[FFmpeg-cvslog] avcodec/prores_aw : add support for prores 444 with alpha
Martin Vignali
git at videolan.org
Thu Nov 8 23:22:17 EET 2018
ffmpeg | branch: master | Martin Vignali <martin.vignali at gmail.com> | Thu Nov 8 22:10:29 2018 +0100| [ff3b4f91cd56896366a044f8e45a88956dd66d39] | committer: Martin Vignali
avcodec/prores_aw : add support for prores 444 with alpha
only 16b alpha is supported (not 8 bits)
following official encoder, alpha data doesn't impact
yuv plane quality.
So the alpha data encoding is done after the yuv part.
It's also avoid to loose quality in yuv part when
alpha is not uniform.
the alpha encoding funcs is mainly take from prores_ks
encoder, except for the alpha data reorganization
> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=ff3b4f91cd56896366a044f8e45a88956dd66d39
---
libavcodec/proresenc_anatoliy.c | 165 +++++++++++++++++++++++++++++++++++++---
1 file changed, 155 insertions(+), 10 deletions(-)
diff --git a/libavcodec/proresenc_anatoliy.c b/libavcodec/proresenc_anatoliy.c
index bbcb03294a..f2fe98dbd9 100644
--- a/libavcodec/proresenc_anatoliy.c
+++ b/libavcodec/proresenc_anatoliy.c
@@ -155,11 +155,13 @@ typedef struct {
uint8_t* fill_y;
uint8_t* fill_u;
uint8_t* fill_v;
+ uint8_t* fill_a;
int qmat_luma[16][64];
int qmat_chroma[16][64];
int is_422;
+ int need_alpha;
} ProresContext;
static void encode_codeword(PutBitContext *pb, int val, int codebook)
@@ -361,6 +363,80 @@ static av_always_inline unsigned encode_slice_data(AVCodecContext *avctx,
return *y_data_size + *u_data_size + *v_data_size;
}
+static void put_alpha_diff(PutBitContext *pb, int cur, int prev)
+{
+ const int abits = 16;
+ const int dbits = 7;
+ const int dsize = 1 << dbits - 1;
+ int diff = cur - prev;
+
+ diff = av_mod_uintp2(diff, abits);
+ if (diff >= (1 << abits) - dsize)
+ diff -= 1 << abits;
+ if (diff < -dsize || diff > dsize || !diff) {
+ put_bits(pb, 1, 1);
+ put_bits(pb, abits, diff);
+ } else {
+ put_bits(pb, 1, 0);
+ put_bits(pb, dbits - 1, FFABS(diff) - 1);
+ put_bits(pb, 1, diff < 0);
+ }
+}
+
+static inline void put_alpha_run(PutBitContext *pb, int run)
+{
+ if (run) {
+ put_bits(pb, 1, 0);
+ if (run < 0x10)
+ put_bits(pb, 4, run);
+ else
+ put_bits(pb, 15, run);
+ } else {
+ put_bits(pb, 1, 1);
+ }
+}
+
+static av_always_inline int encode_alpha_slice_data(AVCodecContext *avctx, int8_t * src_a,
+ unsigned mb_count, uint8_t *buf, unsigned data_size, unsigned* a_data_size)
+{
+ const int abits = 16;
+ const int mask = (1 << abits) - 1;
+ const int num_coeffs = mb_count * 256;
+ int prev = mask, cur;
+ int idx = 0;
+ int run = 0;
+ int16_t * blocks = (int16_t *)src_a;
+ PutBitContext pb;
+ init_put_bits(&pb, buf, data_size);
+
+ cur = blocks[idx++];
+ put_alpha_diff(&pb, cur, prev);
+ prev = cur;
+ do {
+ cur = blocks[idx++];
+ if (cur != prev) {
+ put_alpha_run (&pb, run);
+ put_alpha_diff(&pb, cur, prev);
+ prev = cur;
+ run = 0;
+ } else {
+ run++;
+ }
+ } while (idx < num_coeffs);
+ if (run)
+ put_alpha_run(&pb, run);
+ flush_put_bits(&pb);
+ *a_data_size = put_bits_count(&pb) >> 3;
+
+ if (put_bits_left(&pb) < 0) {
+ av_log(avctx, AV_LOG_ERROR,
+ "Underestimated required buffer size.\n");
+ return AVERROR_BUG;
+ } else {
+ return 0;
+ }
+}
+
static void subimage_with_fill(uint16_t *src, unsigned x, unsigned y,
unsigned stride, unsigned width, unsigned height, uint16_t *dst,
unsigned dst_width, unsigned dst_height)
@@ -391,15 +467,46 @@ static void subimage_with_fill(uint16_t *src, unsigned x, unsigned y,
}
}
+/* reorganize alpha data and convert 10b -> 16b */
+static void subimage_alpha_with_fill(uint16_t *src, unsigned x, unsigned y,
