Merge "Remove the argb* files."

This commit is contained in:
Vincent Rabaud 2017-06-25 15:57:17 +00:00 committed by Gerrit Code Review
commit 28914528e1
12 changed files with 77 additions and 269 deletions

View File

@ -51,9 +51,6 @@ dsp_dec_srcs := \
src/dsp/alpha_processing_neon.$(NEON) \ src/dsp/alpha_processing_neon.$(NEON) \
src/dsp/alpha_processing_sse2.c \ src/dsp/alpha_processing_sse2.c \
src/dsp/alpha_processing_sse41.c \ src/dsp/alpha_processing_sse41.c \
src/dsp/argb.c \
src/dsp/argb_mips_dsp_r2.c \
src/dsp/argb_sse2.c \
src/dsp/cpu.c \ src/dsp/cpu.c \
src/dsp/dec.c \ src/dsp/dec.c \
src/dsp/dec_clip_tables.c \ src/dsp/dec_clip_tables.c \

View File

@ -233,9 +233,6 @@ DSP_DEC_OBJS = \
$(DIROBJ)\dsp\yuv_sse2.obj \ $(DIROBJ)\dsp\yuv_sse2.obj \
DSP_ENC_OBJS = \ DSP_ENC_OBJS = \
$(DIROBJ)\dsp\argb.obj \
$(DIROBJ)\dsp\argb_mips_dsp_r2.obj \
$(DIROBJ)\dsp\argb_sse2.obj \
$(DIROBJ)\dsp\cost.obj \ $(DIROBJ)\dsp\cost.obj \
$(DIROBJ)\dsp\cost_mips32.obj \ $(DIROBJ)\dsp\cost_mips32.obj \
$(DIROBJ)\dsp\cost_mips_dsp_r2.obj \ $(DIROBJ)\dsp\cost_mips_dsp_r2.obj \

View File

@ -119,9 +119,6 @@ model {
include "alpha_processing_neon.$NEON" include "alpha_processing_neon.$NEON"
include "alpha_processing_sse2.c" include "alpha_processing_sse2.c"
include "alpha_processing_sse41.c" include "alpha_processing_sse41.c"
include "argb.c"
include "argb_mips_dsp_r2.c"
include "argb_sse2.c"
include "cpu.c" include "cpu.c"
include "dec.c" include "dec.c"
include "dec_clip_tables.c" include "dec_clip_tables.c"

View File

@ -173,9 +173,6 @@ DSP_DEC_OBJS = \
src/dsp/yuv_sse2.o \ src/dsp/yuv_sse2.o \
DSP_ENC_OBJS = \ DSP_ENC_OBJS = \
src/dsp/argb.o \
src/dsp/argb_mips_dsp_r2.o \
src/dsp/argb_sse2.o \
src/dsp/cost.o \ src/dsp/cost.o \
src/dsp/cost_mips32.o \ src/dsp/cost_mips32.o \
src/dsp/cost_mips_dsp_r2.o \ src/dsp/cost_mips_dsp_r2.o \

View File

@ -39,8 +39,6 @@ COMMON_SOURCES += yuv_mips32.c
COMMON_SOURCES += yuv_mips_dsp_r2.c COMMON_SOURCES += yuv_mips_dsp_r2.c
ENC_SOURCES = ENC_SOURCES =
ENC_SOURCES += argb.c
ENC_SOURCES += argb_mips_dsp_r2.c
ENC_SOURCES += cost.c ENC_SOURCES += cost.c
ENC_SOURCES += cost_mips32.c ENC_SOURCES += cost_mips32.c
ENC_SOURCES += cost_mips_dsp_r2.c ENC_SOURCES += cost_mips_dsp_r2.c
@ -97,7 +95,6 @@ libwebpdspdecode_msa_la_CPPFLAGS = $(libwebpdsp_msa_la_CPPFLAGS)
libwebpdspdecode_msa_la_CFLAGS = $(libwebpdsp_msa_la_CFLAGS) libwebpdspdecode_msa_la_CFLAGS = $(libwebpdsp_msa_la_CFLAGS)
libwebpdsp_sse2_la_SOURCES = libwebpdsp_sse2_la_SOURCES =
libwebpdsp_sse2_la_SOURCES += argb_sse2.c
libwebpdsp_sse2_la_SOURCES += cost_sse2.c libwebpdsp_sse2_la_SOURCES += cost_sse2.c
libwebpdsp_sse2_la_SOURCES += enc_sse2.c libwebpdsp_sse2_la_SOURCES += enc_sse2.c
libwebpdsp_sse2_la_SOURCES += lossless_enc_sse2.c libwebpdsp_sse2_la_SOURCES += lossless_enc_sse2.c

