video: move standard c, sse and lsx implementations of yuv2rgb to its own source

This commit is contained in:
Anonymous Maarten 2024-01-19 00:01:36 +01:00 committed by Sam Lantinga
parent 0d9719964f
commit 472a415c8f
14 changed files with 1222 additions and 731 deletions

View file

@ -81,7 +81,7 @@ SRCS+= SDL_events.c SDL_quit.c SDL_keyboard.c SDL_mouse.c SDL_windowevents.c &
SDL_clipboardevents.c SDL_dropevents.c SDL_displayevents.c SDL_gesture.c & SDL_clipboardevents.c SDL_dropevents.c SDL_displayevents.c SDL_gesture.c &
SDL_sensor.c SDL_touch.c SDL_sensor.c SDL_touch.c
SRCS+= SDL_haptic.c SDL_hidapi.c SDL_gamecontroller.c SDL_joystick.c controller_type.c SRCS+= SDL_haptic.c SDL_hidapi.c SDL_gamecontroller.c SDL_joystick.c controller_type.c
SRCS+= SDL_render.c yuv_rgb.c SDL_yuv.c SDL_yuv_sw.c SDL_blendfillrect.c & SRCS+= SDL_render.c yuv_rgb_lsx.c yuv_rgb_sse.c yuv_rgb_std.c SDL_yuv.c SDL_yuv_sw.c SDL_blendfillrect.c &
SDL_blendline.c SDL_blendpoint.c SDL_drawline.c SDL_drawpoint.c & SDL_blendline.c SDL_blendpoint.c SDL_drawline.c SDL_drawpoint.c &
SDL_render_sw.c SDL_rotate.c SDL_triangle.c SDL_render_sw.c SDL_rotate.c SDL_triangle.c
SRCS+= SDL_blit.c SDL_blit_0.c SDL_blit_1.c SDL_blit_A.c SDL_blit_auto.c & SRCS+= SDL_blit.c SDL_blit_0.c SDL_blit_1.c SDL_blit_A.c SDL_blit_auto.c &
@ -158,6 +158,9 @@ SDL_hidapi.obj: SDL_hidapi.c
wcc386 $(CFLAGS_DLL) -za99 -fo=$^@ $< wcc386 $(CFLAGS_DLL) -za99 -fo=$^@ $<
!endif !endif
yuv_rgb_sse.obj: yuv_rgb_sse.c
wcc386 $(CFLAGS_DLL) -wcd=202 -fo=$^@ $<
$(LIBICONV_LIB): "src/core/os2/iconv2.lbc" $(LIBICONV_LIB): "src/core/os2/iconv2.lbc"
@echo * Creating: $@ @echo * Creating: $@
wlib -q -b -n -c -pa -s -t -zld -ii -io $@ @$< wlib -q -b -n -c -pa -s -t -zld -ii -io $@ @$<

View file

@ -60,7 +60,7 @@ SRCS+= SDL_events.c SDL_quit.c SDL_keyboard.c SDL_mouse.c SDL_windowevents.c &
SDL_clipboardevents.c SDL_dropevents.c SDL_displayevents.c SDL_gesture.c & SDL_clipboardevents.c SDL_dropevents.c SDL_displayevents.c SDL_gesture.c &
SDL_sensor.c SDL_touch.c SDL_sensor.c SDL_touch.c
SRCS+= SDL_haptic.c SDL_hidapi.c SDL_gamecontroller.c SDL_joystick.c controller_type.c SRCS+= SDL_haptic.c SDL_hidapi.c SDL_gamecontroller.c SDL_joystick.c controller_type.c
SRCS+= SDL_render.c yuv_rgb.c SDL_yuv.c SDL_yuv_sw.c SDL_blendfillrect.c & SRCS+= SDL_render.c yuv_rgb_lsx.c yuv_rgb_sse.c yuv_rgb_std.c SDL_yuv.c SDL_yuv_sw.c SDL_blendfillrect.c &
SDL_blendline.c SDL_blendpoint.c SDL_drawline.c SDL_drawpoint.c & SDL_blendline.c SDL_blendpoint.c SDL_drawline.c SDL_drawpoint.c &
SDL_render_sw.c SDL_rotate.c SDL_triangle.c SDL_render_sw.c SDL_rotate.c SDL_triangle.c
SRCS+= SDL_blit.c SDL_blit_0.c SDL_blit_1.c SDL_blit_A.c SDL_blit_auto.c & SRCS+= SDL_blit.c SDL_blit_0.c SDL_blit_1.c SDL_blit_A.c SDL_blit_auto.c &
@ -147,6 +147,9 @@ SDL_RLEaccel.obj: SDL_RLEaccel.c
SDL_malloc.obj: SDL_malloc.c SDL_malloc.obj: SDL_malloc.c
wcc386 $(CFLAGS_DLL) -wcd=201 -fo=$^@ $< wcc386 $(CFLAGS_DLL) -wcd=201 -fo=$^@ $<
yuv_rgb_sse.obj: yuv_rgb_sse.c
wcc386 $(CFLAGS_DLL) -wcd=202 -fo=$^@ $<
# SDL2libm # SDL2libm
MSRCS= e_atan2.c e_exp.c e_fmod.c e_log10.c e_log.c e_pow.c e_rem_pio2.c e_sqrt.c & MSRCS= e_atan2.c e_exp.c e_fmod.c e_log10.c e_log.c e_pow.c e_rem_pio2.c e_sqrt.c &
k_cos.c k_rem_pio2.c k_sin.c k_tan.c & k_cos.c k_rem_pio2.c k_sin.c k_tan.c &