+ unsigned stride, unsigned width, unsigned height, uint16_t *dst,
+ unsigned dst_width, unsigned dst_height)
+{
+ int box_width = FFMIN(width - x, dst_width);
+ int box_height = FFMIN(height - y, dst_height);
+ int i, j, src_stride = stride >> 1;
+ uint16_t last_pix, *last_line;
+
+ src += y * src_stride + x;
+ for (i = 0; i < box_height; ++i) {
+ for (j = 0; j < box_width; ++j) {
+ dst[j] = src[j] << 6; /* 10b to 16b */
+ }
+ last_pix = dst[j - 1] << 6; /* 10b to 16b */
+ for (; j < dst_width; j++)
+ dst[j] = last_pix;
+ src += src_stride;
+ dst += dst_width;
+ }
+ last_line = dst - dst_width;
+ for (; i < dst_height; i++) {
+ for (j = 0; j < dst_width; ++j) {
+ dst[j] = last_line[j];
+ }
+ dst += dst_width;
+ }
+}
+
static int encode_slice(AVCodecContext *avctx, const AVFrame *pic, int mb_x,
int mb_y, unsigned mb_count, uint8_t *buf, unsigned data_size,
int unsafe, int *qp)
{
- int luma_stride, chroma_stride;
- int hdr_size = 6, slice_size;
- uint8_t *dest_y, *dest_u, *dest_v;
- unsigned y_data_size = 0, u_data_size = 0, v_data_size = 0;
+ int luma_stride, chroma_stride, alpha_stride = 0;
ProresContext* ctx = avctx->priv_data;
+ int hdr_size = 6 + (ctx->need_alpha * 2); /* v data size is write when there is alpha */
+ int ret = 0, slice_size;
+ uint8_t *dest_y, *dest_u, *dest_v;
+ unsigned y_data_size = 0, u_data_size = 0, v_data_size = 0, a_data_size = 0;
FDCTDSPContext *fdsp = &ctx->fdsp;
int tgt_bits = (mb_count * bitrate_table[avctx->profile]) >> 2;
int low_bytes = (tgt_bits - (tgt_bits >> 3)) >> 3; // 12% bitrate fluctuation
@@ -412,6 +519,9 @@ static int encode_slice(AVCodecContext *avctx, const AVFrame *pic, int mb_x,
luma_stride = pic->linesize[0];
chroma_stride = pic->linesize[1];
+ if (ctx->need_alpha)
+ alpha_stride = pic->linesize[3];
+
dest_y = pic->data[0] + (mb_y << 4) * luma_stride + (mb_x << 5);
dest_u = pic->data[1] + (mb_y << 4) * chroma_stride + (mb_x << (5 - ctx->is_422));
dest_v = pic->data[2] + (mb_y << 4) * chroma_stride + (mb_x << (5 - ctx->is_422));
@@ -431,7 +541,7 @@ static int encode_slice(AVCodecContext *avctx, const AVFrame *pic, int mb_x,
calc_plane_dct(fdsp, ctx->fill_u, blocks_u, mb_count << (5 - ctx->is_422), mb_count, 1, ctx->is_422);
calc_plane_dct(fdsp, ctx->fill_v, blocks_v, mb_count << (5 - ctx->is_422), mb_count, 1, ctx->is_422);
- encode_slice_data(avctx, blocks_y, blocks_u, blocks_v,
+ slice_size = encode_slice_data(avctx, blocks_y, blocks_u, blocks_v,
mb_count, buf + hdr_size, data_size - hdr_size,
&y_data_size, &u_data_size, &v_data_size,
*qp);
@@ -470,7 +580,21 @@ static int encode_slice(AVCodecContext *avctx, const AVFrame *pic, int mb_x,
AV_WB16(buf + 2, y_data_size);
AV_WB16(buf + 4, u_data_size);
- return hdr_size + y_data_size + u_data_size + v_data_size;
+ if (ctx->need_alpha) {
+ AV_WB16(buf + 6, v_data_size); /* write v data size only if there is alpha */
+
+ subimage_alpha_with_fill((uint16_t *) pic->data[3], mb_x << 4, mb_y << 4,
+ alpha_stride, avctx->width, avctx->height,
+ (uint16_t *) ctx->fill_a, mb_count << 4, 16);
+ ret = encode_alpha_slice_data(avctx, ctx->fill_a, mb_count,
+ buf + hdr_size + slice_size,
+ data_size - hdr_size - slice_size, &a_data_size);
+ }
+
+ if (ret != 0) {
+ return ret;
+ }
+ return hdr_size + y_data_size + u_data_size + v_data_size + a_data_size;
}
static int prores_encode_picture(AVCodecContext *avctx, const AVFrame *pic,
@@ -505,6 +629,9 @@ static int prores_encode_picture(AVCodecContext *avctx, const AVFrame *pic,
sl_size = encode_slice(avctx, pic, mb_x, mb_y, slice_mb_count,
sl_data, sl_data_size, unsafe_bot || unsafe_right, &qp);
