]> git.sesse.net Git - x264/blobdiff - encoder/slicetype.c
Hotfix for high bit depth
[x264] / encoder / slicetype.c
index 84a82de114adf08c82b203ec3d3aed577ff8f60e..894c9f4c9fdd971c74bff90b534e4cf1898e818f 100644 (file)
@@ -1,7 +1,7 @@
 /*****************************************************************************
- * slicetype.c: h264 encoder library
+ * slicetype.c: lookahead analysis
  *****************************************************************************
- * Copyright (C) 2005-2008 x264 project
+ * Copyright (C) 2005-2010 x264 project
  *
  * Authors: Fiona Glaser <fiona@x264.com>
  *          Loren Merritt <lorenm@u.washington.edu>
@@ -20,6 +20,9 @@
  * You should have received a copy of the GNU General Public License
  * along with this program; if not, write to the Free Software
  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02111, USA.
+ *
+ * This program is also available under a commercial proprietary license.
+ * For more information, contact us at licensing@x264.com.
  *****************************************************************************/
 
 #include <math.h>
@@ -54,7 +57,7 @@ static void x264_lowres_context_init( x264_t *h, x264_mb_analysis_t *a )
 }
 
 /* makes a non-h264 weight (i.e. fix7), into an h264 weight */
-static void x264_weight_get_h264( unsigned int weight_nonh264, int offset, x264_weight_t *w )
+static void x264_weight_get_h264( int weight_nonh264, int offset, x264_weight_t *w )
 {
     w->i_offset = offset;
     w->i_denom = 7;
@@ -95,14 +98,68 @@ static NOINLINE pixel *x264_weight_cost_init_luma( x264_t *h, x264_frame_t *fenc
     return ref->lowres[0];
 }
 
-static NOINLINE unsigned int x264_weight_cost( x264_t *h, x264_frame_t *fenc, pixel *src, x264_weight_t *w )
+/* How data is organized for chroma weightp:
+ * [U: ref] [U: fenc]
+ * [V: ref] [V: fenc]
+ * fenc = ref + offset
+ * v = u + stride * chroma height
+ * We'll need more room if we do 4:2:2 or 4:4:4. */
+
+static NOINLINE void x264_weight_cost_init_chroma( x264_t *h, x264_frame_t *fenc, x264_frame_t *ref, pixel *dstu, pixel *dstv )
+{
+    int ref0_distance = fenc->i_frame - ref->i_frame - 1;
+    int i_stride = fenc->i_stride[1];
+    int i_offset = i_stride / 2;
+    int i_lines = fenc->i_lines[1];
+    int i_width = fenc->i_width[1];
+    int cw = h->mb.i_mb_width  << 3;
+    int ch = h->mb.i_mb_height << 3;
+
+    if( fenc->lowres_mvs[0][ref0_distance][0][0] != 0x7FFF )
+    {
+        for( int y = 0, mb_xy = 0, pel_offset_y = 0; y < i_lines; y += 8, pel_offset_y = y*i_stride )
+            for( int x = 0, pel_offset_x = 0; x < i_width; x += 8, mb_xy++, pel_offset_x += 8 )
+            {
+                pixel *pixu = dstu + pel_offset_y + pel_offset_x;
+                pixel *pixv = dstv + pel_offset_y + pel_offset_x;
+                pixel *src1 =  ref->plane[1] + pel_offset_y + pel_offset_x*2; /* NV12 */
+                int mvx = fenc->lowres_mvs[0][ref0_distance][mb_xy][0];
+                int mvy = fenc->lowres_mvs[0][ref0_distance][mb_xy][1];
+                h->mc.mc_chroma( pixu, pixv, i_stride, src1, i_stride, mvx, mvy, 8, 8 );
+            }
+    }
+    else
+        h->mc.plane_copy_deinterleave( dstu, i_stride, dstv, i_stride, ref->plane[1], i_stride, cw, ch );
+    h->mc.plane_copy_deinterleave( dstu+i_offset, i_stride, dstv+i_offset, i_stride, fenc->plane[1], i_stride, cw, ch );
+    x264_emms();
+}
+
+static int x264_weight_slice_header_cost( x264_t *h, x264_weight_t *w, int b_chroma )
+{
+    /* Add cost of weights in the slice header. */
+    int lambda = x264_lambda_tab[X264_LOOKAHEAD_QP];
+    int numslices;
+    if( h->param.i_slice_count )
+        numslices = h->param.i_slice_count;
+    else if( h->param.i_slice_max_mbs )
+        numslices = (h->mb.i_mb_width * h->mb.i_mb_height + h->param.i_slice_max_mbs-1) / h->param.i_slice_max_mbs;
+    else
+        numslices = 1;
+    /* FIXME: find a way to account for --slice-max-size?
+     * Multiply by 2 as there will be a duplicate. 10 bits added as if there is a weighted frame, then an additional duplicate is used.
+     * Cut denom cost in half if chroma, since it's shared between the two chroma planes. */
+    int denom_cost = bs_size_ue( w[0].i_denom ) * (2 - b_chroma);
+    return lambda * numslices * ( 10 + denom_cost + 2 * (bs_size_se( w[0].i_scale ) + bs_size_se( w[0].i_offset )) );
+}
+
+static NOINLINE unsigned int x264_weight_cost_luma( x264_t *h, x264_frame_t *fenc, pixel *src, x264_weight_t *w )
 {
     unsigned int cost = 0;
     int i_stride = fenc->i_stride_lowres;
     int i_lines = fenc->i_lines_lowres;
     int i_width = fenc->i_width_lowres;
     pixel *fenc_plane = fenc->lowres[0];
-    ALIGNED_ARRAY_8( pixel, buf,[8*8] );
+    ALIGNED_ARRAY_16( pixel, buf,[8*8] );
     int pixoff = 0;
     int i_mb = 0;
 