View file

@ -520,7 +520,13 @@
<ClInclude Include="..\..\src\video\windows\SDL_windowswindow.h" /> <ClInclude Include="..\..\src\video\windows\SDL_windowswindow.h" />
<ClInclude Include="..\..\src\video\windows\wmmsg.h" /> <ClInclude Include="..\..\src\video\windows\wmmsg.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb.h" /> <ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_common.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_internal.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_lsx.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_lsx_func.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_sse.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_sse_func.h" /> <ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_sse_func.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_std.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_std_func.h" /> <ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_std_func.h" />
</ItemGroup> </ItemGroup>
<ItemGroup> <ItemGroup>
@ -755,7 +761,9 @@
<ClCompile Include="..\..\src\video\windows\SDL_windowsvideo.c" /> <ClCompile Include="..\..\src\video\windows\SDL_windowsvideo.c" />
<ClCompile Include="..\..\src\video\windows\SDL_windowsvulkan.c" /> <ClCompile Include="..\..\src\video\windows\SDL_windowsvulkan.c" />
<ClCompile Include="..\..\src\video\windows\SDL_windowswindow.c" /> <ClCompile Include="..\..\src\video\windows\SDL_windowswindow.c" />
<ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb.c" /> <ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb_lsx.c" />
<ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb_sse.c" />
<ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb_std.c" />
</ItemGroup> </ItemGroup>
<ItemGroup> <ItemGroup>
<ResourceCompile Include="..\..\src\main\windows\version.rc" /> <ResourceCompile Include="..\..\src\main\windows\version.rc" />

View file

@ -185,6 +185,14 @@
<ClInclude Include="..\src\video\winrt\SDL_winrtopengles.h" /> <ClInclude Include="..\src\video\winrt\SDL_winrtopengles.h" />
<ClInclude Include="..\src\video\winrt\SDL_winrtvideo_cpp.h" /> <ClInclude Include="..\src\video\winrt\SDL_winrtvideo_cpp.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb.h" /> <ClInclude Include="..\src\video\yuv2rgb\yuv_rgb.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_common.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_internal.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_lsx.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_lsx_func.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_sse.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_sse_func.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_std.h" />
<ClInclude Include="..\src\video\yuv2rgb\yuv_rgb_std_func.h" />
</ItemGroup> </ItemGroup>
<ItemGroup> <ItemGroup>
<ClCompile Include="..\src\atomic\SDL_atomic.c" /> <ClCompile Include="..\src\atomic\SDL_atomic.c" />
@ -346,7 +354,9 @@
<ClCompile Include="..\src\video\winrt\SDL_winrtvideo.cpp"> <ClCompile Include="..\src\video\winrt\SDL_winrtvideo.cpp">
<CompileAsWinRT>true</CompileAsWinRT> <CompileAsWinRT>true</CompileAsWinRT>
</ClCompile> </ClCompile>
<ClCompile Include="..\src\video\yuv2rgb\yuv_rgb.c" /> <ClCompile Include="..\src\video\yuv2rgb\yuv_rgb_lsx.c" />
<ClCompile Include="..\src\video\yuv2rgb\yuv_rgb_sse.c" />
<ClCompile Include="..\src\video\yuv2rgb\yuv_rgb_std.c" />
</ItemGroup> </ItemGroup>
<PropertyGroup Label="Globals"> <PropertyGroup Label="Globals">
<ProjectGuid>{89e9b32e-a86a-47c3-a948-d2b1622925ce}</ProjectGuid> <ProjectGuid>{89e9b32e-a86a-47c3-a948-d2b1622925ce}</ProjectGuid>

View file

@ -439,7 +439,13 @@
<ClInclude Include="..\..\src\video\windows\SDL_windowswindow.h" /> <ClInclude Include="..\..\src\video\windows\SDL_windowswindow.h" />
<ClInclude Include="..\..\src\video\windows\wmmsg.h" /> <ClInclude Include="..\..\src\video\windows\wmmsg.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb.h" /> <ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_common.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_internal.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_lsx.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_lsx_func.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_sse.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_sse_func.h" /> <ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_sse_func.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_std.h" />
<ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_std_func.h" /> <ClInclude Include="..\..\src\video\yuv2rgb\yuv_rgb_std_func.h" />
</ItemGroup> </ItemGroup>
<ItemGroup> <ItemGroup>
@ -624,7 +630,9 @@
<ClCompile Include="..\..\src\video\windows\SDL_windowsvideo.c" /> <ClCompile Include="..\..\src\video\windows\SDL_windowsvideo.c" />
<ClCompile Include="..\..\src\video\windows\SDL_windowsvulkan.c" /> <ClCompile Include="..\..\src\video\windows\SDL_windowsvulkan.c" />
<ClCompile Include="..\..\src\video\windows\SDL_windowswindow.c" /> <ClCompile Include="..\..\src\video\windows\SDL_windowswindow.c" />
<ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb.c" /> <ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb_lsx.c" />
<ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb_sse.c" />
<ClCompile Include="..\..\src\video\yuv2rgb\yuv_rgb_std.c" />
</ItemGroup> </ItemGroup>
<ItemGroup> <ItemGroup>
<ResourceCompile Include="..\..\src\main\windows\version.rc" /> <ResourceCompile Include="..\..\src\main\windows\version.rc" />

