backport libx264.c from trunk

now compiles with x264 API versions 65 up to 85

patch prepared by darkshikari


Originally committed as revision 22042 to svn://svn.ffmpeg.org/ffmpeg/branches/0.5

configure

@@ -2017,7 +2017,7 @@ enabled libschroedinger && add_cflags $(pkg-config --cflags schroedinger-1.0) &&
 enabled libspeex   && require  libspeex speex/speex.h speex_decoder_init -lspeex
 enabled libtheora  && require  libtheora theora/theora.h theora_info_init -ltheora -logg
 enabled libvorbis  && require  libvorbis vorbis/vorbisenc.h vorbis_info_init -lvorbisenc -lvorbis -logg
-enabled libx264    && require  libx264 x264.h x264_encoder_open -lx264 -lm &&
+enabled libx264    && require  libx264 x264.h x264_encoder_encode -lx264 -lm &&
                       { check_cpp_condition x264.h "X264_BUILD >= 65" ||
                         die "ERROR: libx264 version must be >= 0.65."; }
 enabled libxvid    && require  libxvid xvid.h xvid_global -lxvidcore

ffpresets/libx264-baseline.ffpreset

@@ -1,3 +1,4 @@
 coder=0
 bf=0
-flags2=-wpred-dct8x8
+flags2=-wpred-dct8x8+mbtree
+wpredp=0

ffpresets/libx264-default.ffpreset

@@ -3,7 +3,7 @@ flags=+loop
 cmp=+chroma
 partitions=+parti8x8+parti4x4+partp8x8+partb8x8
 me_method=hex
-subq=6
+subq=7
 me_range=16
 g=250
 keyint_min=25
@@ -14,5 +14,9 @@ qcomp=0.6
 qmin=10
 qmax=51
 qdiff=4
+bf=3
+refs=3
 directpred=1
-flags2=+fastpskip
+trellis=1
+flags2=+mixed_refs+wpred+dct8x8+fastpskip+mbtree
+wpredp=2

ffpresets/libx264-fastfirstpass.ffpreset

@@ -3,7 +3,7 @@ flags=+loop
 cmp=+chroma
 partitions=-parti8x8-parti4x4-partp8x8-partp4x4-partb8x8
 me_method=dia
-subq=1
+subq=2
 me_range=16
 g=250
 keyint_min=25
@@ -14,8 +14,9 @@ qcomp=0.6
 qmin=10
 qmax=51
 qdiff=4
-bf=4
+bf=3
 refs=1
 directpred=3
 trellis=0
-flags2=-bpyramid-wpred-mixed_refs-dct8x8+fastpskip
+flags2=-bpyramid-wpred-mixed_refs-dct8x8+fastpskip+mbtree
+wpredp=2

ffpresets/libx264-hq.ffpreset

@@ -14,8 +14,9 @@ qcomp=0.6
 qmin=10
 qmax=51
 qdiff=4
-bf=4
+bf=3
 refs=4
 directpred=3
 trellis=1
-flags2=+bpyramid+wpred+mixed_refs+dct8x8+fastpskip
+flags2=+wpred+mixed_refs+dct8x8+fastpskip+mbtree
+wpredp=2

ffpresets/libx264-ipod320.ffpreset

@@ -1,6 +1,7 @@
 coder=0
 bf=0
-flags2=-wpred-dct8x8
+flags2=-wpred-dct8x8+mbtree
 level=13
 maxrate=768000
 bufsize=3000000
+wpredp=0

ffpresets/libx264-ipod640.ffpreset

@@ -1,7 +1,8 @@
 coder=0
 bf=0
 refs=1
-flags2=-wpred-dct8x8
+flags2=-wpred-dct8x8+mbtree
 level=30
 maxrate=10000000
 bufsize=10000000
+wpredp=0

ffpresets/libx264-lossless_fast.ffpreset

@@ -15,5 +15,6 @@ qmin=10
 qmax=51
 qdiff=4
 directpred=1
-flags2=+fastpskip
+flags2=+fastpskip+mbtree
 cqp=0
+wpredp=0

