H.261 encoder by (Maarten Daniels <maarten dot daniels at luc dot ac dot be>)

Originally committed as revision 3643 to svn://svn.ffmpeg.org/ffmpeg/trunk

libavcodec/allcodecs.c

@@ -59,6 +59,7 @@ void avcodec_register_all(void)
 //    register_avcodec(&h264_encoder);
 #ifdef CONFIG_RISKY
     register_avcodec(&mpeg2video_encoder);
+    register_avcodec(&h261_encoder);
     register_avcodec(&h263_encoder);
     register_avcodec(&h263p_encoder);
     register_avcodec(&flv_encoder);

libavcodec/avcodec.h

@@ -1786,6 +1786,7 @@ extern AVCodec faac_encoder;
 extern AVCodec xvid_encoder;
 extern AVCodec mpeg1video_encoder;
 extern AVCodec mpeg2video_encoder;
+extern AVCodec h261_encoder;
 extern AVCodec h263_encoder;
 extern AVCodec h263p_encoder;
 extern AVCodec flv_encoder;

libavcodec/h261.c

@@ -47,6 +47,7 @@ typedef struct H261Context{
     MpegEncContext s;
 
     int current_mba;
+    int previous_mba;
     int mba_diff;
     int mtype;
     int current_mv_x;
@@ -76,11 +77,330 @@ void ff_h261_loop_filter(MpegEncContext *s){
     s->dsp.h261_loop_filter(dest_cr, uvlinesize);
 }
 
+int ff_h261_get_picture_format(int width, int height){
+    // QCIF
+    if (width == 176 && height == 144)
+        return 0;
+    // CIF
+    else if (width == 352 && height == 288)
+        return 1;
+    // ERROR
+    else
+        return -1;
+}
+
+static void h261_encode_block(H261Context * h, DCTELEM * block,
+                              int n);
 static int h261_decode_block(H261Context *h, DCTELEM *block,
                              int n, int coded);
 static int h261_decode_mb(H261Context *h);
 void ff_set_qscale(MpegEncContext * s, int qscale);
 
