Mercurial > mplayer.hg
view libaf/af_resample.h @ 26815:fd3b28579cee
r26675: update paragraphs related to x264, and update its checkout command
r26729: MPlayer uses Subversion, not GIT, 10L to me, and thanks to Mizda for spotting this
r26287: remove excessive space character
r26451: As of r19025, the "above link" refers to an article, not a guide.
r26452: Refer to where encoding quality is described.
r26453: typo: crahes --> crashes
r26474: add better information about inverse-telecining with vf_filmdint
r26711: Consistency fix: all DVD encoding examples had ":aspect=16/9" option, so put
r26258: fix typo: lavcoptc --> lavcopts
author | voroshil |
---|---|
date | Wed, 21 May 2008 17:12:57 +0000 |
parents | 32e21d1beb48 |
children |
line wrap: on
line source
/*============================================================================= // // This software has been released under the terms of the GNU General Public // license. See http://www.gnu.org/copyleft/gpl.html for details. // // Copyright 2002 Anders Johansson ajh@atri.curtin.edu.au // //============================================================================= */ /* This file contains the resampling engine, the sample format is controlled by the FORMAT parameter, the filter length by the L parameter and the resampling type by UP and DN. This file should only be included by af_resample.c */ /* This header intentionally has no multiple inclusion guards. It is meant to * be included multiple times and generates different code depending on the * value of certain #defines. */ #undef L #undef SHIFT #undef FORMAT #undef FIR #undef ADDQUE /* The length Lxx definition selects the length of each poly phase component. Valid definitions are L8 and L16 where the number defines the nuber of taps. This definition affects the computational complexity, the performance and the memory usage. */ /* The FORMAT_x parameter selects the sample format type currently float and int16 are supported. Thes two formats are selected by defining eiter FORMAT_F or FORMAT_I. The advantage of using float is that the amplitude and therefore the SNR isn't affected by the filtering, the disadvantage is that it is a lot slower. */ #if defined(FORMAT_I) #define SHIFT >>16 #define FORMAT int16_t #else #define SHIFT #define FORMAT float #endif // Short filter #if defined(L8) #define L 8 // Filter length // Unrolled loop to speed up execution #define FIR(x,w,y) \ (y[0]) = ( w[0]*x[0]+w[1]*x[1]+w[2]*x[2]+w[3]*x[3] \ + w[4]*x[4]+w[5]*x[5]+w[6]*x[6]+w[7]*x[7] ) SHIFT #else /* L8/L16 */ #define L 16 // Unrolled loop to speed up execution #define FIR(x,w,y) \ y[0] = ( w[0] *x[0] +w[1] *x[1] +w[2] *x[2] +w[3] *x[3] \ + w[4] *x[4] +w[5] *x[5] +w[6] *x[6] +w[7] *x[7] \ + w[8] *x[8] +w[9] *x[9] +w[10]*x[10]+w[11]*x[11] \ + w[12]*x[12]+w[13]*x[13]+w[14]*x[14]+w[15]*x[15] ) SHIFT #endif /* L8/L16 */ // Macro to add data to circular que #define ADDQUE(xi,xq,in)\ xq[xi]=xq[(xi)+L]=*(in);\ xi=((xi)-1)&(L-1); #if defined(UP) uint32_t ci = l->nch; // Index for channels uint32_t nch = l->nch; // Number of channels uint32_t inc = s->up/s->dn; uint32_t level = s->up%s->dn; uint32_t up = s->up; uint32_t dn = s->dn; uint32_t ns = c->len/l->bps; register FORMAT* w = s->w; register uint32_t wi = 0; register uint32_t xi = 0; // Index current channel while(ci--){ // Temporary pointers register FORMAT* x = s->xq[ci]; register FORMAT* in = ((FORMAT*)c->audio)+ci; register FORMAT* out = ((FORMAT*)l->audio)+ci; FORMAT* end = in+ns; // Block loop end wi = s->wi; xi = s->xi; while(in < end){ register uint32_t i = inc; if(wi<level) i++; ADDQUE(xi,x,in); in+=nch; while(i--){ // Run the FIR filter FIR((&x[xi]),(&w[wi*L]),out); len++; out+=nch; // Update wi to point at the correct polyphase component wi=(wi+dn)%up; } } } // Save values that needs to be kept for next time s->wi = wi; s->xi = xi; #endif /* UP */ #if defined(DN) /* DN */ uint32_t ci = l->nch; // Index for channels uint32_t nch = l->nch; // Number of channels uint32_t inc = s->dn/s->up; uint32_t level = s->dn%s->up; uint32_t up = s->up; uint32_t dn = s->dn; uint32_t ns = c->len/l->bps; FORMAT* w = s->w; register int32_t i = 0; register uint32_t wi = 0; register uint32_t xi = 0; // Index current channel while(ci--){ // Temporary pointers register FORMAT* x = s->xq[ci]; register FORMAT* in = ((FORMAT*)c->audio)+ci; register FORMAT* out = ((FORMAT*)l->audio)+ci; register FORMAT* end = in+ns; // Block loop end i = s->i; wi = s->wi; xi = s->xi; while(in < end){ ADDQUE(xi,x,in); in+=nch; if((--i)<=0){ // Run the FIR filter FIR((&x[xi]),(&w[wi*L]),out); len++; out+=nch; // Update wi to point at the correct polyphase component wi=(wi+dn)%up; // Insert i number of new samples in queue i = inc; if(wi<level) i++; } } } // Save values that needs to be kept for next time s->wi = wi; s->xi = xi; s->i = i; #endif /* DN */