+ if (sl_size < 0){
+ return sl_size;
+ }
bytestream_put_be16(&sl_data_sizes, sl_size);
sl_data += sl_size;
@@ -536,6 +663,9 @@ static int prores_encode_frame(AVCodecContext *avctx, AVPacket *pkt,
buf = pkt->data;
pic_size = prores_encode_picture(avctx, pict, buf + header_size + 8,
pkt->size - header_size - 8);
+ if (pic_size < 0) {
+ return pic_size;
+ }
bytestream_put_be32(&buf, pic_size + 8 + header_size);
bytestream_put_buffer(&buf, "icpf", 4);
@@ -589,6 +719,7 @@ static av_cold int prores_encode_init(AVCodecContext *avctx)
ProresContext* ctx = avctx->priv_data;
avctx->bits_per_raw_sample = 10;
+ ctx->need_alpha = 0;
if (avctx->width & 0x1) {
av_log(avctx, AV_LOG_ERROR,
@@ -611,8 +742,14 @@ static av_cold int prores_encode_init(AVCodecContext *avctx)
avctx->profile = FF_PROFILE_PRORES_4444;
av_log(avctx, AV_LOG_INFO,
"encoding with ProRes 4444 (ap4h) profile\n");
+ } else if (avctx->pix_fmt == AV_PIX_FMT_YUVA444P10) {
+ avctx->profile = FF_PROFILE_PRORES_4444;
+ av_log(avctx, AV_LOG_INFO,
+ "encoding with ProRes 4444+ (ap4h) profile\n");
+ } else {
+ av_log(avctx, AV_LOG_ERROR, "Unknown pixel format\n");
+ return AVERROR(EINVAL);
}
-
} else if (avctx->profile < FF_PROFILE_PRORES_PROXY
|| avctx->profile > FF_PROFILE_PRORES_4444) {
av_log(
@@ -625,7 +762,8 @@ static av_cold int prores_encode_init(AVCodecContext *avctx)
av_log(avctx, AV_LOG_ERROR,
"encoding with ProRes 444 (ap4h) profile, need YUV444P10 input\n");
return AVERROR(EINVAL);
- } else if ((avctx->pix_fmt == AV_PIX_FMT_YUV444P10) && (avctx->profile < FF_PROFILE_PRORES_4444)){
+ } else if ((avctx->pix_fmt == AV_PIX_FMT_YUV444P10 || avctx->pix_fmt == AV_PIX_FMT_YUVA444P10)
+ && (avctx->profile < FF_PROFILE_PRORES_4444)){
av_log(avctx, AV_LOG_ERROR,
"encoding with ProRes Proxy/LT/422/422 HQ (apco, apcs, apcn, ap4h) profile, need YUV422P10 input\n");
return AVERROR(EINVAL);
@@ -649,6 +787,12 @@ static av_cold int prores_encode_init(AVCodecContext *avctx)
ctx->fill_u = ctx->fill_y + (DEFAULT_SLICE_MB_WIDTH << 9);
ctx->fill_v = ctx->fill_u + (DEFAULT_SLICE_MB_WIDTH << 9);
}
+ if (avctx->pix_fmt == AV_PIX_FMT_YUVA444P10) {
+ ctx->need_alpha = 1;
+ ctx->fill_a = av_malloc(DEFAULT_SLICE_MB_WIDTH << 9); /* 8 blocks x 16px x 16px x sizeof (uint16) */
+ if (!ctx->fill_a)
+ return AVERROR(ENOMEM);
+ }
}
ff_fdctdsp_init(&ctx->fdsp, avctx);
@@ -667,6 +811,7 @@ static av_cold int prores_encode_close(AVCodecContext *avctx)
{
ProresContext* ctx = avctx->priv_data;
av_freep(&ctx->fill_y);
+ av_freep(&ctx->fill_a);
return 0;
}
@@ -680,7 +825,7 @@ AVCodec ff_prores_aw_encoder = {
.init = prores_encode_init,
.close = prores_encode_close,
.encode2 = prores_encode_frame,
- .pix_fmts = (const enum AVPixelFormat[]){AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10, AV_PIX_FMT_NONE},
+ .pix_fmts = (const enum AVPixelFormat[]){AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10, AV_PIX_FMT_YUVA444P10, AV_PIX_FMT_NONE},
.capabilities = AV_CODEC_CAP_FRAME_THREADS | AV_CODEC_CAP_INTRA_ONLY,
.profiles = NULL_IF_CONFIG_SMALL(ff_prores_profiles),
};
@@ -694,7 +839,7 @@ AVCodec ff_prores_encoder = {
.init = prores_encode_init,
.close = prores_encode_close,
.encode2 = prores_encode_frame,
- .pix_fmts = (const enum AVPixelFormat[]){AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10, AV_PIX_FMT_NONE},
+ .pix_fmts = (const enum AVPixelFormat[]){AV_PIX_FMT_YUV422P10, AV_PIX_FMT_YUV444P10, AV_PIX_FMT_YUVA444P10, AV_PIX_FMT_NONE},
.capabilities = AV_CODEC_CAP_FRAME_THREADS | AV_CODEC_CAP_INTRA_ONLY,
.profiles = NULL_IF_CONFIG_SMALL(ff_prores_profiles),
};
More information about the ffmpeg-cvslog
mailing list