[x264-devel] Re: x264 improvement

[Andrea Barbieri]

Hello,

I have attached two patch files of the MSU start_qp proposed change...

best regards
andrea
-- 
Andrea Barbieri
KeyID=0x034DFD5A
KeyFingerprint=C1 68 EA 9A 71 89 53 8D  21 4F 12 81 A7 52 9F 32  03 4D FD 5A
Moving Image Research, The Workshop, Hampton Lane, Bristol, BS6 6LE, UK
Tel +44 117 9732200, FAX +44 117 9732210
http://www.movingimageresearch.com/
-------------- next part --------------
*** x264/x264.h	Mon Mar  6 10:03:52 2006
--- x264_new/x264.h	Fri Mar 10 12:24:14 2006
***************
*** 215,220 ****
--- 215,221 ----
          int         i_qp_min;       /* min allowed QP value */
          int         i_qp_max;       /* max allowed QP value */
          int         i_qp_step;      /* max QP step between frames */
+         int         i_qp_start;
  
          int         b_cbr;          /* use bitrate instead of CQP */
          int         i_bitrate;
-------------- next part --------------
*** x264/encoder/ratecontrol.c	Fri Mar 10 10:19:25 2006
--- x264_new/encoder/ratecontrol.c	Fri Mar 10 12:27:54 2006
***************
*** 254,262 ****
  
      if( rc->b_abr )
      {
          /* FIXME shouldn't need to arbitrarily specify a QP,
           * but this is more robust than BPP measures */
! #define ABR_INIT_QP ( h->param.rc.i_rf_constant > 0 ? h->param.rc.i_rf_constant : 24 )
          rc->accum_p_norm = .01;
          rc->accum_p_qp = ABR_INIT_QP * rc->accum_p_norm;
          rc->cplxr_sum = .01;
--- 254,263 ----
  
      if( rc->b_abr )
      {
+         h->param.rc.i_qp_start = (int) (-0.287 * exp ( 0.501 * log ( h->param.rc.i_bitrate ) ) + 39.148);
          /* FIXME shouldn't need to arbitrarily specify a QP,
           * but this is more robust than BPP measures */
! #define ABR_INIT_QP ( h->param.rc.i_rf_constant > 0 ? h->param.rc.i_rf_constant : h->param.rc.i_qp_start )
          rc->accum_p_norm = .01;
          rc->accum_p_qp = ABR_INIT_QP * rc->accum_p_norm;
          rc->cplxr_sum = .01;