@@ -114,18 +171,7 @@ static NOINLINE unsigned int x264_weight_cost( x264_t *h, x264_frame_t *fenc, pi
                 w->weightfn[8>>2]( buf, 8, &src[pixoff], i_stride, w, 8 );
                 cost += X264_MIN( h->pixf.mbcmp[PIXEL_8x8]( buf, 8, &fenc_plane[pixoff], i_stride ), fenc->i_intra_cost[i_mb] );
             }
-        /* Add cost of weights in the slice header. */
-        int numslices;
-        if( h->param.i_slice_count )
-            numslices = h->param.i_slice_count;
-        else if( h->param.i_slice_max_mbs )
-            numslices = (h->mb.i_mb_width * h->mb.i_mb_height + h->param.i_slice_max_mbs-1) / h->param.i_slice_max_mbs;
-        else
-            numslices = 1;
-        /* FIXME: find a way to account for --slice-max-size?
-         * Multiply by 2 as there will be a duplicate. 10 bits added as if there is a weighted frame, then an additional duplicate is used.
-         * Since using lowres frames, assume lambda = 1. */
-        cost += numslices * ( 10 + 2 * ( bs_size_ue( w[0].i_denom ) + bs_size_se( w[0].i_scale ) + bs_size_se( w[0].i_offset ) ) );
+        cost += x264_weight_slice_header_cost( h, w, 0 );
     }
     else
         for( int y = 0; y < i_lines; y += 8, pixoff = y*i_stride )
@@ -135,78 +181,159 @@ static NOINLINE unsigned int x264_weight_cost( x264_t *h, x264_frame_t *fenc, pi
     return cost;
 }
 
