avcodec/libaom: Add HDR10+ metadata support

Signed-off-by: Maryla Ustarroz-Calonge <maryla@google.com>
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

Changelog

@@ -39,6 +39,8 @@ version 8.0:
 - AV1 Vulkan encoder
 - ProRes RAW decoder
 - ProRes RAW Vulkan hwaccel
+- ffprobe -codec option
+- HDR10+ metadata passthrough when decoding/encoding with libaom-av1
 
 
 version 7.1:

libavcodec/libaomdec.c

@@ -28,11 +28,14 @@
 
 #include "libavutil/common.h"
 #include "libavutil/cpu.h"
+#include "libavutil/hdr_dynamic_metadata.h"
 #include "libavutil/imgutils.h"
 
 #include "avcodec.h"
+#include "bytestream.h"
 #include "codec_internal.h"
 #include "decode.h"
+#include "itut35.h"
 #include "libaom.h"
 #include "profiles.h"
 
@@ -137,6 +140,61 @@ static int set_pix_fmt(AVCodecContext *avctx, struct aom_image *img)
     }
 }
 
+static int decode_metadata_itu_t_t35(AVFrame *frame,
+                                     const uint8_t *buffer, size_t buffer_size)
+{
+    if (buffer_size < 6)
+        return AVERROR(EINVAL);
+
+    GetByteContext bc;
+    bytestream2_init(&bc, buffer, buffer_size);
+
+    const int country_code = bytestream2_get_byteu(&bc);
+    const int provider_code = bytestream2_get_be16u(&bc);
+    const int provider_oriented_code = bytestream2_get_be16u(&bc);
+    const int application_identifier = bytestream2_get_byteu(&bc);
+
+    // See "HDR10+ AV1 Metadata Handling Specification" v1.0.1, Section 2.1.
+    if (country_code == ITU_T_T35_COUNTRY_CODE_US
+        && provider_code == ITU_T_T35_PROVIDER_CODE_SAMSUNG
+        && provider_oriented_code == 0x0001
+        && application_identifier == 0x04) {
+        // HDR10+
+        AVDynamicHDRPlus *hdr_plus = av_dynamic_hdr_plus_create_side_data(frame);
+        if (!hdr_plus)
+            return AVERROR(ENOMEM);
+
+        int res = av_dynamic_hdr_plus_from_t35(hdr_plus, bc.buffer,
+                                               bytestream2_get_bytes_left(&bc));
+        if (res < 0)
+            return res;
+    }
+
+    return 0;
+}
+
+static int decode_metadata(AVFrame *frame, const struct aom_image *img)
+{
+    const size_t num_metadata = aom_img_num_metadata(img);
+    for (size_t i = 0; i < num_metadata; ++i) {
+        const aom_metadata_t *metadata = aom_img_get_metadata(img, i);
+        if (!metadata)
+            continue;
+
+        switch (metadata->type) {
+        case OBU_METADATA_TYPE_ITUT_T35: {
+            int res = decode_metadata_itu_t_t35(frame, metadata->payload, metadata->sz);
+            if (res < 0)
+                return res;
+            break;
+        }
+        default:
+            break;
+        }
+    }
+    return 0;
+}
+
 static int aom_decode(AVCodecContext *avctx, AVFrame *picture,
                       int *got_frame, AVPacket *avpkt)
 {
@@ -215,6 +273,11 @@ static int aom_decode(AVCodecContext *avctx, AVFrame *picture,
             av_image_copy(picture->data, picture->linesize, planes,
                           stride, avctx->pix_fmt, img->d_w, img->d_h);
         }
+        ret = decode_metadata(picture, img);
+        if (ret < 0) {
+            av_log(avctx, AV_LOG_ERROR, "Failed to decode metadata\n");
+            return ret;
+        }
         *got_frame = 1;
     }
     return avpkt->size;

libavcodec/libaomenc.c

@@ -33,6 +33,7 @@
 #include "libavutil/base64.h"
 #include "libavutil/common.h"
 #include "libavutil/cpu.h"
+#include "libavutil/hdr_dynamic_metadata.h"
 #include "libavutil/imgutils.h"
 #include "libavutil/mathematics.h"
 #include "libavutil/mem.h"
@@ -42,10 +43,12 @@
 #include "av1.h"
 #include "avcodec.h"
 #include "bsf.h"
+#include "bytestream.h"
 #include "codec_internal.h"
 #include "dovi_rpu.h"
 #include "encode.h"
 #include "internal.h"
+#include "itut35.h"
 #include "libaom.h"
 #include "packet_internal.h"
 #include "profiles.h"
@@ -328,6 +331,53 @@ static av_cold int codecctl_int(AVCodecContext *avctx,
     return 0;
 }
 