View file

@ -1,3 +1,6 @@
#ifndef YUV_RGB_H_
#define YUV_RGB_H_
// Copyright 2016 Adrien Descamps // Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License // Distributed under BSD 3-Clause License
@ -16,397 +19,15 @@
// For all methods, width and height should be even, if not, the last row/column of the result image won't be affected. // For all methods, width and height should be even, if not, the last row/column of the result image won't be affected.
// For sse methods, if the width if not divisable by 32, the last (width%32) pixels of each line won't be affected. // For sse methods, if the width if not divisable by 32, the last (width%32) pixels of each line won't be affected.
#include "SDL_stdinc.h"
/*#include <stdint.h>*/ /*#include <stdint.h>*/
typedef enum
{
YCBCR_JPEG,
YCBCR_601,
YCBCR_709
} YCbCrType;
// yuv to rgb, standard c implementation // yuv to rgb, standard c implementation
void yuv420_rgb565_std( #include "yuv_rgb_std.h"
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_std( // yuv to rgb, sse2 implementation
uint32_t width, uint32_t height, #include "yuv_rgb_sse.h"
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_std( // yuv to rgb, lsx implementation
uint32_t width, uint32_t height, #include "yuv_rgb_lsx.h"
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_std( #endif /* YUV_RGB_H_ */
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// yuv to rgb, sse implementation
// pointers must be 16 byte aligned, and strides must be divisable by 16
void yuv420_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// yuv to rgb, sse implementation
// pointers do not need to be 16 byte aligned
void yuv420_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// rgb to yuv, standard c implementation
void rgb24_yuv420_std(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
// rgb to yuv, sse implementation
// pointers must be 16 byte aligned, and strides must be divisible by 16
void rgb24_yuv420_sse(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
// rgb to yuv, sse implementation
// pointers do not need to be 16 byte aligned
void rgb24_yuv420_sseu(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
//yuv420 to bgra, lsx implementation
void yuv420_rgb24_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);

View file

@ -0,0 +1,13 @@
#ifndef YUV_RGB_COMMON_H_
#define YUV_RGB_COMMON_H_
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
typedef enum
{
YCBCR_JPEG,
YCBCR_601,
YCBCR_709
} YCbCrType;
#endif /* YUV_RGB_COMMON_H_ */

View file

@ -0,0 +1,74 @@
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
#define PRECISION 6
#define PRECISION_FACTOR (1<<PRECISION)
typedef struct
{
uint8_t y_shift;
int16_t matrix[3][3];
} RGB2YUVParam;
// |Y| |y_shift| |matrix[0][0] matrix[0][1] matrix[0][2]| |R|
// |U| = | 128 | + 1/PRECISION_FACTOR * |matrix[1][0] matrix[1][1] matrix[1][2]| * |G|
// |V| | 128 | |matrix[2][0] matrix[2][1] matrix[2][2]| |B|
typedef struct
{
uint8_t y_shift;
int16_t y_factor;
int16_t v_r_factor;
int16_t u_g_factor;
int16_t v_g_factor;
int16_t u_b_factor;
} YUV2RGBParam;
// |R| |y_factor 0 v_r_factor| |Y-y_shift|
// |G| = 1/PRECISION_FACTOR * |y_factor u_g_factor v_g_factor| * | U-128 |
// |B| |y_factor u_b_factor 0 | | V-128 |
#ifdef _MSC_VER
#pragma warning(push)
#pragma warning(disable : 26451)
#endif
#define V(value) (int16_t)((value*PRECISION_FACTOR)+0.5)
// for ITU-T T.871, values can be found in section 7
// for ITU-R BT.601-7 values are derived from equations in sections 2.5.1-2.5.3, assuming RGB is encoded using full range ([0-1]<->[0-255])
// for ITU-R BT.709-6 values are derived from equations in sections 3.2-3.4, assuming RGB is encoded using full range ([0-1]<->[0-255])
// all values are rounded to the fourth decimal
static const YUV2RGBParam YUV2RGB[3] = {
// ITU-T T.871 (JPEG)
{/*.y_shift=*/ 0, /*.y_factor=*/ V(1.0), /*.v_r_factor=*/ V(1.402), /*.u_g_factor=*/ -V(0.3441), /*.v_g_factor=*/ -V(0.7141), /*.u_b_factor=*/ V(1.772)},
// ITU-R BT.601-7
{/*.y_shift=*/ 16, /*.y_factor=*/ V(1.1644), /*.v_r_factor=*/ V(1.596), /*.u_g_factor=*/ -V(0.3918), /*.v_g_factor=*/ -V(0.813), /*.u_b_factor=*/ V(2.0172)},
// ITU-R BT.709-6
{/*.y_shift=*/ 16, /*.y_factor=*/ V(1.1644), /*.v_r_factor=*/ V(1.7927), /*.u_g_factor=*/ -V(0.2132), /*.v_g_factor=*/ -V(0.5329), /*.u_b_factor=*/ V(2.1124)}
};
static const RGB2YUVParam RGB2YUV[3] = {
// ITU-T T.871 (JPEG)
{/*.y_shift=*/ 0, /*.matrix=*/ {{V(0.299), V(0.587), V(0.114)}, {-V(0.1687), -V(0.3313), V(0.5)}, {V(0.5), -V(0.4187), -V(0.0813)}}},
// ITU-R BT.601-7
{/*.y_shift=*/ 16, /*.matrix=*/ {{V(0.2568), V(0.5041), V(0.0979)}, {-V(0.1482), -V(0.291), V(0.4392)}, {V(0.4392), -V(0.3678), -V(0.0714)}}},
// ITU-R BT.709-6
{/*.y_shift=*/ 16, /*.matrix=*/ {{V(0.1826), V(0.6142), V(0.062)}, {-V(0.1006), -V(0.3386), V(0.4392)}, {V(0.4392), -V(0.3989), -V(0.0403)}}}
};
#ifdef _MSC_VER
#pragma warning(pop)
#endif
/* The various layouts of YUV data we support */
#define YUV_FORMAT_420 1
#define YUV_FORMAT_422 2
#define YUV_FORMAT_NV12 3
/* The various formats of RGB pixel that we support */
#define RGB_FORMAT_RGB565 1
#define RGB_FORMAT_RGB24 2
#define RGB_FORMAT_RGBA 3
#define RGB_FORMAT_BGRA 4
#define RGB_FORMAT_ARGB 5
#define RGB_FORMAT_ABGR 6