+static NOINLINE unsigned int x264_weight_cost_chroma( x264_t *h, x264_frame_t *fenc, pixel *ref, x264_weight_t *w )
+{
+    unsigned int cost = 0;
+    int i_stride = fenc->i_stride[1];
+    int i_offset = i_stride / 2;
+    int i_lines = fenc->i_lines[1];
+    int i_width = fenc->i_width[1];
+    pixel *src = ref + i_offset;
+    ALIGNED_ARRAY_16( pixel, buf, [8*8] );
+    int pixoff = 0;
+    ALIGNED_16( static pixel flat[9] ) = {0,0,0,0,0,0,0,0,1}; //hack for win32
+    if( w )
+    {
+        for( int y = 0; y < i_lines; y += 8, pixoff = y*i_stride )
+            for( int x = 0; x < i_width; x += 8, pixoff += 8 )
+            {
+                w->weightfn[8>>2]( buf, 8, &ref[pixoff], i_stride, w, 8 );
+                /* The naive and seemingly sensible algorithm is to use mbcmp as in luma.
+                 * But testing shows that for chroma the DC coefficient is by far the most
+                 * important part of the coding cost.  Thus a more useful chroma weight is
+                 * obtained by comparing each block's DC coefficient instead of the actual
+                 * pixels.
+                 *
+                 * FIXME: add a (faster) asm sum function to replace sad. */
+                cost += abs( h->pixf.sad_aligned[PIXEL_8x8](          buf,        8, flat, 0 ) -
+                             h->pixf.sad_aligned[PIXEL_8x8]( &src[pixoff], i_stride, flat, 0 ) );
+            }
+        cost += x264_weight_slice_header_cost( h, w, 1 );
+    }
+    else
+        for( int y = 0; y < i_lines; y += 8, pixoff = y*i_stride )
+            for( int x = 0; x < i_width; x += 8, pixoff += 8 )
+                cost += abs( h->pixf.sad_aligned[PIXEL_8x8]( &ref[pixoff], i_stride, flat, 0 ) -
+                             h->pixf.sad_aligned[PIXEL_8x8]( &src[pixoff], i_stride, flat, 0 ) );
+    x264_emms();
+    return cost;
+}
+
 void x264_weights_analyse( x264_t *h, x264_frame_t *fenc, x264_frame_t *ref, int b_lookahead )
 {
-    float fenc_mean, ref_mean, fenc_var, ref_var;
-    int offset_search;
-    int minoff, minscale, mindenom;
-    unsigned int minscore, origscore;
     int i_delta_index = fenc->i_frame - ref->i_frame - 1;
     /* epsilon is chosen to require at least a numerator of 127 (with denominator = 128) */
-    const float epsilon = 1.0/128.0;
-    float guess_scale;
-    int found;
+    const float epsilon = 1.f/128.f;
     x264_weight_t *weights = fenc->weight[0];
+    SET_WEIGHT( weights[0], 0, 1, 0, 0 );
+    SET_WEIGHT( weights[1], 0, 1, 0, 0 );
+    SET_WEIGHT( weights[2], 0, 1, 0, 0 );
+    /* Don't check chroma in lookahead, or if there wasn't a luma weight. */
+    for( int plane = 0; plane <= 2  && !( plane && ( !weights[0].weightfn || b_lookahead ) ); plane++ )
+    {
+        int offset_search;
+        int minoff, minscale, mindenom;
+        unsigned int minscore, origscore;
+        int found;
+        float fenc_var = fenc->i_pixel_ssd[plane] + !ref->i_pixel_ssd[plane];
+        float ref_var  =  ref->i_pixel_ssd[plane] + !ref->i_pixel_ssd[plane];
+        float guess_scale = sqrtf( fenc_var / ref_var );
+        float fenc_mean = (float)fenc->i_pixel_sum[plane] / (fenc->i_lines[!!plane] * fenc->i_width[!!plane]);
+        float ref_mean  = (float) ref->i_pixel_sum[plane] / (fenc->i_lines[!!plane] * fenc->i_width[!!plane]);
+
+        //early termination
+        if( fabsf( ref_mean - fenc_mean ) < 0.5f && fabsf( 1.f - guess_scale ) < epsilon )
+        {
+            SET_WEIGHT( weights[plane], 0, 1, 0, 0 );
+            continue;
+        }
 
-    fenc_var = round( sqrt( fenc->i_pixel_ssd[0] ) );
-    ref_var  = round( sqrt(  ref->i_pixel_ssd[0] ) );
-    fenc_mean = (float)fenc->i_pixel_sum[0] / (fenc->i_lines[0] * fenc->i_width[0]);
-    ref_mean  = (float) ref->i_pixel_sum[0] / (fenc->i_lines[0] * fenc->i_width[0]);
+        if( plane )
+        {
+            weights[plane].i_denom = 6;
+            weights[plane].i_scale = x264_clip3( round( guess_scale * 64 ), 0, 255 );
+            if( weights[plane].i_scale > 127 )
+            {
+                weights[1].weightfn = weights[2].weightfn = NULL;
+                break;
+            }
+        }
+        else
+            x264_weight_get_h264( round( guess_scale * 128 ), 0, &weights[plane] );
 
-    //early termination
-    if( fabs( ref_mean - fenc_mean ) < 0.5 && fabs( 1 - fenc_var / ref_var ) < epsilon )
-    {
-        SET_WEIGHT( weights[0], 0, 1, 0, 0 );
-        return;
-    }
+        found = 0;
+        mindenom = weights[plane].i_denom;
+        minscale = weights[plane].i_scale;
+        minoff = 0;
 
