0
0
mirror of https://github.com/mpv-player/mpv.git synced 2024-09-20 20:03:10 +02:00
mpv/video/out/opengl/common.h

225 lines
10 KiB
C
Raw Normal View History

/*
* This file is part of mpv.
*
* mpv is free software; you can redistribute it and/or
* modify it under the terms of the GNU Lesser General Public
* License as published by the Free Software Foundation; either
* version 2.1 of the License, or (at your option) any later version.
*
* mpv is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU Lesser General Public License for more details.
*
* You should have received a copy of the GNU Lesser General Public
* License along with mpv. If not, see <http://www.gnu.org/licenses/>.
*/
#ifndef MPLAYER_GL_COMMON_H
#define MPLAYER_GL_COMMON_H
#include <stdio.h>
#include <stdint.h>
#include <stdbool.h>
#include "config.h"
#include "common/msg.h"
#include "misc/bstr.h"
#include "video/out/vo.h"
#include "video/csputils.h"
#include "video/mp_image.h"
#if HAVE_GL_COCOA
#define GL_DO_NOT_WARN_IF_MULTI_GL_VERSION_HEADERS_INCLUDED 1
#include <OpenGL/gl.h>
#include <OpenGL/gl3.h>
#include <OpenGL/glext.h>
#elif HAVE_ANDROID_GL
#include <GLES3/gl3.h>
#else
#include <GL/gl.h>
#include <GL/glext.h>
#endif
#define MP_GET_GL_WORKAROUNDS
#include "header_fixes.h"
struct GL;
typedef struct GL GL;
enum {
MPGL_CAP_ROW_LENGTH = (1 << 4), // GL_[UN]PACK_ROW_LENGTH
MPGL_CAP_FB = (1 << 5),
MPGL_CAP_VAO = (1 << 6),
MPGL_CAP_FLOAT_TEX = (1 << 9),
MPGL_CAP_TEX_RG = (1 << 10), // GL_ARB_texture_rg / GL 3.x
MPGL_CAP_VDPAU = (1 << 11), // GL_NV_vdpau_interop
MPGL_CAP_APPLE_RGB_422 = (1 << 12), // GL_APPLE_rgb_422
MPGL_CAP_1D_TEX = (1 << 14),
MPGL_CAP_3D_TEX = (1 << 15),
MPGL_CAP_DEBUG = (1 << 16),
MPGL_CAP_DXINTEROP = (1 << 17), // WGL_NV_DX_interop
MPGL_CAP_EXT16 = (1 << 18), // GL_EXT_texture_norm16
MPGL_CAP_SW = (1 << 30), // indirect or sw renderer
};
// E.g. 310 means 3.1
// Code doesn't have to use the macros; they are for convenience only.
#define MPGL_VER(major, minor) (((major) * 100) + (minor) * 10)
#define MPGL_VER_GET_MAJOR(ver) ((unsigned)(ver) / 100)
#define MPGL_VER_GET_MINOR(ver) ((unsigned)(ver) % 100 / 10)
#define MPGL_VER_P(ver) MPGL_VER_GET_MAJOR(ver), MPGL_VER_GET_MINOR(ver)
void mpgl_load_functions(GL *gl, void *(*getProcAddress)(const GLubyte *),
const char *ext2, struct mp_log *log);
void mpgl_load_functions2(GL *gl, void *(*get_fn)(void *ctx, const char *n),
void *fn_ctx, const char *ext2, struct mp_log *log);
typedef void (GLAPIENTRY *MP_GLDEBUGPROC)(GLenum, GLenum, GLuint, GLenum,
GLsizei, const GLchar *,const void *);
//function pointers loaded from the OpenGL library
struct GL {
int version; // MPGL_VER() mangled (e.g. 210 for 2.1)
int es; // es version (e.g. 300), 0 for desktop GL
int glsl_version; // e.g. 130 for GLSL 1.30
char *extensions; // Equivalent to GL_EXTENSIONS
int mpgl_caps; // Bitfield of MPGL_CAP_* constants
bool debug_context; // use of e.g. GLX_CONTEXT_DEBUG_BIT_ARB
int fb_r, fb_g, fb_b; // frame buffer bit depth (0 if unknown)
void (GLAPIENTRY *Viewport)(GLint, GLint, GLsizei, GLsizei);
void (GLAPIENTRY *Clear)(GLbitfield);
void (GLAPIENTRY *GenTextures)(GLsizei, GLuint *);
void (GLAPIENTRY *DeleteTextures)(GLsizei, const GLuint *);
void (GLAPIENTRY *ClearColor)(GLclampf, GLclampf, GLclampf, GLclampf);
