sdl_blit_a.c

MPEG-4编解码的实现(包括MPEG4视音频编解码)

/*
    SDL - Simple DirectMedia Layer
    Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002  Sam Lantinga

    This library is free software; you can redistribute it and/or
    modify it under the terms of the GNU Library General Public
    License as published by the Free Software Foundation; either
    version 2 of the License, or (at your option) any later version.

    This library is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
    Library General Public License for more details.

    You should have received a copy of the GNU Library General Public
    License along with this library; if not, write to the Free
    Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA

    Sam Lantinga
    [e-mail]
*/

#ifdef SAVE_RCSID
static char rcsid =
 "@(#) $Id: SDL_blit_A.c,v 1.4 2002/04/22 21:38:03 wmay Exp $";
#endif

#include <stdio.h>

#include "SDL_types.h"
#include "SDL_video.h"
#include "SDL_blit.h"

/* Functions to perform alpha blended blitting */

/* N->1 blending with per-surface alpha */
static void BlitNto1SurfaceAlpha(SDL_BlitInfo *info)
{
	int width = info->d_width;
	int height = info->d_height;
	Uint8 *src = info->s_pixels;
	int srcskip = info->s_skip;
	Uint8 *dst = info->d_pixels;
	int dstskip = info->d_skip;
	Uint8 *palmap = info->table;
	SDL_PixelFormat *srcfmt = info->src;
	SDL_PixelFormat *dstfmt = info->dst;
	int srcbpp = srcfmt->BytesPerPixel;

	const unsigned A = srcfmt->alpha;

	while ( height-- ) {
	    DUFFS_LOOP4(
	    {
		Uint32 pixel;
		unsigned sR;
		unsigned sG;
		unsigned sB;
		unsigned dR;
		unsigned dG;
		unsigned dB;
		DISEMBLE_RGB(src, srcbpp, srcfmt, pixel, sR, sG, sB);
		dR = dstfmt->palette->colors[*dst].r;
		dG = dstfmt->palette->colors[*dst].g;
		dB = dstfmt->palette->colors[*dst].b;
		ALPHA_BLEND(sR, sG, sB, A, dR, dG, dB);
		dR &= 0xff;
		dG &= 0xff;
		dB &= 0xff;
		/* Pack RGB into 8bit pixel */
		if ( palmap == NULL ) {
		    *dst =((dR>>5)<<(3+2))|
			  ((dG>>5)<<(2))|
			  ((dB>>6)<<(0));
		} else {
		    *dst = palmap[((dR>>5)<<(3+2))|
				  ((dG>>5)<<(2))  |
				  ((dB>>6)<<(0))];
		}
		dst++;
		src += srcbpp;
	    },
	    width);
	    src += srcskip;
	    dst += dstskip;
	}
}

/* N->1 blending with pixel alpha */
static void BlitNto1PixelAlpha(SDL_BlitInfo *info)
{
	int width = info->d_width;
	int height = info->d_height;
	Uint8 *src = info->s_pixels;
	int srcskip = info->s_skip;
	Uint8 *dst = info->d_pixels;
	int dstskip = info->d_skip;
	Uint8 *palmap = info->table;
	SDL_PixelFormat *srcfmt = info->src;
	SDL_PixelFormat *dstfmt = info->dst;
	int srcbpp = srcfmt->BytesPerPixel;

	/* FIXME: fix alpha bit field expansion here too? */
	while ( height-- ) {
	    DUFFS_LOOP4(
	    {
		Uint32 pixel;
		unsigned sR;
		unsigned sG;
		unsigned sB;
		unsigned sA;
		unsigned dR;
		unsigned dG;
		unsigned dB;
		DISEMBLE_RGBA(src,srcbpp,srcfmt,pixel,sR,sG,sB,sA);
		dR = dstfmt->palette->colors[*dst].r;
		dG = dstfmt->palette->colors[*dst].g;
		dB = dstfmt->palette->colors[*dst].b;
		ALPHA_BLEND(sR, sG, sB, sA, dR, dG, dB);
		dR &= 0xff;
		dG &= 0xff;
		dB &= 0xff;
		/* Pack RGB into 8bit pixel */
		if ( palmap == NULL ) {
		    *dst =((dR>>5)<<(3+2))|
			  ((dG>>5)<<(2))|
			  ((dB>>6)<<(0));
		} else {
		    *dst = palmap[((dR>>5)<<(3+2))|
				  ((dG>>5)<<(2))  |
				  ((dB>>6)<<(0))  ];
		}
		dst++;
		src += srcbpp;
	    },
	    width);
	    src += srcskip;
	    dst += dstskip;
	}
}