View file

@ -0,0 +1,44 @@
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
#include "../../SDL_internal.h"
#if SDL_HAVE_YUV
#include "yuv_rgb.h"
#include "SDL_cpuinfo.h"
#ifdef __loongarch_sx
#define LSX_FUNCTION_NAME yuv420_rgb24_lsx
#define STD_FUNCTION_NAME yuv420_rgb24_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_rgba_lsx
#define STD_FUNCTION_NAME yuv420_rgba_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_bgra_lsx
#define STD_FUNCTION_NAME yuv420_bgra_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_argb_lsx
#define STD_FUNCTION_NAME yuv420_argb_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_abgr_lsx
#define STD_FUNCTION_NAME yuv420_abgr_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_lsx_func.h"
#endif //__loongarch_sx
#endif /* SDL_HAVE_YUV */

View file

@ -0,0 +1,407 @@
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
// Provide optimized functions to convert images from 8bits yuv420 to rgb24 format
// There are a few slightly different variations of the YCbCr color space with different parameters that
// change the conversion matrix.
// The three most common YCbCr color space, defined by BT.601, BT.709 and JPEG standard are implemented here.
// See the respective standards for details
// The matrix values used are derived from http://www.equasys.de/colorconversion.html
// YUV420 is stored as three separate channels, with U and V (Cb and Cr) subsampled by a 2 factor
// For conversion from yuv to rgb, no interpolation is done, and the same UV value are used for 4 rgb pixels. This
// is suboptimal for image quality, but by far the fastest method.
// For all methods, width and height should be even, if not, the last row/column of the result image won't be affected.
// For sse methods, if the width if not divisable by 32, the last (width%32) pixels of each line won't be affected.
/*#include <stdint.h>*/
#include "yuv_rgb_common.h"
#include "SDL_stdinc.h"
// yuv to rgb, standard c implementation
void yuv420_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// yuv to rgb, sse implementation
// pointers must be 16 byte aligned, and strides must be divisable by 16
void yuv420_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// yuv to rgb, sse implementation
// pointers do not need to be 16 byte aligned
void yuv420_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// rgb to yuv, standard c implementation
void rgb24_yuv420_std(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
// rgb to yuv, sse implementation
// pointers must be 16 byte aligned, and strides must be divisible by 16
void rgb24_yuv420_sse(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
// rgb to yuv, sse implementation
// pointers do not need to be 16 byte aligned
void rgb24_yuv420_sseu(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
//yuv420 to bgra, lsx implementation
void yuv420_rgb24_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_lsx(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *v, const uint8_t *u, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);