-    guess_scale = ref_var ? fenc_var/ref_var : 0;
-    x264_weight_get_h264( round( guess_scale * 128 ), 0, &weights[0] );
+        pixel *mcbuf;
+        if( !plane )
+        {
+            if( !fenc->b_intra_calculated )
+            {
+                x264_mb_analysis_t a;
+                x264_lowres_context_init( h, &a );
+                x264_slicetype_frame_cost( h, &a, &fenc, 0, 0, 0, 0 );
+            }
+            mcbuf = x264_weight_cost_init_luma( h, fenc, ref, h->mb.p_weight_buf[0] );
+            origscore = minscore = x264_weight_cost_luma( h, fenc, mcbuf, NULL );
+        }
+        else
+        {
+            pixel *dstu = h->mb.p_weight_buf[0];
+            pixel *dstv = h->mb.p_weight_buf[0]+fenc->i_stride[1]*fenc->i_lines[1];
+            /* Only initialize chroma data once. */
+            if( plane == 1 )
+                x264_weight_cost_init_chroma( h, fenc, ref, dstu, dstv );
+            mcbuf = plane == 1 ? dstu : dstv;
+            origscore = minscore = x264_weight_cost_chroma( h, fenc, mcbuf, NULL );
+        }
 
-    found = 0;
-    mindenom = weights[0].i_denom;
-    minscale = weights[0].i_scale;
-    minoff = 0;
-    offset_search = x264_clip3( floor( fenc_mean - ref_mean * minscale / (1 << mindenom) + 0.5f*b_lookahead ), -128, 126 );
+        if( !minscore )
+            continue;
 
-    if( !fenc->b_intra_calculated )
-    {
-        x264_mb_analysis_t a;
-        x264_lowres_context_init( h, &a );
-        x264_slicetype_frame_cost( h, &a, &fenc, 0, 0, 0, 0 );
-    }
-    pixel *mcbuf = x264_weight_cost_init_luma( h, fenc, ref, h->mb.p_weight_buf[0] );
-    origscore = minscore = x264_weight_cost( h, fenc, mcbuf, 0 );
+        // This gives a slight improvement due to rounding errors but only tests
+        // one offset on lookahead.
+        // TODO: currently searches only offset +1. try other offsets/multipliers/combinations thereof?
+        offset_search = x264_clip3( fenc_mean - ref_mean * minscale / (1 << mindenom) + 0.5f * b_lookahead, -128, 126 );
+        for( int i_off = offset_search; i_off <= offset_search+!b_lookahead; i_off++ )
+        {
+            SET_WEIGHT( weights[plane], 1, minscale, mindenom, i_off );
+            unsigned int s;
+            if( plane )
+                s = x264_weight_cost_chroma( h, fenc, mcbuf, &weights[plane] );
+            else
+                s = x264_weight_cost_luma( h, fenc, mcbuf, &weights[plane] );
+            COPY3_IF_LT( minscore, s, minoff, i_off, found, 1 );
+        }
+        x264_emms();
 
-    if( !minscore )
-    {
-        SET_WEIGHT( weights[0], 0, 1, 0, 0 );
-        return;
-    }
+        /* FIXME: More analysis can be done here on SAD vs. SATD termination. */
+        /* 0.2% termination derived experimentally to avoid weird weights in frames that are mostly intra. */
+        if( !found || (minscale == 1 << mindenom && minoff == 0) || (float)minscore / origscore > 0.998f )
+        {
+            SET_WEIGHT( weights[plane], 0, 1, 0, 0 );
+            continue;
+        }
+        else
+            SET_WEIGHT( weights[plane], 1, minscale, mindenom, minoff );
 
-    // This gives a slight improvement due to rounding errors but only tests
-    // one offset on lookahead.
-    // TODO: currently searches only offset +1. try other offsets/multipliers/combinations thereof?
-    for( int i_off = offset_search; i_off <= offset_search+!b_lookahead; i_off++ )
-    {
-        SET_WEIGHT( weights[0], 1, minscale, mindenom, i_off );
-        unsigned int s = x264_weight_cost( h, fenc, mcbuf, &weights[0] );
-        COPY3_IF_LT( minscore, s, minoff, i_off, found, 1 );
+        if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_FAKE && weights[0].weightfn && !plane )
+            fenc->f_weighted_cost_delta[i_delta_index] = (float)minscore / origscore;
     }
-    x264_emms();
 