+void ff_h261_encode_picture_header(MpegEncContext * s, int picture_number){
+    H261Context * h = (H261Context *) s;
+    int format, coded_frame_rate, coded_frame_rate_base, temp_ref;
+    int best_clock_code=1;
+    int best_divisor=60;
+    coded_frame_rate= 1800000;
+    coded_frame_rate_base= (1000+best_clock_code)*best_divisor;
+
+    align_put_bits(&s->pb);
+
+    /* Update the pointer to last GOB */
+    s->ptr_lastgob = pbBufPtr(&s->pb);
+
+    put_bits(&s->pb, 20, 0x10); /* PSC */
+
+    temp_ref= s->picture_number * (int64_t)coded_frame_rate * s->avctx->frame_rate_base / 
+                         (coded_frame_rate_base * (int64_t)s->avctx->frame_rate);
+    put_bits(&s->pb, 5, temp_ref & 0x1f); /* TemporalReference */
+
+    put_bits(&s->pb, 1, 0); /* split screen off */
+    put_bits(&s->pb, 1, 0); /* camera  off */
+    put_bits(&s->pb, 1, 0); /* freeze picture release off */
+    
+    format = ff_h261_get_picture_format(s->width, s->height);
+    
+    put_bits(&s->pb, 1, format); /* 0 == QCIF, 1 == CIF */
+
+    put_bits(&s->pb, 1, 0); /* still image mode */
+    put_bits(&s->pb, 1, 0); /* reserved */
+
+    put_bits(&s->pb, 1, 0); /* no PEI */    
+    if(format == 0)
+        h->gob_number = -1;
+    else
+        h->gob_number = 0;
+    h->current_mba = 0;
+}
+
+/**
+ * Encodes a group of blocks header.
+ */
+static void h261_encode_gob_header(MpegEncContext * s, int mb_line){
+    H261Context * h = (H261Context *)s;
+    if(ff_h261_get_picture_format(s->width, s->height) == 0){
+        h->gob_number+=2; // QCIF
+    }
+    else{
+        h->gob_number++; // CIF
+    }
+    put_bits(&s->pb, 16, 1); /* GBSC */
+    put_bits(&s->pb, 4, h->gob_number); /* GN */
+    put_bits(&s->pb, 5, s->qscale); /* GQUANT */
+    put_bits(&s->pb, 1, 0); /* no GEI */
+    h->current_mba = 0;
+    h->previous_mba = 0;
+    h->current_mv_x=0;
+    h->current_mv_y=0;
+}
+
+void ff_h261_reorder_mb_index(MpegEncContext* s){
+    /* for CIF the GOB's are fragmented in the middle of a scanline
+       that's why we need to adjust the x and y index of the macroblocks */
+    if(ff_h261_get_picture_format(s->width,s->height) == 1){ // CIF
+        if((s->mb_x == 0 && (s->mb_y % 3 == 0) ) || (s->mb_x == 11 && ((s->mb_y -1 )% 3 == 0) ))
+            h261_encode_gob_header(s,0);
+        if(s->mb_x < 11 ){
+            if((s->mb_y % 3) == 1 ){
+                s->mb_x += 0;
+                s->mb_y += 1;
+            }
+            else if( (s->mb_y % 3) == 2 ){
+                s->mb_x += 11;
+                s->mb_y -= 1;
+            }
+        }
+        else{
+            if((s->mb_y % 3) == 1 ){
+                s->mb_x += 0;
+                s->mb_y -= 1;
+            }
+            else if( (s->mb_y % 3) == 0 ){
+                s->mb_x -= 11;
+                s->mb_y += 1;
+            }
+        }
+        ff_init_block_index(s);
+        ff_update_block_index(s);
+    /* for QCIF we don't need to reorder MB's
+       there the GOB's aren't fragmented in the middle of a scanline */
+    }else if(ff_h261_get_picture_format(s->width,s->height) == 0){ // QCIF
+        if(s->mb_y % 3 == 0 && s->mb_x == 0)
+            h261_encode_gob_header(s,0);
+    }
+}
+
+static void h261_encode_motion(H261Context * h, int val){
+    MpegEncContext * const s = &h->s;
+    int sign, code;
+    if(val==0){
+        code = 0;
+        put_bits(&s->pb,h261_mv_tab[code][1],h261_mv_tab[code][0]);
+    } 
+    else{
+        if(val > 16)
+            val -=32;
+        if(val < -16)
+            val+=32;
+        sign = val < 0;
+        code = sign ? -val : val; 
+        put_bits(&s->pb,h261_mv_tab[code][1],h261_mv_tab[code][0]);
+        put_bits(&s->pb,1,sign);
+    }
+}
+
+static inline int get_cbp(MpegEncContext * s,
+                      DCTELEM block[6][64])
+{
+    int i, cbp;
+    cbp= 0;
+    for (i = 0; i < 6; i++) {
+        if (s->block_last_index[i] >= 0)
+            cbp |= 1 << (5 - i);
+    }
+    return cbp;
+}
+void ff_h261_encode_mb(MpegEncContext * s,
+         DCTELEM block[6][64],
+         int motion_x, int motion_y)
+{
+    H261Context * h = (H261Context *)s;
+    int old_mtype, mvd, mv_diff_x, mv_diff_y, i, cbp;
+    cbp = 63; // avoid warning
+    mvd = 0;
+ 
+    h->current_mba++;
+    old_mtype = h->mtype;
+    h->mtype = 0;
+ 
+    if (!s->mb_intra){
+        /* compute cbp */
+        cbp= get_cbp(s, block);
+   
+        /* mvd indicates if this block is motion compensated */
+        if(((motion_x >> 1) - h->current_mv_x != 0) || ((motion_y >> 1 ) - h->current_mv_y) != 0){
+            mvd = 1;
