This is a library providing OpenGL 1.x functionality for OpenGL ES accelerated hardware.
This is a fork a glshim (https://github.com/lunixbochs/glshim). Go check this lib if you need things like RemoteGL or TinyGLES (for 100% software rendering).
The focus is on compatibility with a wide selection of game and software, as well as speed.
It has been tested successfully of a large selection of games and software, including: Mincraft, OpenMW, SeriousSam, RVGL, TSMC, TORCS, SpeedDreams, GL-117, Foobillard(plus), Blender 2.68 and many more.
Most function of OpenGL up to 1.5 are supported, with some notable exceptions:
- Reading of Depth or Stencil buffer will not work
- GL_FEEDBACK mode is not implemented
Some know limitations:
- GL_SELECT as some limitation in its implementation (for exemple, current Depth buffer or binded texture are not taken into account)
- NPOT texture are supported, but not with GL_REPEAT / GL_MIRRORED, only GL_CLAMP will work properly (unless the GLES Hardware support NPOT)
- Framebuffer use FRAMEBUFFER_OES extension (that must be present in the GLES 1.1 stack)
- Multiple Color attachment on Framebuffer are not supported
- OcclusionQuery is implemented, but with a 0 bits precision
- Probably many other things
for Pandora
cmake . -DPANDORA=1; make GL
or for the Raspberry Pi
cmake . -DBCMHOST=1; make GL
or for the ODroid
cmake . -DODROID=1; make GL
or for CHIP machines
cmake . -DCHIP=1; make GL
or for Android
An Android.mk is provided that should compile with an NDK
or use ccmake
Alternatively, you can use the curses-bases ccmake (or any other gui frontend for cmake) to select wich platform to use interactively.
Standard GLU do works without any issues.
Put lib/libGL.so.1 in your LD_LIBRARY_PATH
.
There are many environnement variable to control gl4es behavour. All are numeric, except LIBGL_VERSION that take a string.
Controls the Framebuffer output
- 0 : Default, using standard x11 rendering
- 1 : Use Framebuffer output (x11 bypassed, only fullscreen)
- 2 : Use Framebuffer, but also an intermediary FBO
- 3 : Use PBuffer, allowing x11 (and windowed) rendering even if driver doesn't support it
Debug helper in specific cases
- 0 : Default, nothing special
- 1 : xrefresh will be called on cleanup
Automatic Backtrace log
- 0 : Default, nothing special
- 1 : stacktrace will be printed on crash
VSync control
- 0 : Default, nothing special
- 1 : vsync enabled
Recycling FBO special case (don't delete a created FBO, but recycle it if needed)
- 0 : Default, nothing special
- 1 : Recycling of FBO enabled
Handling of Manual and Automatic MIPMAP
- 0 : Default, nothing special
- 1 : AutoMipMap forced
- 2 : guess AutoMipMap (based on manual mipmaping on textures)
- 3 : ignore MipMap (mipmap creation / use entirely disabled)
- 4 : ignore AutoMipMap on non-squared textures
Make a local copy of every texture for easy glGetTexImage2D
- 0 : Default, nothing special
- 1 : Texture copy enabled
Texture shrinking control
- 0 : Default, nothing special
- 1 : everything / 2 (using original algorithm for size reduction, all other shink mode use a refined algorithm)
- 2 : only textures wich one size > 512 are / 2
- 3 : only textures wich one size > 256 are / 2
- 4 : only textures wich one size > 256 are / 2, and the one > 1024 are / 4
- 5 : only textures wich one size > 256 are resized to 256 (if possible, because only /2 and /4 exists), but empty texture are not shrinked
- 6 : only textures wich one size > 128 are / 2, thoses >= 512 are resized to 256 (if possible, because only /2 and /4 exists), but empty texture are not shrinked
- 7 : only textures wich one size > 512 are / 2, but empty texture are not shrinked
- 8 : advertise a max texture size of 8192, but every texture wich one size > 2048 are shrinked to 2048
- 9 : advertise a max texture size of 8192, but every texture wich one size > 4096 are / 4 and the one > 512 are / 2, but empty texture are not shrinked
- 10: advertise a max texture size of 8192, but every texture wich one size > 2048 are / 4 and the one > 512 are / 2, but empty texture are not shrinked
Texture dump
- 0 : Default, nothing special
- 1 : Texture dump enabled
Experimental: enable Alpha test only when using texture that contains an alpha channel
- 0 : Default, nothing special
- 1 : Alpha Hack enabled
Texture downsampling control
- 0 : Default, DXTc texture are downsampled to 16bits
- 1 : DXTc texture are left as 32bits RGBA
PANDORA only: enable Texture Streaming (works only on RGB textures)
- 0 : Default, nothing special
- 1 : Enabled on empty RGB textures
- 2 : Enabled on all RGB textures
Control the glCopyTex(Sub)Image2D hack (they are buggy on pandora and don't work most of the time)
- 0 : Don't use native glCopyTex(Sub)Image2D, but a workaround function using FBO
- 1 : No glCopyTexImage2D / glCopyTexSubImage2D hack, use native ones
Control the availability of the LUMUNANCE_ALPHA format (can be buggy on Pandora model CC)
- 0 : Default,GL_LUMINANCE_ALPHA is available and used if needed
- 1 : GL_LUMINANCE_ALPHA hardware support disabled (a GL_RGBA texture will be used instead)
Experimental: Change Blend GL_SRC_ALPHA, GL_ONE to GL_ONE, GL_ONE
- 0 : Default, nothing special
- 1 : Change Blend GL_SRC_ALPHA, GL_ONE to GL_ONE, GL_ONE (can be usefull for Xash3D engine)
Hack: Export a (faked) glBlendColor
- 0 : Default, don't expose gBlendColor
- 1 : Exposed the function (if no hadware support, faked function willbe used)
Hack: Control the glGetString version. Overide version string (should be in the form of "1.x")
Experimental: Batch mode (fuse of contigous Display list, to limit Draw calls)
- 0 : Default, only Draw list created during a glList are fused (i.e. contigous glBegin / glEnd inside a List)
- 1 : Force a maximum of call to be batched (like if all was inside a big glList)
- 2 : Disable Batch mode completly, no fuse of draw list
Hack: glGetError() always return GL_NOERROR
- 0 : Default, glGetError behave as it should
- 1 : glGetError never fail.