-    /* FIXME: More analysis can be done here on SAD vs. SATD termination. */
-    /* 0.2% termination derived experimentally to avoid weird weights in frames that are mostly intra. */
-    if( !found || (minscale == 1<<mindenom && minoff == 0) || (float)minscore / origscore > 0.998 )
+    //FIXME, what is the correct way to deal with this?
+    if( weights[1].weightfn && weights[2].weightfn && weights[1].i_denom != weights[2].i_denom )
     {
-        SET_WEIGHT( weights[0], 0, 1, 0, 0 );
-        return;
+        int denom = X264_MIN( weights[1].i_denom, weights[2].i_denom );
+        int i;
+        for( i = 1; i <= 2; i++ )
+        {
+            weights[i].i_scale = x264_clip3( weights[i].i_scale >> ( weights[i].i_denom - denom ), 0, 255 );
+            weights[i].i_denom = denom;
+            h->mc.weight_cache( h, &weights[i] );
+        }
     }
-    else
-        SET_WEIGHT( weights[0], 1, minscale, mindenom, minoff );
-
-    if( h->param.analyse.i_weighted_pred == X264_WEIGHTP_FAKE && weights[0].weightfn )
-        fenc->f_weighted_cost_delta[i_delta_index] = (float)minscore / origscore;
 
     if( weights[0].weightfn && b_lookahead )
     {
@@ -242,7 +369,7 @@ static void x264_slicetype_mb_cost( x264_t *h, x264_mb_analysis_t *a,
                             i_mb_y > 0 && i_mb_y < h->mb.i_mb_height - 1) ||
                             h->mb.i_mb_width <= 2 || h->mb.i_mb_height <= 2;
 
-    ALIGNED_ARRAY_8( pixel, pix1,[9*FDEC_STRIDE] );
+    ALIGNED_ARRAY_16( pixel, pix1,[9*FDEC_STRIDE] );
     pixel *pix2 = pix1+8;
     x264_me_t m[2];
     int i_bcost = COST_MAX;
@@ -303,7 +430,7 @@ static void x264_slicetype_mb_cost( x264_t *h, x264_mb_analysis_t *a,
                                   (mv1)[0], (mv1)[1], 8, 8, w ); \
             h->mc.avg[PIXEL_8x8]( pix1, 16, src1, stride1, src2, stride2, i_bipred_weight ); \
         } \
-        i_cost = penalty + h->pixf.mbcmp[PIXEL_8x8]( \
+        i_cost = penalty * a->i_lambda + h->pixf.mbcmp[PIXEL_8x8]( \
                            m[0].p_fenc[0], FENC_STRIDE, pix1, 16 ); \
         COPY2_IF_LT( i_bcost, i_cost, list_used, 3 ); \
     }
@@ -393,9 +520,9 @@ static void x264_slicetype_mb_cost( x264_t *h, x264_mb_analysis_t *a,
             }
 
             x264_me_search( h, &m[l], mvc, i_mvc );
-            m[l].cost -= 2; // remove mvcost from skip mbs
+            m[l].cost -= 2 * a->i_lambda; // remove mvcost from skip mbs
             if( M32( m[l].mv ) )
-                m[l].cost += 5;
+                m[l].cost += 5 * a->i_lambda;
 
 skip_motionest:
             CP32( fenc_mvs[l], m[l].mv );
@@ -418,7 +545,7 @@ lowres_intra_mb:
         ALIGNED_ARRAY_16( pixel, edge,[33] );
         pixel *pix = &pix1[8+FDEC_STRIDE - 1];
         pixel *src = &fenc->lowres[0][i_pel_offset - 1];
-        const int intra_penalty = 5;
+        const int intra_penalty = 5 * a->i_lambda;
         int satds[3];
 
         memcpy( pix-FDEC_STRIDE, src-i_stride, 17 * sizeof(pixel) );
@@ -426,16 +553,7 @@ lowres_intra_mb:
             pix[i*FDEC_STRIDE] = src[i*i_stride];
         pix++;
 