View file

@ -3,254 +3,12 @@
#include "../../SDL_internal.h" #include "../../SDL_internal.h"
#if SDL_HAVE_YUV #if SDL_HAVE_YUV
#include "yuv_rgb.h" #include "yuv_rgb.h"
#include "yuv_rgb_internal.h"
#include "SDL_cpuinfo.h" #include "SDL_cpuinfo.h"
/*#include <x86intrin.h>*/ /*#include <x86intrin.h>*/
#define PRECISION 6
#define PRECISION_FACTOR (1<<PRECISION)
typedef struct
{
uint8_t y_shift;
int16_t matrix[3][3];
} RGB2YUVParam;
// |Y| |y_shift| |matrix[0][0] matrix[0][1] matrix[0][2]| |R|
// |U| = | 128 | + 1/PRECISION_FACTOR * |matrix[1][0] matrix[1][1] matrix[1][2]| * |G|
// |V| | 128 | |matrix[2][0] matrix[2][1] matrix[2][2]| |B|
typedef struct
{
uint8_t y_shift;
int16_t y_factor;
int16_t v_r_factor;
int16_t u_g_factor;
int16_t v_g_factor;
int16_t u_b_factor;
} YUV2RGBParam;
// |R| |y_factor 0 v_r_factor| |Y-y_shift|
// |G| = 1/PRECISION_FACTOR * |y_factor u_g_factor v_g_factor| * | U-128 |
// |B| |y_factor u_b_factor 0 | | V-128 |
#ifdef _MSC_VER
#pragma warning(push)
#pragma warning(disable : 26451)
#endif
#define V(value) (int16_t)((value*PRECISION_FACTOR)+0.5)
// for ITU-T T.871, values can be found in section 7
// for ITU-R BT.601-7 values are derived from equations in sections 2.5.1-2.5.3, assuming RGB is encoded using full range ([0-1]<->[0-255])
// for ITU-R BT.709-6 values are derived from equations in sections 3.2-3.4, assuming RGB is encoded using full range ([0-1]<->[0-255])
// all values are rounded to the fourth decimal
static const YUV2RGBParam YUV2RGB[3] = {
// ITU-T T.871 (JPEG)
{/*.y_shift=*/ 0, /*.y_factor=*/ V(1.0), /*.v_r_factor=*/ V(1.402), /*.u_g_factor=*/ -V(0.3441), /*.v_g_factor=*/ -V(0.7141), /*.u_b_factor=*/ V(1.772)},
// ITU-R BT.601-7
{/*.y_shift=*/ 16, /*.y_factor=*/ V(1.1644), /*.v_r_factor=*/ V(1.596), /*.u_g_factor=*/ -V(0.3918), /*.v_g_factor=*/ -V(0.813), /*.u_b_factor=*/ V(2.0172)},
// ITU-R BT.709-6
{/*.y_shift=*/ 16, /*.y_factor=*/ V(1.1644), /*.v_r_factor=*/ V(1.7927), /*.u_g_factor=*/ -V(0.2132), /*.v_g_factor=*/ -V(0.5329), /*.u_b_factor=*/ V(2.1124)}
};
static const RGB2YUVParam RGB2YUV[3] = {
// ITU-T T.871 (JPEG)
{/*.y_shift=*/ 0, /*.matrix=*/ {{V(0.299), V(0.587), V(0.114)}, {-V(0.1687), -V(0.3313), V(0.5)}, {V(0.5), -V(0.4187), -V(0.0813)}}},
// ITU-R BT.601-7
{/*.y_shift=*/ 16, /*.matrix=*/ {{V(0.2568), V(0.5041), V(0.0979)}, {-V(0.1482), -V(0.291), V(0.4392)}, {V(0.4392), -V(0.3678), -V(0.0714)}}},
// ITU-R BT.709-6
{/*.y_shift=*/ 16, /*.matrix=*/ {{V(0.1826), V(0.6142), V(0.062)}, {-V(0.1006), -V(0.3386), V(0.4392)}, {V(0.4392), -V(0.3989), -V(0.0403)}}}
};
#ifdef _MSC_VER
#pragma warning(pop)
#endif
/* The various layouts of YUV data we support */
#define YUV_FORMAT_420 1
#define YUV_FORMAT_422 2
#define YUV_FORMAT_NV12 3
/* The various formats of RGB pixel that we support */
#define RGB_FORMAT_RGB565 1
#define RGB_FORMAT_RGB24 2
#define RGB_FORMAT_RGBA 3
#define RGB_FORMAT_BGRA 4
#define RGB_FORMAT_ARGB 5
#define RGB_FORMAT_ABGR 6
// divide by PRECISION_FACTOR and clamp to [0:255] interval
// input must be in the [-128*PRECISION_FACTOR:384*PRECISION_FACTOR] range
static uint8_t clampU8(int32_t v)
{
static const uint8_t lut[512] =
{0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,
47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,
91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,
126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,
159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,
192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,
225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255
};
return lut[((v+128*PRECISION_FACTOR)>>PRECISION)&511];
}
#define STD_FUNCTION_NAME yuv420_rgb565_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGB565
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_rgb24_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_rgba_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_bgra_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_argb_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_abgr_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_rgb565_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_RGB565
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_rgb24_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_rgba_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_bgra_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_argb_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_abgr_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_rgb565_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_RGB565
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_rgb24_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_rgba_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_bgra_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_argb_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_abgr_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_std_func.h"
void rgb24_yuv420_std(
uint32_t width, uint32_t height,
const uint8_t *RGB, uint32_t RGB_stride,
uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride,
YCbCrType yuv_type)
{
const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
uint32_t x, y;
for(y=0; y<(height-1); y+=2)
{
const uint8_t *rgb_ptr1=RGB+y*RGB_stride,
*rgb_ptr2=RGB+(y+1)*RGB_stride;
uint8_t *y_ptr1=Y+y*Y_stride,
*y_ptr2=Y+(y+1)*Y_stride,
*u_ptr=U+(y/2)*UV_stride,
*v_ptr=V+(y/2)*UV_stride;
for(x=0; x<(width-1); x+=2)
{
// compute yuv for the four pixels, u and v values are summed
int32_t y_tmp, u_tmp, v_tmp;
y_tmp = param->matrix[0][0]*rgb_ptr1[0] + param->matrix[0][1]*rgb_ptr1[1] + param->matrix[0][2]*rgb_ptr1[2];
u_tmp = param->matrix[1][0]*rgb_ptr1[0] + param->matrix[1][1]*rgb_ptr1[1] + param->matrix[1][2]*rgb_ptr1[2];
v_tmp = param->matrix[2][0]*rgb_ptr1[0] + param->matrix[2][1]*rgb_ptr1[1] + param->matrix[2][2]*rgb_ptr1[2];
y_ptr1[0]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
y_tmp = param->matrix[0][0]*rgb_ptr1[3] + param->matrix[0][1]*rgb_ptr1[4] + param->matrix[0][2]*rgb_ptr1[5];
u_tmp += param->matrix[1][0]*rgb_ptr1[3] + param->matrix[1][1]*rgb_ptr1[4] + param->matrix[1][2]*rgb_ptr1[5];
v_tmp += param->matrix[2][0]*rgb_ptr1[3] + param->matrix[2][1]*rgb_ptr1[4] + param->matrix[2][2]*rgb_ptr1[5];
y_ptr1[1]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
y_tmp = param->matrix[0][0]*rgb_ptr2[0] + param->matrix[0][1]*rgb_ptr2[1] + param->matrix[0][2]*rgb_ptr2[2];
u_tmp += param->matrix[1][0]*rgb_ptr2[0] + param->matrix[1][1]*rgb_ptr2[1] + param->matrix[1][2]*rgb_ptr2[2];
v_tmp += param->matrix[2][0]*rgb_ptr2[0] + param->matrix[2][1]*rgb_ptr2[1] + param->matrix[2][2]*rgb_ptr2[2];
y_ptr2[0]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