View File

@ -339,12 +339,30 @@ static void ExtractGreen_C(const uint32_t* argb, uint8_t* alpha, int size) {
for (i = 0; i < size; ++i) alpha[i] = argb[i] >> 8; for (i = 0; i < size; ++i) alpha[i] = argb[i] >> 8;
} }
//------------------------------------------------------------------------------
// Simple channel manipulations.
static WEBP_INLINE uint32_t MakeARGB32(int a, int r, int g, int b) {
return (((uint32_t)a << 24) | (r << 16) | (g << 8) | b);
}
static void PackRGB_C(const uint8_t* r, const uint8_t* g, const uint8_t* b,
int len, int step, uint32_t* out) {
int i, offset = 0;
for (i = 0; i < len; ++i) {
out[i] = MakeARGB32(0xff, r[offset], g[offset], b[offset]);
offset += step;
}
}
void (*WebPApplyAlphaMultiply)(uint8_t*, int, int, int, int); void (*WebPApplyAlphaMultiply)(uint8_t*, int, int, int, int);
void (*WebPApplyAlphaMultiply4444)(uint8_t*, int, int, int); void (*WebPApplyAlphaMultiply4444)(uint8_t*, int, int, int);
int (*WebPDispatchAlpha)(const uint8_t*, int, int, int, uint8_t*, int); int (*WebPDispatchAlpha)(const uint8_t*, int, int, int, uint8_t*, int);
void (*WebPDispatchAlphaToGreen)(const uint8_t*, int, int, int, uint32_t*, int); void (*WebPDispatchAlphaToGreen)(const uint8_t*, int, int, int, uint32_t*, int);
int (*WebPExtractAlpha)(const uint8_t*, int, int, int, uint8_t*, int); int (*WebPExtractAlpha)(const uint8_t*, int, int, int, uint8_t*, int);
void (*WebPExtractGreen)(const uint32_t* argb, uint8_t* alpha, int size); void (*WebPExtractGreen)(const uint32_t* argb, uint8_t* alpha, int size);
void (*WebPPackRGB)(const uint8_t* r, const uint8_t* g, const uint8_t* b,
int len, int step, uint32_t* out);
//------------------------------------------------------------------------------ //------------------------------------------------------------------------------
// Init function // Init function
@ -370,6 +388,8 @@ WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessing(void) {
WebPExtractAlpha = ExtractAlpha_C; WebPExtractAlpha = ExtractAlpha_C;
WebPExtractGreen = ExtractGreen_C; WebPExtractGreen = ExtractGreen_C;
WebPPackRGB = PackRGB_C;
// If defined, use CPUInfo() to overwrite some pointers with faster versions. // If defined, use CPUInfo() to overwrite some pointers with faster versions.
if (VP8GetCPUInfo != NULL) { if (VP8GetCPUInfo != NULL) {
#if defined(WEBP_USE_SSE2) #if defined(WEBP_USE_SSE2)