-        if( h->pixf.intra_mbcmp_x3_8x8c )
-            h->pixf.intra_mbcmp_x3_8x8c( h->mb.pic.p_fenc[0], pix, satds );
-        else
-        {
-            for( int i = 0; i < 3; i++ )
-            {
-                h->predict_8x8c[i]( pix );
-                satds[i] = h->pixf.mbcmp[PIXEL_8x8]( pix, FDEC_STRIDE, h->mb.pic.p_fenc[0], FENC_STRIDE );
-            }
-        }
+        h->pixf.intra_mbcmp_x3_8x8c( h->mb.pic.p_fenc[0], pix, satds );
         int i_icost = X264_MIN3( satds[0], satds[1], satds[2] );
 
         if( h->param.analyse.i_subpel_refine > 1 )
@@ -496,7 +614,7 @@ lowres_intra_mb:
         }
     }
 
-    fenc->lowres_costs[b-p0][p1-b][i_mb_xy] = i_bcost + (list_used << LOWRES_COST_SHIFT);
+    fenc->lowres_costs[b-p0][p1-b][i_mb_xy] = X264_MIN( i_bcost, LOWRES_COST_MASK ) + (list_used << LOWRES_COST_SHIFT);
 }
 #undef TRY_BIDIR
 
@@ -529,8 +647,7 @@ static int x264_slicetype_frame_cost( x264_t *h, x264_mb_analysis_t *a,
         do_search[1] = b != p1 && frames[b]->lowres_mvs[1][p1-b-1][0][0] == 0x7FFF;
         if( do_search[0] )
         {
-            if( ( h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART ||
-                  h->param.analyse.i_weighted_pred == X264_WEIGHTP_FAKE ) && b == p1 )
+            if( h->param.analyse.i_weighted_pred && b == p1 )
             {
                 x264_emms();
                 x264_weights_analyse( h, frames[b], frames[p0], 1 );
@@ -1008,13 +1125,12 @@ static int scenecut_internal( x264_t *h, x264_mb_analysis_t *a, x264_frame_t **f
     int i_gop_size = frame->i_frame - h->lookahead->i_last_keyframe;
     float f_thresh_max = h->param.i_scenecut_threshold / 100.0;
     /* magic numbers pulled out of thin air */
-    float f_thresh_min = f_thresh_max * h->param.i_keyint_min
-                         / ( h->param.i_keyint_max * 4. );
+    float f_thresh_min = f_thresh_max * 0.25;
     int res;
 
     if( h->param.i_keyint_min == h->param.i_keyint_max )
-        f_thresh_min= f_thresh_max;
-    if( i_gop_size < h->param.i_keyint_min / 4 || h->param.b_intra_refresh )
+        f_thresh_min = f_thresh_max;
+    if( i_gop_size <= h->param.i_keyint_min / 4 || h->param.b_intra_refresh )
         f_bias = f_thresh_min / 4;
     else if( i_gop_size <= h->param.i_keyint_min )
         f_bias = f_thresh_min * i_gop_size / h->param.i_keyint_min;
@@ -1022,8 +1138,8 @@ static int scenecut_internal( x264_t *h, x264_mb_analysis_t *a, x264_frame_t **f
     {
         f_bias = f_thresh_min
                  + ( f_thresh_max - f_thresh_min )
-                    * ( i_gop_size - h->param.i_keyint_min )
-                   / ( h->param.i_keyint_max - h->param.i_keyint_min ) ;
+                 * ( i_gop_size - h->param.i_keyint_min )
+                 / ( h->param.i_keyint_max - h->param.i_keyint_min );
     }
 
     res = pcost >= (1.0 - f_bias) * icost;
@@ -1039,19 +1155,19 @@ static int scenecut_internal( x264_t *h, x264_mb_analysis_t *a, x264_frame_t **f
     return res;
 }
 
-static int scenecut( x264_t *h, x264_mb_analysis_t *a, x264_frame_t **frames, int p0, int p1, int real_scenecut, int num_frames )
+static int scenecut( x264_t *h, x264_mb_analysis_t *a, x264_frame_t **frames, int p0, int p1, int real_scenecut, int num_frames, int i_max_search )
 {
     /* Only do analysis during a normal scenecut check. */
     if( real_scenecut && h->param.i_bframe )
     {
-        int maxp1 = p0 + 1;
+        int origmaxp1 = p0 + 1;
         /* Look ahead to avoid coding short flashes as scenecuts. */
         if( h->param.i_bframe_adaptive == X264_B_ADAPT_TRELLIS )
             /* Don't analyse any more frames than the trellis would have covered. */
-            maxp1 += h->param.i_bframe;
+            origmaxp1 += h->param.i_bframe;
         else
-            maxp1++;
-        maxp1 = X264_MIN( maxp1, num_frames );
+            origmaxp1++;
+        int maxp1 = X264_MIN( origmaxp1, num_frames );
 