y_tmp = param->matrix[0][0]*rgb_ptr2[3] + param->matrix[0][1]*rgb_ptr2[4] + param->matrix[0][2]*rgb_ptr2[5];
u_tmp += param->matrix[1][0]*rgb_ptr2[3] + param->matrix[1][1]*rgb_ptr2[4] + param->matrix[1][2]*rgb_ptr2[5];
v_tmp += param->matrix[2][0]*rgb_ptr2[3] + param->matrix[2][1]*rgb_ptr2[4] + param->matrix[2][2]*rgb_ptr2[5];
y_ptr2[1]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
u_ptr[0] = clampU8(u_tmp/4+(128<<PRECISION));
v_ptr[0] = clampU8(v_tmp/4+(128<<PRECISION));
rgb_ptr1 += 6;
rgb_ptr2 += 6;
y_ptr1 += 2;
y_ptr2 += 2;
u_ptr += 1;
v_ptr += 1;
}
}
}
#ifdef __SSE2__ #ifdef __SSE2__
/* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan. /* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan.
@ -516,17 +274,17 @@ UNPACK_RGB24_32_STEP1(RGB1, RGB2, RGB3, RGB4, RGB5, RGB6, R1, R2, G1, G2, B1, B2
#define RGB2YUV_16(R, G, B, Y, U, V) \ #define RGB2YUV_16(R, G, B, Y, U, V) \
Y = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[0][0])), \ Y = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[0][0])), \
_mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[0][1]))); \ _mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[0][1]))); \
Y = _mm_add_epi16(Y, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[0][2]))); \ Y = _mm_add_epi16(Y, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[0][2]))); \
Y = _mm_add_epi16(Y, _mm_set1_epi16((param->y_shift)<<PRECISION)); \ Y = _mm_add_epi16(Y, _mm_set1_epi16((param->y_shift)<<PRECISION)); \
Y = _mm_srai_epi16(Y, PRECISION); \ Y = _mm_srai_epi16(Y, PRECISION); \
U = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[1][0])), \ U = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[1][0])), \
_mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[1][1]))); \ _mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[1][1]))); \
U = _mm_add_epi16(U, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[1][2]))); \ U = _mm_add_epi16(U, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[1][2]))); \
U = _mm_add_epi16(U, _mm_set1_epi16(128<<PRECISION)); \ U = _mm_add_epi16(U, _mm_set1_epi16(128<<PRECISION)); \
U = _mm_srai_epi16(U, PRECISION); \ U = _mm_srai_epi16(U, PRECISION); \
V = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[2][0])), \ V = _mm_add_epi16(_mm_mullo_epi16(R, _mm_set1_epi16(param->matrix[2][0])), \
_mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[2][1]))); \ _mm_mullo_epi16(G, _mm_set1_epi16(param->matrix[2][1]))); \
V = _mm_add_epi16(V, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[2][2]))); \ V = _mm_add_epi16(V, _mm_mullo_epi16(B, _mm_set1_epi16(param->matrix[2][2]))); \
V = _mm_add_epi16(V, _mm_set1_epi16(128<<PRECISION)); \ V = _mm_add_epi16(V, _mm_set1_epi16(128<<PRECISION)); \
V = _mm_srai_epi16(V, PRECISION); V = _mm_srai_epi16(V, PRECISION);
@ -629,112 +387,78 @@ V = _mm_srai_epi16(V, PRECISION);
/* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan. /* SDL doesn't use these atm and compiling them adds seconds onto the build. --ryan.
void rgb24_yuv420_sse(uint32_t width, uint32_t height, void rgb24_yuv420_sse(uint32_t width, uint32_t height,
const uint8_t *RGB, uint32_t RGB_stride, const uint8_t *RGB, uint32_t RGB_stride,
uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride,
YCbCrType yuv_type) YCbCrType yuv_type)
{ {
#define LOAD_SI128 _mm_load_si128 #define LOAD_SI128 _mm_load_si128
#define SAVE_SI128 _mm_stream_si128 #define SAVE_SI128 _mm_stream_si128
const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]); const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
uint32_t xpos, ypos; uint32_t xpos, ypos;
for(ypos=0; ypos<(height-1); ypos+=2) for(ypos=0; ypos<(height-1); ypos+=2)
{ {
const uint8_t *rgb_ptr1=RGB+ypos*RGB_stride, const uint8_t *rgb_ptr1=RGB+ypos*RGB_stride,
*rgb_ptr2=RGB+(ypos+1)*RGB_stride; *rgb_ptr2=RGB+(ypos+1)*RGB_stride;
uint8_t *y_ptr1=Y+ypos*Y_stride, uint8_t *y_ptr1=Y+ypos*Y_stride,
*y_ptr2=Y+(ypos+1)*Y_stride, *y_ptr2=Y+(ypos+1)*Y_stride,
*u_ptr=U+(ypos/2)*UV_stride, *u_ptr=U+(ypos/2)*UV_stride,
*v_ptr=V+(ypos/2)*UV_stride; *v_ptr=V+(ypos/2)*UV_stride;
for(xpos=0; xpos<(width-31); xpos+=32) for(xpos=0; xpos<(width-31); xpos+=32)
{ {
RGB2YUV_32 RGB2YUV_32
rgb_ptr1+=96; rgb_ptr1+=96;
rgb_ptr2+=96; rgb_ptr2+=96;
y_ptr1+=32; y_ptr1+=32;
y_ptr2+=32; y_ptr2+=32;
u_ptr+=16; u_ptr+=16;
v_ptr+=16; v_ptr+=16;
} }
} }
#undef LOAD_SI128 #undef LOAD_SI128
#undef SAVE_SI128 #undef SAVE_SI128
} }
void rgb24_yuv420_sseu(uint32_t width, uint32_t height, void rgb24_yuv420_sseu(uint32_t width, uint32_t height,
const uint8_t *RGB, uint32_t RGB_stride, const uint8_t *RGB, uint32_t RGB_stride,
uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride, uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride,
YCbCrType yuv_type) YCbCrType yuv_type)
{ {
#define LOAD_SI128 _mm_loadu_si128 #define LOAD_SI128 _mm_loadu_si128
#define SAVE_SI128 _mm_storeu_si128 #define SAVE_SI128 _mm_storeu_si128
const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]); const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
uint32_t xpos, ypos; uint32_t xpos, ypos;
for(ypos=0; ypos<(height-1); ypos+=2) for(ypos=0; ypos<(height-1); ypos+=2)
{ {
const uint8_t *rgb_ptr1=RGB+ypos*RGB_stride, const uint8_t *rgb_ptr1=RGB+ypos*RGB_stride,
*rgb_ptr2=RGB+(ypos+1)*RGB_stride; *rgb_ptr2=RGB+(ypos+1)*RGB_stride;
uint8_t *y_ptr1=Y+ypos*Y_stride, uint8_t *y_ptr1=Y+ypos*Y_stride,
*y_ptr2=Y+(ypos+1)*Y_stride, *y_ptr2=Y+(ypos+1)*Y_stride,
*u_ptr=U+(ypos/2)*UV_stride, *u_ptr=U+(ypos/2)*UV_stride,
*v_ptr=V+(ypos/2)*UV_stride; *v_ptr=V+(ypos/2)*UV_stride;
for(xpos=0; xpos<(width-31); xpos+=32) for(xpos=0; xpos<(width-31); xpos+=32)
{ {
RGB2YUV_32 RGB2YUV_32
rgb_ptr1+=96; rgb_ptr1+=96;
rgb_ptr2+=96; rgb_ptr2+=96;
y_ptr1+=32; y_ptr1+=32;
y_ptr2+=32; y_ptr2+=32;
u_ptr+=16; u_ptr+=16;
v_ptr+=16; v_ptr+=16;
} }
} }
#undef LOAD_SI128 #undef LOAD_SI128
#undef SAVE_SI128 #undef SAVE_SI128
} }
*/ */
#endif //__SSE2__ #endif //__SSE2__
#ifdef __loongarch_sx
#define LSX_FUNCTION_NAME yuv420_rgb24_lsx
#define STD_FUNCTION_NAME yuv420_rgb24_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_rgba_lsx
#define STD_FUNCTION_NAME yuv420_rgba_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_bgra_lsx
#define STD_FUNCTION_NAME yuv420_bgra_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_argb_lsx
#define STD_FUNCTION_NAME yuv420_argb_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_lsx_func.h"
#define LSX_FUNCTION_NAME yuv420_abgr_lsx
#define STD_FUNCTION_NAME yuv420_abgr_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_lsx_func.h"
#endif //__loongarch_sx
#endif /* SDL_HAVE_YUV */ #endif /* SDL_HAVE_YUV */