void (GLAPIENTRY *Enable)(GLenum);
void (GLAPIENTRY *Disable)(GLenum);
const GLubyte *(GLAPIENTRY * GetString)(GLenum);
void (GLAPIENTRY *DrawBuffer)(GLenum);
void (GLAPIENTRY *BlendFuncSeparate)(GLenum, GLenum, GLenum, GLenum);
void (GLAPIENTRY *Flush)(void);
void (GLAPIENTRY *Finish)(void);
void (GLAPIENTRY *PixelStorei)(GLenum, GLint);
void (GLAPIENTRY *TexImage1D)(GLenum, GLint, GLint, GLsizei, GLint,
GLenum, GLenum, const GLvoid *);
void (GLAPIENTRY *TexImage2D)(GLenum, GLint, GLint, GLsizei, GLsizei,
GLint, GLenum, GLenum, const GLvoid *);
void (GLAPIENTRY *TexSubImage2D)(GLenum, GLint, GLint, GLint,
GLsizei, GLsizei, GLenum, GLenum,
const GLvoid *);
void (GLAPIENTRY *TexParameteri)(GLenum, GLenum, GLint);
void (GLAPIENTRY *GetIntegerv)(GLenum, GLint *);
void (GLAPIENTRY *ReadPixels)(GLint, GLint, GLsizei, GLsizei, GLenum,
GLenum, GLvoid *);
void (GLAPIENTRY *ReadBuffer)(GLenum);
void (GLAPIENTRY *DrawArrays)(GLenum, GLint, GLsizei);
GLenum (GLAPIENTRY *GetError)(void);
void (GLAPIENTRY *GetTexLevelParameteriv)(GLenum, GLint, GLenum, GLint *);
void (GLAPIENTRY *GenBuffers)(GLsizei, GLuint *);
void (GLAPIENTRY *DeleteBuffers)(GLsizei, const GLuint *);
void (GLAPIENTRY *BindBuffer)(GLenum, GLuint);
vo_opengl: implement NNEDI3 prescaler Implement NNEDI3, a neural network based deinterlacer. The shader is reimplemented in GLSL and supports both 8x4 and 8x6 sampling window now. This allows the shader to be licensed under LGPL2.1 so that it can be used in mpv. The current implementation supports uploading the NN weights (up to 51kb with placebo setting) in two different way, via uniform buffer object or hard coding into shader source. UBO requires OpenGL 3.1, which only guarantee 16kb per block. But I find that 64kb seems to be a default setting for recent card/driver (which nnedi3 is targeting), so I think we're fine here (with default nnedi3 setting the size of weights is 9kb). Hard-coding into shader requires OpenGL 3.3, for the "intBitsToFloat()" built-in function. This is necessary to precisely represent these weights in GLSL. I tried several human readable floating point number format (with really high precision as for single precision float), but for some reason they are not working nicely, bad pixels (with NaN value) could be produced with some weights set. We could also add support to upload these weights with texture, just for compatibility reason (etc. upscaling a still image with a low end graphics card). But as I tested, it's rather slow even with 1D texture (we probably had to use 2D texture due to dimension size limitation). Since there is always better choice to do NNEDI3 upscaling for still image (vapoursynth plugin), it's not implemented in this commit. If this turns out to be a popular demand from the user, it should be easy to add it later. For those who wants to optimize the performance a bit further, the bottleneck seems to be: 1. overhead to upload and access these weights, (in particular, the shader code will be regenerated for each frame, it's on CPU though). 2. "dot()" performance in the main loop. 3. "exp()" performance in the main loop, there are various fast implementation with some bit tricks (probably with the help of the intBitsToFloat function). The code is tested with nvidia card and driver (355.11), on Linux. Closes #2230