ffpresets/libx264-lossless_max.ffpreset

@@ -16,5 +16,6 @@ qmax=51
 qdiff=4
 refs=16
 directpred=1
-flags2=+mixed_refs+dct8x8+fastpskip
+flags2=+mixed_refs+dct8x8+fastpskip+mbtree
 cqp=0
+wpredp=2

ffpresets/libx264-lossless_medium.ffpreset

@@ -15,5 +15,6 @@ qmin=10
 qmax=51
 qdiff=4
 directpred=1
-flags2=+fastpskip
+flags2=+fastpskip+mbtree
 cqp=0
+wpredp=2

ffpresets/libx264-lossless_slow.ffpreset

@@ -16,5 +16,6 @@ qmax=51
 qdiff=4
 refs=2
 directpred=1
-flags2=+dct8x8+fastpskip
+flags2=+dct8x8+fastpskip+mbtree
 cqp=0
+wpredp=2

ffpresets/libx264-lossless_slower.ffpreset

@@ -16,5 +16,6 @@ qmax=51
 qdiff=4
 refs=4
 directpred=1
-flags2=+mixed_refs+dct8x8+fastpskip
+flags2=+mixed_refs+dct8x8+fastpskip+mbtree
 cqp=0
+wpredp=2

ffpresets/libx264-lossless_ultrafast.ffpreset

@@ -15,5 +15,5 @@ qmin=10
 qmax=51
 qdiff=4
 directpred=1
-flags2=+fastpskip
+flags2=+fastpskip+mbtree
 cqp=0

ffpresets/libx264-main.ffpreset

@@ -1 +1 @@
-flags2=-dct8x8
+flags2=-dct8x8+mbtree

ffpresets/libx264-max.ffpreset

@@ -3,8 +3,8 @@ flags=+loop
 cmp=+chroma
 partitions=+parti8x8+parti4x4+partp8x8+partp4x4+partb8x8
 me_method=tesa
-subq=9
+subq=10
-me_range=32
+me_range=24
 g=250
 keyint_min=25
 sc_threshold=40
@@ -14,8 +14,9 @@ qcomp=0.6
 qmin=10
 qmax=51
 qdiff=4
-bf=4
+bf=3
 refs=16
 directpred=3
 trellis=2
-flags2=+bpyramid+wpred+mixed_refs+dct8x8-fastpskip
+flags2=+wpred+mixed_refs+dct8x8-fastpskip+mbtree
+wpredp=2

ffpresets/libx264-normal.ffpreset

@@ -14,8 +14,9 @@ qcomp=0.6
 qmin=10
 qmax=51
 qdiff=4
-bf=4
+bf=3
 refs=2
 directpred=3
 trellis=0
-flags2=+bpyramid+wpred+dct8x8+fastpskip
+flags2=+wpred+dct8x8+fastpskip+mbtree
+wpredp=2

ffpresets/libx264-slowfirstpass.ffpreset

@@ -14,8 +14,9 @@ qcomp=0.6
 qmin=10
 qmax=51
 qdiff=4
-bf=4
+bf=3
 refs=1
 directpred=3
 trellis=0
-flags2=+bpyramid+wpred+dct8x8+fastpskip
+flags2=+wpred+dct8x8+fastpskip+mbtree
+wpredp=2

libavcodec/avcodec.h

@@ -501,6 +501,7 @@ typedef struct RcOverride{
 #define CODEC_FLAG2_CHUNKS        0x00008000 ///< Input bitstream might be truncated at a packet boundaries instead of only at frame boundaries.
 #define CODEC_FLAG2_NON_LINEAR_QUANT 0x00010000 ///< Use MPEG-2 nonlinear quantizer.
 #define CODEC_FLAG2_BIT_RESERVOIR 0x00020000 ///< Use a bit reservoir when encoding if possible
+#define CODEC_FLAG2_MBTREE        0x00040000 ///< Use macroblock tree ratecontrol (x264 only)
 
 /* Unsupported options :
  *              Syntax Arithmetic coding (SAC)
@@ -2332,6 +2333,16 @@ typedef struct AVCodecContext {
      * Set to time_base ticks per frame. Default 1, e.g., H.264/MPEG-2 set it to 2.
      */
     int ticks_per_frame;
+
+    /**
+     * explicit P-frame weighted prediction analysis method
+     * 0: off
+     * 1: fast blind weighting (one reference duplicate with -1 offset)
+     * 2: smart weighting (full fade detection analysis)
+     * - encoding: Set by user.
+     * - decoding: unused
+     */
+    int weighted_p_pred;
 } AVCodecContext;
 