/* colorkeyed N->1 blending with per-surface alpha */
static void BlitNto1SurfaceAlphaKey(SDL_BlitInfo *info)
{
	int width = info->d_width;
	int height = info->d_height;
	Uint8 *src = info->s_pixels;
	int srcskip = info->s_skip;
	Uint8 *dst = info->d_pixels;
	int dstskip = info->d_skip;
	Uint8 *palmap = info->table;
	SDL_PixelFormat *srcfmt = info->src;
	SDL_PixelFormat *dstfmt = info->dst;
	int srcbpp = srcfmt->BytesPerPixel;
	Uint32 ckey = srcfmt->colorkey;

	const int A = srcfmt->alpha;

	while ( height-- ) {
	    DUFFS_LOOP(
	    {
		Uint32 pixel;
		unsigned sR;
		unsigned sG;
		unsigned sB;
		unsigned dR;
		unsigned dG;
		unsigned dB;
		DISEMBLE_RGB(src, srcbpp, srcfmt, pixel, sR, sG, sB);
		if ( pixel != ckey ) {
		    dR = dstfmt->palette->colors[*dst].r;
		    dG = dstfmt->palette->colors[*dst].g;
		    dB = dstfmt->palette->colors[*dst].b;
		    ALPHA_BLEND(sR, sG, sB, A, dR, dG, dB);
		    dR &= 0xff;
		    dG &= 0xff;
		    dB &= 0xff;
		    /* Pack RGB into 8bit pixel */
		    if ( palmap == NULL ) {
			*dst =((dR>>5)<<(3+2))|
			      ((dG>>5)<<(2)) |
			      ((dB>>6)<<(0));
		    } else {
			*dst = palmap[((dR>>5)<<(3+2))|
				      ((dG>>5)<<(2))  |
				      ((dB>>6)<<(0))  ];
		    }
		}
		dst++;
		src += srcbpp;
	    },
	    width);
	    src += srcskip;
	    dst += dstskip;
	}
}

/* fast RGB888->(A)RGB888 blending with surface alpha=128 special case */
static void BlitRGBtoRGBSurfaceAlpha128(SDL_BlitInfo *info)
{
	int width = info->d_width;
	int height = info->d_height;
	Uint32 *srcp = (Uint32 *)info->s_pixels;
	int srcskip = info->s_skip >> 2;
	Uint32 *dstp = (Uint32 *)info->d_pixels;
	int dstskip = info->d_skip >> 2;

	while(height--) {
	    DUFFS_LOOP4({
		    Uint32 s = *srcp++;
		    Uint32 d = *dstp;
		    *dstp++ = ((((s & 0x00fefefe) + (d & 0x00fefefe)) >> 1)
			       + (s & d & 0x00010101)) | 0xff000000;
	    }, width);
	    srcp += srcskip;
	    dstp += dstskip;
	}
}

/* fast RGB888->(A)RGB888 blending with surface alpha */
static void BlitRGBtoRGBSurfaceAlpha(SDL_BlitInfo *info)
{
	unsigned alpha = info->src->alpha;
	if(alpha == 128) {
		BlitRGBtoRGBSurfaceAlpha128(info);
	} else {
		int width = info->d_width;
		int height = info->d_height;
		Uint32 *srcp = (Uint32 *)info->s_pixels;
		int srcskip = info->s_skip >> 2;
		Uint32 *dstp = (Uint32 *)info->d_pixels;
		int dstskip = info->d_skip >> 2;

		while(height--) {
			DUFFS_LOOP4({
				Uint32 s;
				Uint32 d;
				Uint32 s1;
				Uint32 d1;
				s = *srcp;
				d = *dstp;
				s1 = s & 0xff00ff;
				d1 = d & 0xff00ff;
				d1 = (d1 + ((s1 - d1) * alpha >> 8))
				     & 0xff00ff;
				s &= 0xff00;
				d &= 0xff00;
				d = (d + ((s - d) * alpha >> 8)) & 0xff00;
				*dstp = d1 | d | 0xff000000;
				++srcp;
				++dstp;
			}, width);
			srcp += srcskip;
			dstp += dstskip;
		}
	}
}