         /* Where A and B are scenes: AAAAAABBBAAAAAA
          * If BBB is shorter than (maxp1-p0), it is detected as a flash
@@ -1065,9 +1181,10 @@ static int scenecut( x264_t *h, x264_mb_analysis_t *a, x264_frame_t **frames, in
         /* Where A-F are scenes: AAAAABBCCDDEEFFFFFF
          * If each of BB ... EE are shorter than (maxp1-p0), they are
          * detected as flashes and not considered scenecuts.
-         * Instead, the first F frame becomes a scenecut. */
-        for( int curp0 = p0; curp0 < maxp1; curp0++ )
-            if( scenecut_internal( h, a, frames, curp0, maxp1, 0 ) )
+         * Instead, the first F frame becomes a scenecut.
+         * If the video ends before F, no frame becomes a scenecut. */
+        for( int curp0 = p0; curp0 <= maxp1; curp0++ )
+            if( origmaxp1 > i_max_search || (curp0 < maxp1 && scenecut_internal( h, a, frames, curp0, maxp1, 0 )) )
                 /* If cur_p0 is the p0 of a scenecut, it cannot be the p1 of a scenecut. */
                     frames[curp0]->b_scenecut = 0;
     }
@@ -1125,7 +1242,7 @@ void x264_slicetype_analyse( x264_t *h, int keyframe )
     int num_bframes = 0;
     int num_analysed_frames = num_frames;
     int reset_start;
-    if( h->param.i_scenecut_threshold && scenecut( h, &a, frames, 0, 1, 1, orig_num_frames ) )
+    if( h->param.i_scenecut_threshold && scenecut( h, &a, frames, 0, 1, 1, orig_num_frames, i_max_search ) )
     {
         frames[1]->i_type = X264_TYPE_I;
         return;
@@ -1207,7 +1324,7 @@ void x264_slicetype_analyse( x264_t *h, int keyframe )
 
         /* Check scenecut on the first minigop. */
         for( int j = 1; j < num_bframes+1; j++ )
-            if( h->param.i_scenecut_threshold && scenecut( h, &a, frames, j, j+1, 0, orig_num_frames ) )
+            if( h->param.i_scenecut_threshold && scenecut( h, &a, frames, j, j+1, 0, orig_num_frames, i_max_search ) )
             {
                 frames[j]->i_type = X264_TYPE_P;
                 num_analysed_frames = j;
@@ -1260,32 +1377,32 @@ void x264_slicetype_decide( x264_t *h )
 
     int lookahead_size = h->lookahead->next.i_size;
 