+        }
+        else if((motion_x >> 1 == 0) && (motion_y >> 1 == 0)){
+            mvd = 0;
+        }
+        else
+            mvd = 1;
+        if((cbp | mvd | s->dquant ) == 0) {
+            /* skip macroblock */
+            s->skip_count++;
+            h->current_mv_x=0;
+            h->current_mv_y=0;
+            return;
+        }
+    }
+
+    /* MB is not skipped, encode MBA */
+    put_bits(&s->pb, h261_mba_bits[(h->current_mba-h->previous_mba)-1], h261_mba_code[(h->current_mba-h->previous_mba)-1]);
+ 
+    /* calculate MTYPE */
+    if(!s->mb_intra){
+        h->mtype+=2;
+    if(mvd == 1){   
+        h->mtype+=2;
+        if(cbp!=0)
+            h->mtype+=1;
+        if(s->loop_filter)
+            h->mtype+=3;
+        }
+    }
+
+    if(s->dquant) 
+        h->mtype++;
+
+    put_bits(&s->pb, h261_mtype_bits[h->mtype], h261_mtype_code[h->mtype]);
+ 
+    h->mtype = h261_mtype_map[h->mtype];
+ 
+    if(IS_QUANT(h->mtype)){
+        ff_set_qscale(s,s->qscale+s->dquant);
+        put_bits(&s->pb, 5, s->qscale);
+    }
+ 
+    if(IS_16X16(h->mtype)){
+        mv_diff_x = (motion_x >> 1) - h->current_mv_x;
+        mv_diff_y = (motion_y >> 1) - h->current_mv_y;
+        h->current_mv_x = (motion_x >> 1);
+        h->current_mv_y = (motion_y >> 1);
+        h261_encode_motion(h,mv_diff_x);
+        h261_encode_motion(h,mv_diff_y);
+    }
+ 
+    h->previous_mba = h->current_mba;
+ 
+    if(HAS_CBP(h->mtype)){
+        put_bits(&s->pb,h261_cbp_tab[cbp-1][1],h261_cbp_tab[cbp-1][0]); 
+    }
+    for(i=0; i<6; i++) {
+        /* encode each block */
+        h261_encode_block(h, block[i], i);
+    }
+
+    if ( ( h->current_mba == 11 ) || ( h->current_mba == 22 ) || ( h->current_mba == 33 ) || ( !IS_16X16 ( h->mtype ) )){
+        h->current_mv_x=0;
+        h->current_mv_y=0;
+    }
+}
+
+void ff_h261_encode_init(MpegEncContext *s){
+    static int done = 0;
+    
+    if (!done) {
+        done = 1;
+        init_rl(&h261_rl_tcoeff);
+    }
+
+    s->min_qcoeff= -127;
+    s->max_qcoeff=  127;
+    s->y_dc_scale_table=
+    s->c_dc_scale_table= ff_mpeg1_dc_scale_table;
+}
+
+
+/**
+ * encodes a 8x8 block.
+ * @param block the 8x8 block
+ * @param n block index (0-3 are luma, 4-5 are chroma)
+ */
+static void h261_encode_block(H261Context * h, DCTELEM * block, int n){
+    MpegEncContext * const s = &h->s;
+    int level, run, last, i, j, last_index, last_non_zero, sign, slevel, code;
+    RLTable *rl;
+
+    rl = &h261_rl_tcoeff;
+    if (s->mb_intra) {
+        /* DC coef */
+        level = block[0];
+        /* 255 cannot be represented, so we clamp */
+        if (level > 254) {
+            level = 254;
+            block[0] = 254;
+        }
+        /* 0 cannot be represented also */
+        else if (level < 1) {
+            level = 1;
+            block[0] = 1;
+        }
+        if (level == 128)
+            put_bits(&s->pb, 8, 0xff);
+        else
+            put_bits(&s->pb, 8, level);
+        i = 1;
+    } else if((block[0]==1 || block[0] == -1) && (s->block_last_index[n] > -1)){
+        //special case
+        put_bits(&s->pb,1,1);
+        put_bits(&s->pb,1,block[0]>0 ? 0 : 1 );
+        i = 1;
+    } else {
+        i = 0;
+    }
+   
+    /* AC coefs */
+    last_index = s->block_last_index[n];
+    last_non_zero = i - 1;
+    for (; i <= last_index; i++) {
+        j = s->intra_scantable.permutated[i];
+        level = block[j];
+        if (level) {
+            run = i - last_non_zero - 1;
+            last = (i == last_index);
+            sign = 0;
+            slevel = level;
+            if (level < 0) {
+                sign = 1;
+                level = -level;
+            }
+            code = get_rl_index(rl, 0 /*no last in H.261, EOB is used*/, run, level);
+            if(run==0 && level < 16)
+            code+=1;
+            put_bits(&s->pb, rl->table_vlc[code][1], rl->table_vlc[code][0]);
+            if (code == rl->n) {
+                put_bits(&s->pb, 6, run);
+                assert(slevel != 0);
+                if(slevel < -127){
+                    slevel = -127;
+                }
+                else if(slevel > 127){
+                    slevel = 127;
+                }
+                put_bits(&s->pb, 8, slevel & 0xff);
+            } else {
+                put_bits(&s->pb, 1, sign);
+            }
+            last_non_zero = i;
+        }
+    }
+    if(last_index > -1){
+        put_bits(&s->pb, rl->table_vlc[0][1], rl->table_vlc[0][0]);// END OF BLOCK
+    }
+}
+
 /***********************************************/
 /* decoding */
 