View File

@ -124,6 +124,45 @@ static void MultARGBRow(uint32_t* const ptr, int width, int inverse) {
} }
} }
static void PackRGB_MIPSdspR2(const uint8_t* r, const uint8_t* g,
const uint8_t* b, int len, int step,
uint32_t* out) {
int temp0, temp1, temp2, offset;
const int rest = len & 1;
const int a = 0xff;
const uint32_t* const loop_end = out + len - rest;
__asm__ volatile (
"xor %[offset], %[offset], %[offset] \n\t"
"beq %[loop_end], %[out], 0f \n\t"
"2: \n\t"
"lbux %[temp0], %[offset](%[r]) \n\t"
"lbux %[temp1], %[offset](%[g]) \n\t"
"lbux %[temp2], %[offset](%[b]) \n\t"
"ins %[temp0], %[a], 16, 16 \n\t"
"ins %[temp2], %[temp1], 16, 16 \n\t"
"addiu %[out], %[out], 4 \n\t"
"precr.qb.ph %[temp0], %[temp0], %[temp2] \n\t"
"sw %[temp0], -4(%[out]) \n\t"
"addu %[offset], %[offset], %[step] \n\t"
"bne %[loop_end], %[out], 2b \n\t"
"0: \n\t"
"beq %[rest], $zero, 1f \n\t"
"lbux %[temp0], %[offset](%[r]) \n\t"
"lbux %[temp1], %[offset](%[g]) \n\t"
"lbux %[temp2], %[offset](%[b]) \n\t"
"ins %[temp0], %[a], 16, 16 \n\t"
"ins %[temp2], %[temp1], 16, 16 \n\t"
"precr.qb.ph %[temp0], %[temp0], %[temp2] \n\t"
"sw %[temp0], 0(%[out]) \n\t"
"1: \n\t"
: [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
[offset]"=&r"(offset), [out]"+&r"(out)
: [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
[loop_end]"r"(loop_end), [rest]"r"(rest)
: "memory"
);
}
//------------------------------------------------------------------------------ //------------------------------------------------------------------------------
// Entry point // Entry point
@ -132,6 +171,7 @@ extern void WebPInitAlphaProcessingMIPSdspR2(void);
WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessingMIPSdspR2(void) { WEBP_TSAN_IGNORE_FUNCTION void WebPInitAlphaProcessingMIPSdspR2(void) {
WebPDispatchAlpha = DispatchAlpha; WebPDispatchAlpha = DispatchAlpha;
WebPMultARGBRow = MultARGBRow; WebPMultARGBRow = MultARGBRow;
WebPPackRGB = PackRGB_MIPSdspR2;
} }
#else // !WEBP_USE_MIPS_DSP_R2 #else // !WEBP_USE_MIPS_DSP_R2

View File

@ -1,68 +0,0 @@
// Copyright 2014 Google Inc. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the COPYING file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
// -----------------------------------------------------------------------------
//
// ARGB making functions.
//
// Author: Djordje Pesut (djordje.pesut@imgtec.com)
#include "./dsp.h"
static WEBP_INLINE uint32_t MakeARGB32(int a, int r, int g, int b) {
return (((uint32_t)a << 24) | (r << 16) | (g << 8) | b);
}
static void PackARGB(const uint8_t* a, const uint8_t* r, const uint8_t* g,
const uint8_t* b, int len, uint32_t* out) {
int i;
for (i = 0; i < len; ++i) {
out[i] = MakeARGB32(a[4 * i], r[4 * i], g[4 * i], b[4 * i]);
}
}
static void PackRGB(const uint8_t* r, const uint8_t* g, const uint8_t* b,
int len, int step, uint32_t* out) {
int i, offset = 0;
for (i = 0; i < len; ++i) {
out[i] = MakeARGB32(0xff, r[offset], g[offset], b[offset]);
offset += step;
}
}
void (*VP8PackARGB)(const uint8_t*, const uint8_t*, const uint8_t*,
const uint8_t*, int, uint32_t*);
void (*VP8PackRGB)(const uint8_t*, const uint8_t*, const uint8_t*,
int, int, uint32_t*);
extern void VP8EncDspARGBInitMIPSdspR2(void);
extern void VP8EncDspARGBInitSSE2(void);
static volatile VP8CPUInfo argb_last_cpuinfo_used =
(VP8CPUInfo)&argb_last_cpuinfo_used;
WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspARGBInit(void) {
if (argb_last_cpuinfo_used == VP8GetCPUInfo) return;
VP8PackARGB = PackARGB;
VP8PackRGB = PackRGB;
// If defined, use CPUInfo() to overwrite some pointers with faster versions.
if (VP8GetCPUInfo != NULL) {
#if defined(WEBP_USE_SSE2)
if (VP8GetCPUInfo(kSSE2)) {
VP8EncDspARGBInitSSE2();
}
#endif
#if defined(WEBP_USE_MIPS_DSP_R2)
if (VP8GetCPUInfo(kMIPSdspR2)) {
VP8EncDspARGBInitMIPSdspR2();
}
#endif
}
argb_last_cpuinfo_used = VP8GetCPUInfo;
}