 /**

libavcodec/libx264.c

@@ -30,6 +30,8 @@ typedef struct X264Context {
     x264_param_t    params;
     x264_t         *enc;
     x264_picture_t  pic;
+    uint8_t        *sei;
+    int             sei_size;
     AVFrame         out_pic;
 } X264Context;
 
@@ -48,14 +50,58 @@ static void X264_log(void *p, int level, const char *fmt, va_list args)
     av_vlog(p, level_map[level], fmt, args);
 }
 
-
+#if X264_BUILD >= 76
-static int encode_nals(uint8_t *buf, int size, x264_nal_t *nals, int nnal)
+static int encode_nals(AVCodecContext *ctx, uint8_t *buf, int size,
+                       x264_nal_t *nals, int nnal, int skip_sei)
 {
+    X264Context *x4 = ctx->priv_data;
     uint8_t *p = buf;
     int i;
 
+    /* Write the SEI as part of the first frame. */
+    if (x4->sei_size > 0 && nnal > 0) {
+        memcpy(p, x4->sei, x4->sei_size);
+        p += x4->sei_size;
+        x4->sei_size = 0;
+    }
+
     for (i = 0; i < nnal; i++){
-        int s = x264_nal_encode(p, &size, 1, nals + i);
+        /* Don't put the SEI in extradata. */
+        if (skip_sei && nals[i].i_type == NAL_SEI) {
+            x4->sei_size = nals[i].i_payload;
+            x4->sei      = av_malloc(x4->sei_size);
+            memcpy(x4->sei, nals[i].p_payload, nals[i].i_payload);
+            continue;
+        }
+        memcpy(p, nals[i].p_payload, nals[i].i_payload);
+        p += nals[i].i_payload;
+    }
+
+    return p - buf;
+}
+#else
+static int encode_nals(AVCodecContext *ctx, uint8_t *buf, int size, x264_nal_t *nals, int nnal, int skip_sei)
+{
+    X264Context *x4 = ctx->priv_data;
+    uint8_t *p = buf;
+    int i, s;
+
+    /* Write the SEI as part of the first frame. */
+    if (x4->sei_size > 0 && nnal > 0) {
+        memcpy(p, x4->sei, x4->sei_size);
+        p += x4->sei_size;
+        x4->sei_size = 0;
+    }
+
+    for (i = 0; i < nnal; i++) {
+        /* Don't put the SEI in extradata. */
+        if (skip_sei && nals[i].i_type == NAL_SEI) {
+            x4->sei = av_malloc( 5 + nals[i].i_payload * 4 / 3 );
+            if(x264_nal_encode(x4->sei, &x4->sei_size, 1, nals + i) < 0)
+                return -1;
+            continue;
+        }
+        s = x264_nal_encode(p, &size, 1, nals + i);
         if (s < 0)
             return -1;
         p += s;
@@ -63,6 +109,7 @@ static int encode_nals(uint8_t *buf, int size, x264_nal_t *nals, int nnal)
 
     return p - buf;
 }
+#endif
 
 static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
                       int bufsize, void *data)
@@ -86,15 +133,14 @@ static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
         x4->pic.i_type = X264_TYPE_AUTO;
     }
 
-    if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL,
+    if (x264_encoder_encode(x4->enc, &nal, &nnal, frame? &x4->pic: NULL, &pic_out) < 0)
-                            &pic_out))
         return -1;
 
-    bufsize = encode_nals(buf, bufsize, nal, nnal);
+    bufsize = encode_nals(ctx, buf, bufsize, nal, nnal, 0);
     if (bufsize < 0)
         return -1;
 
-    /* FIXME: dts */
+    /* FIXME: libx264 now provides DTS, but AVFrame doesn't have a field for it. */
     x4->out_pic.pts = pic_out.i_pts;
 
     switch (pic_out.i_type) {
@@ -111,7 +157,11 @@ static int X264_frame(AVCodecContext *ctx, uint8_t *buf,
         break;
     }
 