2015-10-28 02:37:55 +01:00
void (GLAPIENTRY *BindBufferBase)(GLenum, GLuint, GLuint);
GLvoid * (GLAPIENTRY * MapBuffer)(GLenum, GLenum);
GLboolean (GLAPIENTRY *UnmapBuffer)(GLenum);
void (GLAPIENTRY *BufferData)(GLenum, intptr_t, const GLvoid *, GLenum);
void (GLAPIENTRY *ActiveTexture)(GLenum);
void (GLAPIENTRY *BindTexture)(GLenum, GLuint);
int (GLAPIENTRY *SwapInterval)(int);
void (GLAPIENTRY *TexImage3D)(GLenum, GLint, GLenum, GLsizei, GLsizei,
GLsizei, GLint, GLenum, GLenum,
const GLvoid *);
void (GLAPIENTRY *GenVertexArrays)(GLsizei, GLuint *);
void (GLAPIENTRY *BindVertexArray)(GLuint);
GLint (GLAPIENTRY *GetAttribLocation)(GLuint, const GLchar *);
void (GLAPIENTRY *EnableVertexAttribArray)(GLuint);
void (GLAPIENTRY *DisableVertexAttribArray)(GLuint);
void (GLAPIENTRY *VertexAttribPointer)(GLuint, GLint, GLenum, GLboolean,
GLsizei, const GLvoid *);
void (GLAPIENTRY *DeleteVertexArrays)(GLsizei, const GLuint *);
void (GLAPIENTRY *UseProgram)(GLuint);
GLint (GLAPIENTRY *GetUniformLocation)(GLuint, const GLchar *);
void (GLAPIENTRY *CompileShader)(GLuint);
GLuint (GLAPIENTRY *CreateProgram)(void);
GLuint (GLAPIENTRY *CreateShader)(GLenum);
void (GLAPIENTRY *ShaderSource)(GLuint, GLsizei, const GLchar **,
const GLint *);
void (GLAPIENTRY *LinkProgram)(GLuint);
void (GLAPIENTRY *AttachShader)(GLuint, GLuint);
void (GLAPIENTRY *DeleteShader)(GLuint);
void (GLAPIENTRY *DeleteProgram)(GLuint);
void (GLAPIENTRY *GetShaderInfoLog)(GLuint, GLsizei, GLsizei *, GLchar *);
void (GLAPIENTRY *GetShaderiv)(GLuint, GLenum, GLint *);
void (GLAPIENTRY *GetProgramInfoLog)(GLuint, GLsizei, GLsizei *, GLchar *);
void (GLAPIENTRY *GetProgramiv)(GLenum, GLenum, GLint *);
const GLubyte* (GLAPIENTRY *GetStringi)(GLenum, GLuint);
void (GLAPIENTRY *BindAttribLocation)(GLuint, GLuint, const GLchar *);
void (GLAPIENTRY *BindFramebuffer)(GLenum, GLuint);
void (GLAPIENTRY *GenFramebuffers)(GLsizei, GLuint *);
void (GLAPIENTRY *DeleteFramebuffers)(GLsizei, const GLuint *);
GLenum (GLAPIENTRY *CheckFramebufferStatus)(GLenum);
void (GLAPIENTRY *FramebufferTexture2D)(GLenum, GLenum, GLenum, GLuint,
GLint);
void (GLAPIENTRY *BlitFramebuffer)(GLint, GLint, GLint, GLint, GLint, GLint,
GLint, GLint, GLbitfield, GLenum);
void (GLAPIENTRY *Uniform1f)(GLint, GLfloat);
void (GLAPIENTRY *Uniform2f)(GLint, GLfloat, GLfloat);
void (GLAPIENTRY *Uniform3f)(GLint, GLfloat, GLfloat, GLfloat);
void (GLAPIENTRY *Uniform4f)(GLint, GLfloat, GLfloat, GLfloat, GLfloat);
void (GLAPIENTRY *Uniform1i)(GLint, GLint);
void (GLAPIENTRY *UniformMatrix2fv)(GLint, GLsizei, GLboolean,
const GLfloat *);
void (GLAPIENTRY *UniformMatrix3fv)(GLint, GLsizei, GLboolean,
const GLfloat *);
GLsync (GLAPIENTRY *FenceSync)(GLenum, GLbitfield);
GLenum (GLAPIENTRY *ClientWaitSync)(GLsync, GLbitfield, GLuint64);
void (GLAPIENTRY *DeleteSync)(GLsync sync);
void (GLAPIENTRY *VDPAUInitNV)(const GLvoid *, const GLvoid *);
void (GLAPIENTRY *VDPAUFiniNV)(void);
GLvdpauSurfaceNV (GLAPIENTRY *VDPAURegisterOutputSurfaceNV)
(GLvoid *, GLenum, GLsizei, const GLuint *);
void (GLAPIENTRY *VDPAUUnregisterSurfaceNV)(GLvdpauSurfaceNV);
void (GLAPIENTRY *VDPAUSurfaceAccessNV)(GLvdpauSurfaceNV, GLenum);
void (GLAPIENTRY *VDPAUMapSurfacesNV)(GLsizei, const GLvdpauSurfaceNV *);
void (GLAPIENTRY *VDPAUUnmapSurfacesNV)(GLsizei, const GLvdpauSurfaceNV *);