View File

@ -1,110 +0,0 @@
// Copyright 2014 Google Inc. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the COPYING file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
// -----------------------------------------------------------------------------
//
// ARGB making functions (mips version).
//
// Author: Djordje Pesut (djordje.pesut@imgtec.com)
#include "./dsp.h"
#if defined(WEBP_USE_MIPS_DSP_R2)
static void PackARGB(const uint8_t* a, const uint8_t* r, const uint8_t* g,
const uint8_t* b, int len, uint32_t* out) {
int temp0, temp1, temp2, temp3, offset;
const int rest = len & 1;
const uint32_t* const loop_end = out + len - rest;
const int step = 4;
__asm__ volatile (
"xor %[offset], %[offset], %[offset] \n\t"
"beq %[loop_end], %[out], 0f \n\t"
"2: \n\t"
"lbux %[temp0], %[offset](%[a]) \n\t"
"lbux %[temp1], %[offset](%[r]) \n\t"
"lbux %[temp2], %[offset](%[g]) \n\t"
"lbux %[temp3], %[offset](%[b]) \n\t"
"ins %[temp1], %[temp0], 16, 16 \n\t"
"ins %[temp3], %[temp2], 16, 16 \n\t"
"addiu %[out], %[out], 4 \n\t"
"precr.qb.ph %[temp0], %[temp1], %[temp3] \n\t"
"sw %[temp0], -4(%[out]) \n\t"
"addu %[offset], %[offset], %[step] \n\t"
"bne %[loop_end], %[out], 2b \n\t"
"0: \n\t"
"beq %[rest], $zero, 1f \n\t"
"lbux %[temp0], %[offset](%[a]) \n\t"
"lbux %[temp1], %[offset](%[r]) \n\t"
"lbux %[temp2], %[offset](%[g]) \n\t"
"lbux %[temp3], %[offset](%[b]) \n\t"
"ins %[temp1], %[temp0], 16, 16 \n\t"
"ins %[temp3], %[temp2], 16, 16 \n\t"
"precr.qb.ph %[temp0], %[temp1], %[temp3] \n\t"
"sw %[temp0], 0(%[out]) \n\t"
"1: \n\t"
: [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
[temp3]"=&r"(temp3), [offset]"=&r"(offset), [out]"+&r"(out)
: [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
[loop_end]"r"(loop_end), [rest]"r"(rest)
: "memory"
);
}
static void PackRGB(const uint8_t* r, const uint8_t* g, const uint8_t* b,
int len, int step, uint32_t* out) {
int temp0, temp1, temp2, offset;
const int rest = len & 1;
const int a = 0xff;
const uint32_t* const loop_end = out + len - rest;
__asm__ volatile (
"xor %[offset], %[offset], %[offset] \n\t"
"beq %[loop_end], %[out], 0f \n\t"
"2: \n\t"
"lbux %[temp0], %[offset](%[r]) \n\t"
"lbux %[temp1], %[offset](%[g]) \n\t"
"lbux %[temp2], %[offset](%[b]) \n\t"
"ins %[temp0], %[a], 16, 16 \n\t"
"ins %[temp2], %[temp1], 16, 16 \n\t"
"addiu %[out], %[out], 4 \n\t"
"precr.qb.ph %[temp0], %[temp0], %[temp2] \n\t"
"sw %[temp0], -4(%[out]) \n\t"
"addu %[offset], %[offset], %[step] \n\t"
"bne %[loop_end], %[out], 2b \n\t"
"0: \n\t"
"beq %[rest], $zero, 1f \n\t"
"lbux %[temp0], %[offset](%[r]) \n\t"
"lbux %[temp1], %[offset](%[g]) \n\t"
"lbux %[temp2], %[offset](%[b]) \n\t"
"ins %[temp0], %[a], 16, 16 \n\t"
"ins %[temp2], %[temp1], 16, 16 \n\t"
"precr.qb.ph %[temp0], %[temp0], %[temp2] \n\t"
"sw %[temp0], 0(%[out]) \n\t"
"1: \n\t"
: [temp0]"=&r"(temp0), [temp1]"=&r"(temp1), [temp2]"=&r"(temp2),
[offset]"=&r"(offset), [out]"+&r"(out)
: [a]"r"(a), [r]"r"(r), [g]"r"(g), [b]"r"(b), [step]"r"(step),
[loop_end]"r"(loop_end), [rest]"r"(rest)
: "memory"
);
}
//------------------------------------------------------------------------------
// Entry point
extern void VP8EncDspARGBInitMIPSdspR2(void);
WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspARGBInitMIPSdspR2(void) {
VP8PackARGB = PackARGB;
VP8PackRGB = PackRGB;
}
#else // !WEBP_USE_MIPS_DSP_R2
WEBP_DSP_INIT_STUB(VP8EncDspARGBInitMIPSdspR2)
#endif // WEBP_USE_MIPS_DSP_R2