+#if X264_BUILD < 82
     x4->out_pic.key_frame = pic_out.i_type == X264_TYPE_IDR;
+#else
+    x4->out_pic.key_frame = pic_out.b_keyframe;
+#endif
     x4->out_pic.quality   = (pic_out.i_qpplus1 - 1) * FF_QP2LAMBDA;
 
     return bufsize;
@@ -122,6 +172,7 @@ static av_cold int X264_close(AVCodecContext *avctx)
     X264Context *x4 = avctx->priv_data;
 
     av_freep(&avctx->extradata);
+    av_free(x4->sei);
 
     if (x4->enc)
         x264_encoder_close(x4->enc);
@@ -133,6 +184,7 @@ static av_cold int X264_init(AVCodecContext *avctx)
 {
     X264Context *x4 = avctx->priv_data;
 
+    x4->sei_size = 0;
     x264_param_default(&x4->params);
 
     x4->params.pf_log               = X264_log;
@@ -164,7 +216,11 @@ static av_cold int X264_init(AVCodecContext *avctx)
     x4->params.b_cabac           = avctx->coder_type == FF_CODER_TYPE_AC;
     x4->params.i_bframe_adaptive = avctx->b_frame_strategy;
     x4->params.i_bframe_bias     = avctx->bframebias;
+#if X264_BUILD >= 78
+    x4->params.i_bframe_pyramid  = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? X264_B_PYRAMID_NORMAL : X264_B_PYRAMID_NONE;
+#else
     x4->params.b_bframe_pyramid  = avctx->flags2 & CODEC_FLAG2_BPYRAMID;
+#endif
     avctx->has_b_frames          = avctx->flags2 & CODEC_FLAG2_BPYRAMID ? 2 : !!avctx->max_b_frames;
 
     x4->params.i_keyint_min = avctx->keyint_min;
@@ -191,8 +247,10 @@ static av_cold int X264_init(AVCodecContext *avctx)
     x4->params.i_height             = avctx->height;
     x4->params.vui.i_sar_width      = avctx->sample_aspect_ratio.num;
     x4->params.vui.i_sar_height     = avctx->sample_aspect_ratio.den;
-    x4->params.i_fps_num            = avctx->time_base.den;
+#if X264_BUILD >= 81
-    x4->params.i_fps_den            = avctx->time_base.num;
+    x4->params.i_fps_num = x4->params.i_timebase_den = avctx->time_base.den;
+    x4->params.i_fps_den = x4->params.i_timebase_num = avctx->time_base.num;
+#endif
 
     x4->params.analyse.inter    = 0;
     if (avctx->partitions) {
@@ -211,6 +269,9 @@ static av_cold int X264_init(AVCodecContext *avctx)
     x4->params.analyse.i_direct_mv_pred  = avctx->directpred;
 
     x4->params.analyse.b_weighted_bipred = avctx->flags2 & CODEC_FLAG2_WPRED;
+#if X264_BUILD >= 79
+    x4->params.analyse.i_weighted_pred = avctx->weighted_p_pred;
+#endif
 
     if (avctx->me_method == ME_EPZS)
         x4->params.analyse.i_me_method = X264_ME_DIA;
@@ -248,6 +309,9 @@ static av_cold int X264_init(AVCodecContext *avctx)
     } else
         x4->params.rc.f_vbv_buffer_init = 0.9;
 
+#if X264_BUILD >= 69
+    x4->params.rc.b_mb_tree               = !!(avctx->flags2 & CODEC_FLAG2_MBTREE);
+#endif
     x4->params.rc.f_ip_factor             = 1 / fabs(avctx->i_quant_factor);
     x4->params.rc.f_pb_factor             = avctx->b_quant_factor;
     x4->params.analyse.i_chroma_qp_offset = avctx->chromaoffset;
@@ -270,6 +334,17 @@ static av_cold int X264_init(AVCodecContext *avctx)
 
     avctx->coded_frame = &x4->out_pic;
 
