggressive rate control", 0,              AV_OPT_TYPE_CONST, {.i64 = 2}, 0,  0, VE, "tune_content"},
+#endif
     LEGACY_OPTIONS
     { NULL }
 };