#if HAVE_GL_WIN32
// The HANDLE type might not be present on non-Win32
BOOL (GLAPIENTRY *DXSetResourceShareHandleNV)(void *dxObject,
HANDLE shareHandle);
HANDLE (GLAPIENTRY *DXOpenDeviceNV)(void *dxDevice);
BOOL (GLAPIENTRY *DXCloseDeviceNV)(HANDLE hDevice);
HANDLE (GLAPIENTRY *DXRegisterObjectNV)(HANDLE hDevice, void *dxObject,
GLuint name, GLenum type, GLenum access);
BOOL (GLAPIENTRY *DXUnregisterObjectNV)(HANDLE hDevice, HANDLE hObject);
BOOL (GLAPIENTRY *DXLockObjectsNV)(HANDLE hDevice, GLint count,
HANDLE *hObjects);
BOOL (GLAPIENTRY *DXUnlockObjectsNV)(HANDLE hDevice, GLint count,
HANDLE *hObjects);
#endif
GLint (GLAPIENTRY *GetVideoSync)(GLuint *);
GLint (GLAPIENTRY *WaitVideoSync)(GLint, GLint, unsigned int *);
vo_opengl: implement NNEDI3 prescaler Implement NNEDI3, a neural network based deinterlacer. The shader is reimplemented in GLSL and supports both 8x4 and 8x6 sampling window now. This allows the shader to be licensed under LGPL2.1 so that it can be used in mpv. The current implementation supports uploading the NN weights (up to 51kb with placebo setting) in two different way, via uniform buffer object or hard coding into shader source. UBO requires OpenGL 3.1, which only guarantee 16kb per block. But I find that 64kb seems to be a default setting for recent card/driver (which nnedi3 is targeting), so I think we're fine here (with default nnedi3 setting the size of weights is 9kb). Hard-coding into shader requires OpenGL 3.3, for the "intBitsToFloat()" built-in function. This is necessary to precisely represent these weights in GLSL. I tried several human readable floating point number format (with really high precision as for single precision float), but for some reason they are not working nicely, bad pixels (with NaN value) could be produced with some weights set. We could also add support to upload these weights with texture, just for compatibility reason (etc. upscaling a still image with a low end graphics card). But as I tested, it's rather slow even with 1D texture (we probably had to use 2D texture due to dimension size limitation). Since there is always better choice to do NNEDI3 upscaling for still image (vapoursynth plugin), it's not implemented in this commit. If this turns out to be a popular demand from the user, it should be easy to add it later. For those who wants to optimize the performance a bit further, the bottleneck seems to be: 1. overhead to upload and access these weights, (in particular, the shader code will be regenerated for each frame, it's on CPU though). 2. "dot()" performance in the main loop. 3. "exp()" performance in the main loop, there are various fast implementation with some bit tricks (probably with the help of the intBitsToFloat function). The code is tested with nvidia card and driver (355.11), on Linux. Closes #2230
2015-10-28 02:37:55 +01:00
GLuint (GLAPIENTRY *GetUniformBlockIndex)(GLuint, const GLchar *);
void (GLAPIENTRY *UniformBlockBinding)(GLuint, GLuint, GLuint);
void (GLAPIENTRY *GetTranslatedShaderSourceANGLE)(GLuint, GLsizei,
GLsizei*, GLchar* source);
void (GLAPIENTRY *DebugMessageCallback)(MP_GLDEBUGPROC callback,
const void *userParam);
void *(GLAPIENTRY *MPGetNativeDisplay)(const char *name);
};
#endif /* MPLAYER_GL_COMMON_H */