View file

@ -0,0 +1,266 @@
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
// Provide optimized functions to convert images from 8bits yuv420 to rgb24 format
// There are a few slightly different variations of the YCbCr color space with different parameters that
// change the conversion matrix.
// The three most common YCbCr color space, defined by BT.601, BT.709 and JPEG standard are implemented here.
// See the respective standards for details
// The matrix values used are derived from http://www.equasys.de/colorconversion.html
// YUV420 is stored as three separate channels, with U and V (Cb and Cr) subsampled by a 2 factor
// For conversion from yuv to rgb, no interpolation is done, and the same UV value are used for 4 rgb pixels. This
// is suboptimal for image quality, but by far the fastest method.
// For all methods, width and height should be even, if not, the last row/column of the result image won't be affected.
// For sse methods, if the width if not divisable by 32, the last (width%32) pixels of each line won't be affected.
/*#include <stdint.h>*/
#include "yuv_rgb_common.h"
#include "SDL_stdinc.h"
// yuv to rgb, sse implementation
// pointers must be 16 byte aligned, and strides must be divisable by 16
void yuv420_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_sse(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// yuv to rgb, sse implementation
// pointers do not need to be 16 byte aligned
void yuv420_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_sseu(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
// rgb to yuv, standard c implementation
void rgb24_yuv420_std(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
// rgb to yuv, sse implementation
// pointers must be 16 byte aligned, and strides must be divisible by 16
void rgb24_yuv420_sse(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);
// rgb to yuv, sse implementation
// pointers do not need to be 16 byte aligned
void rgb24_yuv420_sseu(
uint32_t width, uint32_t height,
const uint8_t *rgb, uint32_t rgb_stride,
uint8_t *y, uint8_t *u, uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
YCbCrType yuv_type);