View File

@ -1,53 +0,0 @@
// Copyright 2014 Google Inc. All Rights Reserved.
//
// Use of this source code is governed by a BSD-style license
// that can be found in the COPYING file in the root of the source
// tree. An additional intellectual property rights grant can be found
// in the file PATENTS. All contributing project authors may
// be found in the AUTHORS file in the root of the source tree.
// -----------------------------------------------------------------------------
//
// ARGB making functions (SSE2 version).
//
// Author: Skal (pascal.massimino@gmail.com)
#include "./dsp.h"
#include "./lossless.h"
#if defined(WEBP_USE_SSE2)
#include <assert.h>
#include <emmintrin.h>
#include <string.h>
static void PackARGB(const uint8_t* a, const uint8_t* r, const uint8_t* g,
const uint8_t* b, int len, uint32_t* out) {
(void)a;
if (g == r + 1) { // RGBA input order. Need to swap R and B.
assert(b == r + 2);
assert(a == r + 3);
VP8LConvertBGRAToRGBA((const uint32_t*)r, len, (uint8_t*)out);
} else {
assert(g == b + 1);
assert(r == b + 2);
assert(a == b + 3);
memcpy(out, b, len * 4);
}
}
//------------------------------------------------------------------------------
// Entry point
extern void VP8EncDspARGBInitSSE2(void);
extern void VP8LDspInitSSE2(void);
WEBP_TSAN_IGNORE_FUNCTION void VP8EncDspARGBInitSSE2(void) {
VP8LDspInitSSE2();
VP8PackARGB = PackARGB;
}
#else // !WEBP_USE_SSE2
WEBP_DSP_INIT_STUB(VP8EncDspARGBInitSSE2)
#endif // WEBP_USE_SSE2

View File

@ -542,21 +542,13 @@ void WebPMultRowC(uint8_t* const ptr, const uint8_t* const alpha,
int width, int inverse); int width, int inverse);
void WebPMultARGBRowC(uint32_t* const ptr, int width, int inverse); void WebPMultARGBRowC(uint32_t* const ptr, int width, int inverse);
// RGB packing function. 'step' can be 3 or 4. r/g/b input is rgb or bgr order.
extern void (*WebPPackRGB)(const uint8_t* r, const uint8_t* g, const uint8_t* b,
int len, int step, uint32_t* out);
// To be called first before using the above. // To be called first before using the above.
void WebPInitAlphaProcessing(void); void WebPInitAlphaProcessing(void);
// ARGB packing function: a/r/g/b input is rgba or bgra order.
extern void (*VP8PackARGB)(const uint8_t* a, const uint8_t* r,
const uint8_t* g, const uint8_t* b, int len,
uint32_t* out);
// RGB packing function. 'step' can be 3 or 4. r/g/b input is rgb or bgr order.
extern void (*VP8PackRGB)(const uint8_t* r, const uint8_t* g, const uint8_t* b,
int len, int step, uint32_t* out);
// To be called first before using the above.
void VP8EncDspARGBInit(void);
//------------------------------------------------------------------------------ //------------------------------------------------------------------------------
// Filter functions // Filter functions