-    if( h->param.rc.i_rc_method == X264_RC_ABR || h->param.rc.b_stat_write || h->param.rc.i_vbv_buffer_size )
+    for( int i = 0; i < h->lookahead->next.i_size; i++ )
     {
-        for( int i = 0; i < h->lookahead->next.i_size; i++ )
+        if( h->param.b_vfr_input )
         {
-            if( h->param.b_vfr_input )
-            {
-                if( lookahead_size-- > 1 )
-                    h->lookahead->next.list[i]->i_duration = 2 * (h->lookahead->next.list[i+1]->i_pts - h->lookahead->next.list[i]->i_pts);
-                else
-                    h->lookahead->next.list[i]->i_duration = h->i_prev_duration;
-            }
+            if( lookahead_size-- > 1 )
+                h->lookahead->next.list[i]->i_duration = 2 * (h->lookahead->next.list[i+1]->i_pts - h->lookahead->next.list[i]->i_pts);
             else
-                h->lookahead->next.list[i]->i_duration = delta_tfi_divisor[h->lookahead->next.list[i]->i_pic_struct];
-            h->i_prev_duration = h->lookahead->next.list[i]->i_duration;
-
-            if( h->lookahead->next.list[i]->i_frame > h->i_disp_fields_last_frame && lookahead_size > 0 )
-            {
-                h->lookahead->next.list[i]->i_field_cnt = h->i_disp_fields;
-                h->i_disp_fields += h->lookahead->next.list[i]->i_duration;
-                h->i_disp_fields_last_frame = h->lookahead->next.list[i]->i_frame;
-            }
-            else if( lookahead_size == 0 )
-            {
-                h->lookahead->next.list[i]->i_field_cnt = h->i_disp_fields;
                 h->lookahead->next.list[i]->i_duration = h->i_prev_duration;
-            }
+        }
+        else
+            h->lookahead->next.list[i]->i_duration = delta_tfi_divisor[h->lookahead->next.list[i]->i_pic_struct];
+        h->i_prev_duration = h->lookahead->next.list[i]->i_duration;
+        h->lookahead->next.list[i]->f_duration = (double)h->lookahead->next.list[i]->i_duration
+                                               * h->sps->vui.i_num_units_in_tick
+                                               / h->sps->vui.i_time_scale;
+
+        if( h->lookahead->next.list[i]->i_frame > h->i_disp_fields_last_frame && lookahead_size > 0 )
+        {
+            h->lookahead->next.list[i]->i_field_cnt = h->i_disp_fields;
+            h->i_disp_fields += h->lookahead->next.list[i]->i_duration;
+            h->i_disp_fields_last_frame = h->lookahead->next.list[i]->i_frame;
+        }
+        else if( lookahead_size == 0 )
+        {
+            h->lookahead->next.list[i]->i_field_cnt = h->i_disp_fields;
+            h->lookahead->next.list[i]->i_duration = h->i_prev_duration;
         }
     }
 
@@ -1431,7 +1548,7 @@ void x264_slicetype_decide( x264_t *h )
 
     /* Analyse for weighted P frames */
     if( !h->param.rc.b_stat_read && h->lookahead->next.list[bframes]->i_type == X264_TYPE_P
-        && h->param.analyse.i_weighted_pred == X264_WEIGHTP_SMART )
+        && h->param.analyse.i_weighted_pred >= X264_WEIGHTP_SIMPLE )
     {
         x264_emms();
         x264_weights_analyse( h, h->lookahead->next.list[bframes], h->lookahead->last_nonb, 0 );
@@ -1457,20 +1574,17 @@ void x264_slicetype_decide( x264_t *h )
     for( int i = 0; i <= bframes; i++ )
     {
         h->lookahead->next.list[i]->i_coded = i_coded++;
-        if( h->param.rc.i_rc_method == X264_RC_ABR || h->param.rc.b_stat_write || h->param.rc.i_vbv_buffer_size )
+        if( i )
         {
-            if( i )
-            {
-                x264_calculate_durations( h, h->lookahead->next.list[i], h->lookahead->next.list[i-1], &h->i_cpb_delay, &h->i_coded_fields );
-                h->lookahead->next.list[0]->f_planned_cpb_duration[i-1] = (double)h->lookahead->next.list[i-1]->i_cpb_duration *
-                                                                          h->sps->vui.i_num_units_in_tick / h->sps->vui.i_time_scale;
-            }
-            else
-                x264_calculate_durations( h, h->lookahead->next.list[i], NULL, &h->i_cpb_delay, &h->i_coded_fields );
-
-            h->lookahead->next.list[0]->f_planned_cpb_duration[i] = (double)h->lookahead->next.list[i]->i_cpb_duration *
-                                                                    h->sps->vui.i_num_units_in_tick / h->sps->vui.i_time_scale;
+            x264_calculate_durations( h, h->lookahead->next.list[i], h->lookahead->next.list[i-1], &h->i_cpb_delay, &h->i_coded_fields );
+            h->lookahead->next.list[0]->f_planned_cpb_duration[i-1] = (double)h->lookahead->next.list[i-1]->i_cpb_duration *
+                                                                      h->sps->vui.i_num_units_in_tick / h->sps->vui.i_time_scale;
         }
+        else
+            x264_calculate_durations( h, h->lookahead->next.list[i], NULL, &h->i_cpb_delay, &h->i_coded_fields );
+
+        h->lookahead->next.list[0]->f_planned_cpb_duration[i] = (double)h->lookahead->next.list[i]->i_cpb_duration *
+                                                                h->sps->vui.i_num_units_in_tick / h->sps->vui.i_time_scale;
     }
 }