+#if X264_BUILD >= 76
+    if (avctx->flags & CODEC_FLAG_GLOBAL_HEADER) {
+        x264_nal_t *nal;
+        int nnal, s;
+
+        s = x264_encoder_headers(x4->enc, &nal, &nnal);
+
+        avctx->extradata      = av_malloc(s);
+        avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
+    }
+#else
     if(avctx->flags & CODEC_FLAG_GLOBAL_HEADER){
         x264_nal_t *nal;
         int nnal, i, s = 0;
@@ -281,8 +356,9 @@ static av_cold int X264_init(AVCodecContext *avctx)
              s += 5 + nal[i].i_payload * 4 / 3;
 
         avctx->extradata = av_malloc(s);
-        avctx->extradata_size = encode_nals(avctx->extradata, s, nal, nnal);
+        avctx->extradata_size = encode_nals(avctx, avctx->extradata, s, nal, nnal, 1);
     }
+#endif
 
     return 0;
 }
@@ -296,6 +372,6 @@ AVCodec libx264_encoder = {
     .encode         = X264_frame,
     .close          = X264_close,
     .capabilities   = CODEC_CAP_DELAY,
-    .pix_fmts       = (enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_NONE },
+    .pix_fmts       = (const enum PixelFormat[]) { PIX_FMT_YUV420P, PIX_FMT_NONE },
     .long_name      = NULL_IF_CONFIG_SMALL("libx264 H.264 / AVC / MPEG-4 AVC / MPEG-4 part 10"),
 };

libavcodec/options.c

@@ -122,6 +122,7 @@ static const AVOption options[]={
 {"b_qfactor", "qp factor between p and b frames", OFFSET(b_quant_factor), FF_OPT_TYPE_FLOAT, 1.25, -FLT_MAX, FLT_MAX, V|E},
 {"rc_strategy", "ratecontrol method", OFFSET(rc_strategy), FF_OPT_TYPE_INT, DEFAULT, INT_MIN, INT_MAX, V|E},
 {"b_strategy", "strategy to choose between I/P/B-frames", OFFSET(b_frame_strategy), FF_OPT_TYPE_INT, 0, INT_MIN, INT_MAX, V|E},
+{"wpredp", "weighted prediction analysis method", OFFSET(weighted_p_pred), FF_OPT_TYPE_INT, 0, INT_MIN, INT_MAX, V|E},
 {"hurry_up", NULL, OFFSET(hurry_up), FF_OPT_TYPE_INT, DEFAULT, INT_MIN, INT_MAX, V|D},
 {"ps", "rtp payload size in bits", OFFSET(rtp_payload_size), FF_OPT_TYPE_INT, DEFAULT, INT_MIN, INT_MAX, V|E},
 {"mv_bits", NULL, OFFSET(mv_bits), FF_OPT_TYPE_INT, DEFAULT, INT_MIN, INT_MAX},
@@ -388,6 +389,7 @@ static const AVOption options[]={
 {"request_channels", "set desired number of audio channels", OFFSET(request_channels), FF_OPT_TYPE_INT, DEFAULT, 0, INT_MAX, A|D},
 {"drc_scale", "percentage of dynamic range compression to apply", OFFSET(drc_scale), FF_OPT_TYPE_FLOAT, 1.0, 0.0, 1.0, A|D},
 {"reservoir", "use bit reservoir", 0, FF_OPT_TYPE_CONST, CODEC_FLAG2_BIT_RESERVOIR, INT_MIN, INT_MAX, A|E, "flags2"},
+{"mbtree", "use macroblock tree ratecontrol (x264 only)", 0, FF_OPT_TYPE_CONST, CODEC_FLAG2_MBTREE, INT_MIN, INT_MAX, V|E, "flags2"},
 {"bits_per_raw_sample", NULL, OFFSET(bits_per_raw_sample), FF_OPT_TYPE_INT, DEFAULT, INT_MIN, INT_MAX},
 {"channel_layout", NULL, OFFSET(channel_layout), FF_OPT_TYPE_INT64, DEFAULT, 0, INT64_MAX, A|E|D, "channel_layout"},
 {"request_channel_layout", NULL, OFFSET(request_channel_layout), FF_OPT_TYPE_INT64, DEFAULT, 0, INT64_MAX, A|D, "request_channel_layout"},