View File

@ -18,6 +18,8 @@
#include "./vp8i_enc.h" #include "./vp8i_enc.h"
#include "../utils/random_utils.h" #include "../utils/random_utils.h"
#include "../utils/utils.h" #include "../utils/utils.h"
#include "../dsp/dsp.h"
#include "../dsp/lossless.h"
#include "../dsp/yuv.h" #include "../dsp/yuv.h"
// Uncomment to disable gamma-compression during RGB->U/V averaging // Uncomment to disable gamma-compression during RGB->U/V averaging
@ -1085,45 +1087,45 @@ int WebPPictureYUVAToARGB(WebPPicture* picture) {
// automatic import / conversion // automatic import / conversion
static int Import(WebPPicture* const picture, static int Import(WebPPicture* const picture,
const uint8_t* const rgb, int rgb_stride, const uint8_t* rgb, int rgb_stride,
int step, int swap_rb, int import_alpha) { int step, int swap_rb, int import_alpha) {
int y; int y;
const uint8_t* r_ptr = rgb + (swap_rb ? 2 : 0); const uint8_t* r_ptr = rgb + (swap_rb ? 2 : 0);
const uint8_t* g_ptr = rgb + 1; const uint8_t* g_ptr = rgb + 1;
const uint8_t* b_ptr = rgb + (swap_rb ? 0 : 2); const uint8_t* b_ptr = rgb + (swap_rb ? 0 : 2);
const uint8_t* a_ptr = import_alpha ? rgb + 3 : NULL;
const int width = picture->width; const int width = picture->width;
const int height = picture->height; const int height = picture->height;
if (!picture->use_argb) { if (!picture->use_argb) {
const uint8_t* a_ptr = import_alpha ? rgb + 3 : NULL;
return ImportYUVAFromRGBA(r_ptr, g_ptr, b_ptr, a_ptr, step, rgb_stride, return ImportYUVAFromRGBA(r_ptr, g_ptr, b_ptr, a_ptr, step, rgb_stride,
0.f /* no dithering */, 0, picture); 0.f /* no dithering */, 0, picture);
} }
if (!WebPPictureAlloc(picture)) return 0; if (!WebPPictureAlloc(picture)) return 0;
VP8EncDspARGBInit(); VP8LDspInit();
WebPInitAlphaProcessing();
if (import_alpha) { if (import_alpha) {
uint32_t* dst = picture->argb; uint32_t* dst = picture->argb;
const int do_copy = !swap_rb && !ALPHA_IS_LAST; const int do_copy =
(!swap_rb && !ALPHA_IS_LAST) || (swap_rb && ALPHA_IS_LAST);
assert(step == 4); assert(step == 4);
for (y = 0; y < height; ++y) { for (y = 0; y < height; ++y) {
if (do_copy) { if (do_copy) {
memcpy(dst, r_ptr, width * sizeof(*dst)); memcpy(dst, rgb, width * 4);
} else { } else {
VP8PackARGB(a_ptr, r_ptr, g_ptr, b_ptr, width, dst); // RGBA input order. Need to swap R and B.
VP8LConvertBGRAToRGBA((const uint32_t*)rgb, width, (uint8_t*)dst);
} }
a_ptr += rgb_stride; rgb += rgb_stride;
r_ptr += rgb_stride;
g_ptr += rgb_stride;
b_ptr += rgb_stride;
dst += picture->argb_stride; dst += picture->argb_stride;
} }
} else { } else {
uint32_t* dst = picture->argb; uint32_t* dst = picture->argb;
assert(step >= 3); assert(step >= 3);
for (y = 0; y < height; ++y) { for (y = 0; y < height; ++y) {
VP8PackRGB(r_ptr, g_ptr, b_ptr, width, step, dst); WebPPackRGB(r_ptr, g_ptr, b_ptr, width, step, dst);
r_ptr += rgb_stride; r_ptr += rgb_stride;
g_ptr += rgb_stride; g_ptr += rgb_stride;
b_ptr += rgb_stride; b_ptr += rgb_stride;