@@ -767,6 +1087,16 @@ static int h261_decode_end(AVCodecContext *avctx)
     return 0;
 }
 
+AVCodec h261_encoder = {
+    "h261",
+    CODEC_TYPE_VIDEO,
+    CODEC_ID_H261,
+    sizeof(H261Context),
+    MPV_encode_init,
+    MPV_encode_picture,
+    MPV_encode_end,
+};
+
 AVCodec h261_decoder = {
     "h261",
     CODEC_TYPE_VIDEO,

libavcodec/i386/mpegvideo_mmx_template.c

@@ -88,7 +88,7 @@ static int RENAME(dct_quantize)(MpegEncContext *s,
         qmat = s->q_inter_matrix16[qscale][0];
     }
 
-    if(s->out_format == FMT_H263 && s->mpeg_quant==0){
+    if((s->out_format == FMT_H263 || s->out_format == FMT_H261) && s->mpeg_quant==0){
     
         asm volatile(
             "movd %%"REG_a", %%mm3		\n\t" // last_non_zero_p1

libavcodec/motion_est.c

@@ -279,6 +279,10 @@ void ff_init_me(MpegEncContext *s){
         c->hpel_put[2][2]= c->hpel_put[2][3]= zero_hpel;
     }
 
+    if(s->codec_id == CODEC_ID_H261){
+        c->sub_motion_search= no_sub_motion_search;
+    }
+
     c->temp= c->scratchpad;
 }
       
@@ -691,6 +695,12 @@ static inline void get_limits(MpegEncContext *s, int x, int y)
         c->ymin = - y - 16;
         c->xmax = - x + s->mb_width *16;
         c->ymax = - y + s->mb_height*16;
+    } else if (s->out_format == FMT_H261){
+        // Search range of H261 is different from other codec standards
+        c->xmin = (x > 15) ? - 15 : 0;
+        c->ymin = (y > 15) ? - 15 : 0;
+        c->xmax = (x < s->mb_width * 16 - 16) ? 15 : 0;              
+        c->ymax = (y < s->mb_height * 16 - 16) ? 15 : 0;
     } else {
         c->xmin = - x;
         c->ymin = - y;

libavcodec/motion_est_template.c

@@ -221,6 +221,16 @@ static int hpel_motion_search(MpegEncContext * s,
 }
 #endif
 
+static int no_sub_motion_search(MpegEncContext * s,
+          int *mx_ptr, int *my_ptr, int dmin,
+                                  int src_index, int ref_index,
+                                  int size, int h)
+{
+    (*mx_ptr)<<=1;
+    (*my_ptr)<<=1;
+    return dmin;
+}
+
 int inline ff_get_mb_score(MpegEncContext * s, int mx, int my, int src_index,
                                int ref_index, int size, int h, int add_rate)
 {

libavcodec/mpegvideo.c

@@ -1080,6 +1080,11 @@ int MPV_encode_init(AVCodecContext *avctx)
         s->low_delay=1;
         break;
 #ifdef CONFIG_RISKY
+    case CODEC_ID_H261:
+        s->out_format = FMT_H261;
+        avctx->delay=0;
+        s->low_delay=1;
+        break;
     case CODEC_ID_H263:
         if (h263_get_picture_format(s->width, s->height) == 7) {
             av_log(avctx, AV_LOG_INFO, "Input picture size isn't suitable for h263 codec! try h263+\n");
@@ -1199,6 +1204,8 @@ int MPV_encode_init(AVCodecContext *avctx)
     