+static int add_hdr_plus(AVCodecContext *avctx, struct aom_image *img, const AVFrame *frame)
+{
+    // Check for HDR10+
+    AVFrameSideData *side_data =
+        av_frame_get_side_data(frame, AV_FRAME_DATA_DYNAMIC_HDR_PLUS);
+    if (!side_data)
+        return 0;
+
+    size_t payload_size;
+    AVDynamicHDRPlus *hdr_plus = (AVDynamicHDRPlus *)side_data->buf->data;
+    int res = av_dynamic_hdr_plus_to_t35(hdr_plus, NULL, &payload_size);
+    if (res < 0) {
+        log_encoder_error(avctx, "Error finding the size of HDR10+");
+        return res;
+    }
+
+    uint8_t *hdr_plus_buf;
+    // Extra bytes for the country code, provider code, provider oriented code and app id.
+    const size_t hdr_plus_buf_size = payload_size + 6;
+    hdr_plus_buf = av_malloc(hdr_plus_buf_size);
+    if (!hdr_plus_buf)
+        return AVERROR(ENOMEM);
+
+    uint8_t *payload = hdr_plus_buf;
+    // See "HDR10+ AV1 Metadata Handling Specification" v1.0.1, Section 2.1.
+    bytestream_put_byte(&payload, ITU_T_T35_COUNTRY_CODE_US);
+    bytestream_put_be16(&payload, ITU_T_T35_PROVIDER_CODE_SAMSUNG);
+    bytestream_put_be16(&payload, 0x0001); // provider_oriented_code
+    bytestream_put_byte(&payload, 0x04);   // application_identifier
+
+    res = av_dynamic_hdr_plus_to_t35(hdr_plus, &payload, &payload_size);
+    if (res < 0) {
+        av_free(hdr_plus_buf);
+        log_encoder_error(avctx, "Error encoding HDR10+ from side data");
+        return res;
+    }
+
+    res = aom_img_add_metadata(img, OBU_METADATA_TYPE_ITUT_T35,
+                               hdr_plus_buf, hdr_plus_buf_size, AOM_MIF_ANY_FRAME);
+    av_free(hdr_plus_buf);
+    if (res < 0) {
+        log_encoder_error(avctx, "Error adding HDR10+ to aom_img");
+        return res;
+    }
+    return 0;
+}
+
 #if defined(AOM_CTRL_AV1E_GET_NUM_OPERATING_POINTS) && \
     defined(AOM_CTRL_AV1E_GET_SEQ_LEVEL_IDX) && \
     defined(AOM_CTRL_AV1E_GET_TARGET_SEQ_LEVEL_IDX)
@@ -1234,6 +1284,7 @@ static int aom_encode(AVCodecContext *avctx, AVPacket *pkt,
 
     if (frame) {
         rawimg                      = &ctx->rawimg;
+        aom_img_remove_metadata(rawimg);
         rawimg->planes[AOM_PLANE_Y] = frame->data[0];
         rawimg->planes[AOM_PLANE_U] = frame->data[1];
         rawimg->planes[AOM_PLANE_V] = frame->data[2];
@@ -1284,6 +1335,10 @@ static int aom_encode(AVCodecContext *avctx, AVPacket *pkt,
 
         if (frame->pict_type == AV_PICTURE_TYPE_I)
             flags |= AOM_EFLAG_FORCE_KF;
+
+        res = add_hdr_plus(avctx, rawimg, frame);
+        if (res < 0)
+            return res;
     }
 
     res = aom_codec_encode(&ctx->encoder, rawimg, timestamp, duration, flags);