Pandora Hack: Set a Gamma value (in decimal formal, 1.0 means no gamma boost)
- X.Y : Use X.Y as gamma when creating context (typical value can be 1.6 or 2.0)
ODROID Hack: Enable sRGB Surface (so Gamma corrected), if Hardware support it
- 0 : Default, don't try to use sRGB surface
- 1 : Enable sRGB Surface (but support will be tested first, must have EGL_KHR_gl_colorspace extension)
Hack: Activate some Fast Math in processor/coprocessor
- 0 : Default, nothing special
- 1 : On OpenPandora, activate "RunFast" on Cortex-A8 (mode default NaN, flush-to-zero) : Not implemented on other platforms (will do nothing)
Debug: Hide or Show the Sub / Not found message
- 0 : Default, the messages are shown
- 1 : Silent, don't print the STUB or glXGetProcAddress glXXXXX not found message
Show/Hide initial text
- 0 : Default, print starting message
- 1 : Silent: no LIBGL message at start (combile with LIBGL_SILENTSTUB for more silence)
Expose NPOT (Non Power of Two) Support
- 0 : Default, don't expose the extension (but npot are still supported)
- 1 : Expose limited NPOT extension
- 2 : Expose GL_ARB_texture_non_power_of_two extension
Expose glQueries functions
- 0 : Don't expose the function (fake one will be used if called)
- 1 : Default, expose fake functions (always answer 0)
Handling of Texture Matrix
- 0 : Default, perform handling internaly (better handling of NPOT texture on all hardware)
- 1 : Let the driver handle texmat (can be faster in some cases, but NPOT texture may be broken)
Initial Hardware test
- 0 : Default, perform intial hardware testing (using a PBuffer)
- 1 : Do not perform test (no extensions tested or used)
VAO Caching
- 0 : Default, try to cache vao to avoid memcpy in render list
- 1 : Don't cache VAO
Vertex Array BGRA extension
- 0 : Default, GL_ARB_vertex_array_bgra not exposed (still emulated)
- 1 : Extension exposed may be faster in some cases (Arx Libertatis mainly)
Merge of subsequent glBegin/glEnd blocks (will be non-effective if BATCH mode is used)
- 0 : Don't try to merge
- 1 : Try to merge, even if there is a glColor / glNormal in between (default)
- 2 : Try hard to merge, even if there is a glColor / glNormal or Matrix operations in between
- Added some optimisations for sequencial glBegin/glEnd blocks, with a switch to control them
- Fixed many issue with Lights introduced with 0.9.4
- Fixed Android build introduced with 0.9.4
- Fixed some extended functions (like glBlendFuncSeparate) not working inside list (fixing some issues with Batch mode)
- Added back GL_TEXTURE_RECTANGLE_ARB handling (and using npot texture, even limited, if available)
- Added tracking of Lights and Materials
- Fixed (Added in fact) support for Multisampling on the GLX Context creation (a bit hacky, but seems to works)
- Added LIBGL_NODOWNSAMPLING and associated Hint
- Try to implement some caching on VAO to avoid some memcpy in renderlist (with a way to disable it just in case)
- Added support for Cube Mapping (with hardware support)
- Improved Texture state tracking
- Added LIBGL_NOTEXMAT env. var. switch for Texture Matrix handling
- Added GL_EXT_vertex_array_bgra (and NEONinzed some loop)
- Finished GL_EXT_direct_state_access extension
- Mangled glX function (to be abble to use apitrace to capture GL frames)
- Return some values in glXQueryServerString, coherent with glXGetClientString
- All matrix are tracked now
- Texture Matrix are 100% handled by gl4es. GLES Hardware keep an Identity matrix (TexCoord are transformed if needed). This allows a better handling of NPOT texture on hadware that doesn't support Full NPOT (fixed movies beiing horizontaly shifted in openmw with LIBGL_NPOT=1 for example)
- Added gl4es specifics glHint capabilities. If the extension GL_GL4ES_hint is present, than a few Hint are accessible. Look in include/gl4eshint.h for the list.
- New name: gl4es