/* fast ARGB888->(A)RGB888 blending with pixel alpha */
static void BlitRGBtoRGBPixelAlpha(SDL_BlitInfo *info)
{
	int width = info->d_width;
	int height = info->d_height;
	Uint32 *srcp = (Uint32 *)info->s_pixels;
	int srcskip = info->s_skip >> 2;
	Uint32 *dstp = (Uint32 *)info->d_pixels;
	int dstskip = info->d_skip >> 2;

	while(height--) {
	    DUFFS_LOOP4({
		Uint32 dalpha;
		Uint32 d;
		Uint32 s1;
		Uint32 d1;
		Uint32 s = *srcp;
		Uint32 alpha = s >> 24;
		/* FIXME: Here we special-case opaque alpha since the
		   compositioning used (>>8 instead of /255) doesn't handle
		   it correctly. Also special-case alpha=0 for speed?
		   Benchmark this! */
		if(alpha == SDL_ALPHA_OPAQUE) {
		    *dstp = (s & 0x00ffffff) | (*dstp & 0xff000000);
		} else {
		    /*
		     * take out the middle component (green), and process
		     * the other two in parallel. One multiply less.
		     */
		    d = *dstp;
		    dalpha = d & 0xff000000;
		    s1 = s & 0xff00ff;
		    d1 = d & 0xff00ff;
		    d1 = (d1 + ((s1 - d1) * alpha >> 8)) & 0xff00ff;
		    s &= 0xff00;
		    d &= 0xff00;
		    d = (d + ((s - d) * alpha >> 8)) & 0xff00;
		    *dstp = d1 | d | dalpha;
		}
		++srcp;
		++dstp;
	    }, width);
	    srcp += srcskip;
	    dstp += dstskip;
	}
}

/* 16bpp special case for per-surface alpha=50%: blend 2 pixels in parallel */

/* blend a single 16 bit pixel at 50% */
#define BLEND16_50(d, s, mask)						\
	((((s & mask) + (d & mask)) >> 1) + (s & d & (~mask & 0xffff)))

/* blend two 16 bit pixels at 50% */
#define BLEND2x16_50(d, s, mask)					     \
	(((s & (mask | mask << 16)) >> 1) + ((d & (mask | mask << 16)) >> 1) \
	 + (s & d & (~(mask | mask << 16))))

static void Blit16to16SurfaceAlpha128(SDL_BlitInfo *info, Uint16 mask)
{
	int width = info->d_width;
	int height = info->d_height;
	Uint16 *srcp = (Uint16 *)info->s_pixels;
	int srcskip = info->s_skip >> 1;
	Uint16 *dstp = (Uint16 *)info->d_pixels;
	int dstskip = info->d_skip >> 1;

	while(height--) {
		if(((unsigned long)srcp ^ (unsigned long)dstp) & 2) {
			/*
			 * Source and destination not aligned, pipeline it.
			 * This is mostly a win for big blits but no loss for
			 * small ones
			 */
			Uint32 prev_sw;
			int w = width;

			/* handle odd destination */
			if((unsigned long)dstp & 2) {
				Uint16 d = *dstp, s = *srcp;
				*dstp = BLEND16_50(d, s, mask);
				dstp++;
				srcp++;
				w--;
			}
			srcp++;	/* srcp is now 32-bit aligned */

			/* bootstrap pipeline with first halfword */
			prev_sw = ((Uint32 *)srcp)[-1];

			while(w > 1) {
				Uint32 sw, dw, s;
				sw = *(Uint32 *)srcp;
				dw = *(Uint32 *)dstp;
				if(SDL_BYTEORDER == SDL_BIG_ENDIAN)
					s = (prev_sw << 16) + (sw >> 16);
				else
					s = (prev_sw >> 16) + (sw << 16);
				prev_sw = sw;
				*(Uint32 *)dstp = BLEND2x16_50(dw, s, mask);
				dstp += 2;
				srcp += 2;
				w -= 2;
			}

			/* final pixel if any */
			if(w) {
				Uint16 d = *dstp, s;
				if(SDL_BYTEORDER == SDL_BIG_ENDIAN)
					s = prev_sw;
				else
					s = prev_sw >> 16;
				*dstp = BLEND16_50(d, s, mask);
				srcp++;
				dstp++;
			}
			srcp += srcskip - 1;
			dstp += dstskip;
		} else {
			/* source and destination are aligned */
			int w = width;

			/* first odd pixel? */
			if((unsigned long)srcp & 2) {
				Uint16 d = *dstp, s = *srcp;
				*dstp = BLEND16_50(d, s, mask);
				srcp++;
				dstp++;
				w--;
			}
			/* srcp and dstp are now 32-bit aligned */