 #ifdef CONFIG_ENCODERS
 #ifdef CONFIG_RISKY
+    if (s->out_format == FMT_H261)
+        ff_h261_encode_init(s);
     if (s->out_format == FMT_H263)
         h263_encode_init(s);
     if(s->msmpeg4_version)
@@ -1215,7 +1222,7 @@ int MPV_encode_init(AVCodecContext *avctx)
         if(s->codec_id==CODEC_ID_MPEG4 && s->mpeg_quant){
             s->intra_matrix[j] = ff_mpeg4_default_intra_matrix[i];
             s->inter_matrix[j] = ff_mpeg4_default_non_intra_matrix[i];
-        }else if(s->out_format == FMT_H263){
+        }else if(s->out_format == FMT_H263 || s->out_format == FMT_H261){
             s->intra_matrix[j] =
             s->inter_matrix[j] = ff_mpeg1_default_non_intra_matrix[i];
         }else
@@ -4127,6 +4134,8 @@ static void encode_mb(MpegEncContext *s, int motion_x, int motion_y)
         msmpeg4_encode_mb(s, s->block, motion_x, motion_y); break;
     case CODEC_ID_WMV2:
          ff_wmv2_encode_mb(s, s->block, motion_x, motion_y); break;
+    case CODEC_ID_H261:
+        ff_h261_encode_mb(s, s->block, motion_x, motion_y); break;
     case CODEC_ID_H263:
     case CODEC_ID_H263P:
     case CODEC_ID_FLV1:
@@ -4495,15 +4504,21 @@ static int encode_thread(AVCodecContext *c, void *arg){
         ff_init_block_index(s);
         
         for(mb_x=0; mb_x < s->mb_width; mb_x++) {
-            const int xy= mb_y*s->mb_stride + mb_x;
+            int xy= mb_y*s->mb_stride + mb_x; // removed const, H261 needs to adjust this
             int mb_type= s->mb_type[xy];
 //            int d;
             int dmin= INT_MAX;
             int dir;
 
             s->mb_x = mb_x;
+            s->mb_y = mb_y;  // moved into loop, can get changed by H.261
             ff_update_block_index(s);
 
+            if(s->codec_id == CODEC_ID_H261){
+                ff_h261_reorder_mb_index(s);
+                xy= s->mb_y*s->mb_stride + s->mb_x;
+            }
+
             /* write gob / video packet header  */
 #ifdef CONFIG_RISKY
             if(s->rtp_mode){
@@ -5215,6 +5230,9 @@ static void encode_picture(MpegEncContext *s, int picture_number)
         mjpeg_picture_header(s);
         break;
 #ifdef CONFIG_RISKY
+    case FMT_H261:
+        ff_h261_encode_picture_header(s, picture_number);
+        break;
     case FMT_H263:
         if (s->codec_id == CODEC_ID_WMV2) 
             ff_wmv2_encode_picture_header(s, picture_number);

libavcodec/mpegvideo.h

@@ -865,6 +865,12 @@ extern const uint8_t ff_h263_loop_filter_strength[32];
 
 /* h261.c */
 void ff_h261_loop_filter(MpegEncContext *s);
+void ff_h261_reorder_mb_index(MpegEncContext* s);
+void ff_h261_encode_mb(MpegEncContext *s,
+                    DCTELEM block[6][64],
+                    int motion_x, int motion_y);
+void ff_h261_encode_picture_header(MpegEncContext * s, int picture_number);
+void ff_h261_encode_init(MpegEncContext *s);
 
 
 /* h263.c, h263dec.c */

libavformat/raw.c

@@ -344,6 +344,21 @@ AVInputFormat h261_iformat = {
     .value = CODEC_ID_H261,
 };
 
+#ifdef CONFIG_ENCODERS
+AVOutputFormat h261_oformat = {
+    "h261",
+    "raw h261",
+    "video/x-h261",
+    "h261",
+    0,
+    0,
+    CODEC_ID_H261,
+    raw_write_header,
+    raw_write_packet,
+    raw_write_trailer,
+};
+#endif //CONFIG_ENCODERS
+
 AVInputFormat h263_iformat = {
     "h263",
     "raw h263",
@@ -648,6 +663,7 @@ int raw_init(void)
     av_register_input_format(&dts_iformat);
 
     av_register_input_format(&h261_iformat);
+    av_register_output_format(&h261_oformat);
 
     av_register_input_format(&h263_iformat);
     av_register_output_format(&h263_oformat);