View file

@ -0,0 +1,179 @@
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
#include "../../SDL_internal.h"
#if SDL_HAVE_YUV
#include "yuv_rgb.h"
#include "yuv_rgb_internal.h"
// divide by PRECISION_FACTOR and clamp to [0:255] interval
// input must be in the [-128*PRECISION_FACTOR:384*PRECISION_FACTOR] range
static uint8_t clampU8(int32_t v)
{
static const uint8_t lut[512] =
{0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,
0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,
47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,
91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,
126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,
159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,
192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,
225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,
255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255
};
return lut[((v+128*PRECISION_FACTOR)>>PRECISION)&511];
}
#define STD_FUNCTION_NAME yuv420_rgb565_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGB565
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_rgb24_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_rgba_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_bgra_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_argb_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv420_abgr_std
#define YUV_FORMAT YUV_FORMAT_420
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_rgb565_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_RGB565
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_rgb24_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_rgba_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_bgra_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_argb_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuv422_abgr_std
#define YUV_FORMAT YUV_FORMAT_422
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_rgb565_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_RGB565
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_rgb24_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_RGB24
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_rgba_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_RGBA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_bgra_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_BGRA
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_argb_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_ARGB
#include "yuv_rgb_std_func.h"
#define STD_FUNCTION_NAME yuvnv12_abgr_std
#define YUV_FORMAT YUV_FORMAT_NV12
#define RGB_FORMAT RGB_FORMAT_ABGR
#include "yuv_rgb_std_func.h"
void rgb24_yuv420_std(
uint32_t width, uint32_t height,
const uint8_t *RGB, uint32_t RGB_stride,
uint8_t *Y, uint8_t *U, uint8_t *V, uint32_t Y_stride, uint32_t UV_stride,
YCbCrType yuv_type)
{
const RGB2YUVParam *const param = &(RGB2YUV[yuv_type]);
uint32_t x, y;
for(y=0; y<(height-1); y+=2)
{
const uint8_t *rgb_ptr1=RGB+y*RGB_stride,
*rgb_ptr2=RGB+(y+1)*RGB_stride;
uint8_t *y_ptr1=Y+y*Y_stride,
*y_ptr2=Y+(y+1)*Y_stride,
*u_ptr=U+(y/2)*UV_stride,
*v_ptr=V+(y/2)*UV_stride;
for(x=0; x<(width-1); x+=2)
{
// compute yuv for the four pixels, u and v values are summed
int32_t y_tmp, u_tmp, v_tmp;
y_tmp = param->matrix[0][0]*rgb_ptr1[0] + param->matrix[0][1]*rgb_ptr1[1] + param->matrix[0][2]*rgb_ptr1[2];
u_tmp = param->matrix[1][0]*rgb_ptr1[0] + param->matrix[1][1]*rgb_ptr1[1] + param->matrix[1][2]*rgb_ptr1[2];
v_tmp = param->matrix[2][0]*rgb_ptr1[0] + param->matrix[2][1]*rgb_ptr1[1] + param->matrix[2][2]*rgb_ptr1[2];
y_ptr1[0]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
y_tmp = param->matrix[0][0]*rgb_ptr1[3] + param->matrix[0][1]*rgb_ptr1[4] + param->matrix[0][2]*rgb_ptr1[5];
u_tmp += param->matrix[1][0]*rgb_ptr1[3] + param->matrix[1][1]*rgb_ptr1[4] + param->matrix[1][2]*rgb_ptr1[5];
v_tmp += param->matrix[2][0]*rgb_ptr1[3] + param->matrix[2][1]*rgb_ptr1[4] + param->matrix[2][2]*rgb_ptr1[5];
y_ptr1[1]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
y_tmp = param->matrix[0][0]*rgb_ptr2[0] + param->matrix[0][1]*rgb_ptr2[1] + param->matrix[0][2]*rgb_ptr2[2];
u_tmp += param->matrix[1][0]*rgb_ptr2[0] + param->matrix[1][1]*rgb_ptr2[1] + param->matrix[1][2]*rgb_ptr2[2];
v_tmp += param->matrix[2][0]*rgb_ptr2[0] + param->matrix[2][1]*rgb_ptr2[1] + param->matrix[2][2]*rgb_ptr2[2];
y_ptr2[0]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
y_tmp = param->matrix[0][0]*rgb_ptr2[3] + param->matrix[0][1]*rgb_ptr2[4] + param->matrix[0][2]*rgb_ptr2[5];
u_tmp += param->matrix[1][0]*rgb_ptr2[3] + param->matrix[1][1]*rgb_ptr2[4] + param->matrix[1][2]*rgb_ptr2[5];
v_tmp += param->matrix[2][0]*rgb_ptr2[3] + param->matrix[2][1]*rgb_ptr2[4] + param->matrix[2][2]*rgb_ptr2[5];
y_ptr2[1]=clampU8(y_tmp+((param->y_shift)<<PRECISION));
u_ptr[0] = clampU8(u_tmp/4+(128<<PRECISION));
v_ptr[0] = clampU8(v_tmp/4+(128<<PRECISION));
rgb_ptr1 += 6;
rgb_ptr2 += 6;
y_ptr1 += 2;
y_ptr2 += 2;
u_ptr += 1;
v_ptr += 1;
}
}
}
#endif /* SDL_HAVE_YUV */

View file

@ -0,0 +1,131 @@
// Copyright 2016 Adrien Descamps
// Distributed under BSD 3-Clause License
// Provide optimized functions to convert images from 8bits yuv420 to rgb24 format
// There are a few slightly different variations of the YCbCr color space with different parameters that
// change the conversion matrix.
// The three most common YCbCr color space, defined by BT.601, BT.709 and JPEG standard are implemented here.
// See the respective standards for details
// The matrix values used are derived from http://www.equasys.de/colorconversion.html
// YUV420 is stored as three separate channels, with U and V (Cb and Cr) subsampled by a 2 factor
// For conversion from yuv to rgb, no interpolation is done, and the same UV value are used for 4 rgb pixels. This
// is suboptimal for image quality, but by far the fastest method.
// For all methods, width and height should be even, if not, the last row/column of the result image won't be affected.
// For sse methods, if the width if not divisable by 32, the last (width%32) pixels of each line won't be affected.
/*#include <stdint.h>*/
#include "yuv_rgb_common.h"
#include "SDL_stdinc.h"
// yuv to rgb, standard c implementation
void yuv420_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv420_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuv422_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb565_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgb24_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_rgba_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_bgra_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_argb_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);
void yuvnv12_abgr_std(
uint32_t width, uint32_t height,
const uint8_t *y, const uint8_t *u, const uint8_t *v, uint32_t y_stride, uint32_t uv_stride,
uint8_t *rgb, uint32_t rgb_stride,
YCbCrType yuv_type);