a simple framework for OpenGLES 2.0 written in C
*** Brief API notes at the end of the readme ***
intended to help learning and tutorial writing
Currently supports XORG (linux) Desktop using mesa GLES2.0 libs or various embedded platforms such as Allwinner A20
dependencies, libEGL, libGLES (2.0), libode (0.11) for phystest, Chipmunk-6.1.1 for chiptest
pkg-config, build-esentials and development libraries must be installed to compile the framework
raw mouse, keyboard and joystick on some embedded platforms
When not using xwindows (ie via ssh) input including keyboard is now done entirely via the kernel evdev interface.
you (might) need some udev rules
make a file called /etc/udev/rules.d/99-evdev.rules (as root) it should contain the following
KERNEL=="event*", NAME="input/%k", MODE="0640", GROUP="evdev" KERNEL=="mouse*", NAME="input/%k", MODE="0640", GROUP="evdev" KERNEL=="js*", NAME="input/%k", MODE="0640", GROUP="evdev"
you need to add an new group and add your user account to the group (as root)
groupadd evdev usermod -a -G evdev your_user_name
you'll need to log out and back in again
You can now run your programs from ssh and it will only use the targets attached usb keyboard rather than being confused with the ssh console
editing files via ssh (sftp enabled editor) and compiling with a ssh console is the recommended way of developing with this framework if running with a small LCD and you have old tired eyes...
file structure for external libraries
some examples rely on external libraries they should be extracted and compiled in the same directory that you are working on the frame work like this:
Chipmunk-6.1.1 gles2framework ode-0.13
phystest (ode example)
the rather hacky ODE example is only really for advanced users...
(install Debian package version v0.11sp-dev (single precision package)
If you want some of the extra bits of libODE (trimesh vs cylinder collisions - and other luxuries) then you can always compile ODE from source, see the ODE website for details
suggested compile configuration for ODE
./configure --enable-libccd --with-box-cylinder=libccd --with-drawstuff=none --disable-demos
this should by default provide trimesh collider etc as a static library
chiptest (Chipmonk physics example)
a quick example showing some balls falling on some invisible slopes, niether the sprites or the position of the slopes are scaled depending on the display size, so the sample will look different on different platforms and is a good example of why you should use scaled sizes!
You can pass parameters to cmake or edit CMakeLists.txt so as to NOT compile the demos, you only need compile a static library
option(BUILD_DEMOS "Build the demo applications" OFF) option(INSTALL_DEMOS "Install the demo applications" OFF) option(BUILD_SHARED "Build and install the shared library" OFF) option(BUILD_STATIC "Build as static library" ON) option(INSTALL_STATIC "Install the static library" OFF)
|-include||source code include files|
|-kazmath||source code and docs for the kazmath library|
|-lib||kazmath compiled as static lib goes here. Other libs may follow. The framework may become a library|
|-o||somewhere to put intermediate binary objects|
|-src||source code for the framework|
|-tools||a tool to package up 3d shapes and one to make 2d bitmap fonts|
|-resources||holds textures, shaders and binary 3d models for the samples|
|-examples||example code showing use of the framework|
|Makefile||tells the compiler how to build the examples|
|TODO.md||aide memoire, ideas and inspiration for future development|
there is no need to seperatly compile the kazmath library for your platform kazmath sources are now automatically compiled into a static library
All though the source is unchanged I have deleted everthing except the C source the full distribution of kazmath is available at https://github.com/Kazade/kazmath
In order to create code from OBJ files use this script, ensure you set the output file name to something like shape.c
you will have to manually make shape.h it should look something like this
extern float shapeVerts; extern float shapeTexCoords; extern float shapeNormals; extern unsigned int shapeNumVerts;
in addition you will have to add a rule in the Makefile
this method is depricated you should use a GBO instead...
This is a binary file format, a kind of compiled OBJ file which can be used instead of embedding your objects in the executable (which can be wasteful in terms of ram) Once the object data is passed to the GPU the loaded data its based on is freed from memory...
To make a gbo (Gles Binary Object) file place your wavefront object into the same directory, if for example the shape is called alien.obj then execute ./makeGBO.sh alien - note the lack of the file extension it will output alien.gbo which you can then copy to your resources directory - see loadObj command detailed below. makeGBO.sh relies on obj2opengl which must be in the same directory, it also needs at least the build esentials if going on another (artists) machine.
int loadPNG(const char *filename);
loads a specified png file returning a GLES texture handle
int makeContext(); int makeContextXY(int x, int y);
creates a native window and sets up a GLES context within it makeContextXY takes dimesions for the window, giving -1,-1 will put the context in fullscreen mode, you can the use getDisplayWidth() / getDisplayHeight() to resize things according to this resolution if needed.
closes GLES context and window
GLuint create_shader(const char *filename, GLenum type);
returns a GLES shader handle from a file you must specify what type of shader it is either GL_VERTEX_SHADER or GL_FRAGMENT_SHADER
GLuint getShaderLocation(int type, GLuint prog, const char *name);
given a type of shaderAttrib or shaderUniform, a shader program handle and a name for the attrib or uniform it returns a location handle
void initGlPrint(int w, int h);
This initialises the resources used by the glPrintf you must supply the windows width and height
font_t* createFont(const char* tpath,uint cbase,float tHeight,float tLines, int fWidth, int fHeight);
tpath is the full path and of the texture for this font, cbase is the ascii code of the first character tHeight is the height in pixels of the texture, tLines specifies the number of lines in total the texture contains (I think there is a bug with this!) fWidth and fHeight are the width and height a character
TODO freeFont to release a fonts resources...
void glPrintf(float x, float y, font_t fnt, const char *fmt, ...);
this behaves exactly like a normal printf except for the first two parameters which specify the starting coordinate you must specify a previously created font structure to print with
In order isolate egl and native window handles use this routine instead of eglSwapBuffers
The number of frames EGL should wait before swapBuffers actually swaps the buffer, this freqently does nothing on many EGL implementations however...
this should be called once a frame to update the key down boolean array and the mouse information
this returns a pointer to an array of 3 ints the first 2 are the x and y mouse position the 3rd int is a bit field reflecting the state of the mouse buttons
this is an array of 256 bools, while a key is held down the coresponding bool is true, key values are defined in keys.h
*** deprecated may be removed in later version ***
int createObj(struct obj_t *obj, int numVerts, float verts, float txVert, float norms, char *vertShader, char *fragShader);
pass an empty obj_t struct, the number of vertices and arrays of verts, texture coordinates and normals, finally you need to specify the file names for the vert and frag shaders
*** deprecated may be removed in later version ***
int createObjCopyShader(struct obj_t *obj, int numVerts, float verts, float txVert, float norms, struct obj_t *sdrobj);
this allows you to initialise a obj shape but using an existing obj's shader.
void drawObj(struct obj_t *obj, kmMat4 * combined, kmMat4 * mv, kmVec3 lightDir, kmVec3 viewDir);
this draws an obj, you need to supply a combined model, view and projection matrices as well as a combined model, view matrix for the lighting, light and view direction vectors are also needed for lighting
returns full screen width and height, for now when not on Raspberry PI the "screen" is fixed to a 640x480 window
int loadObj(struct obj_t *obj,const char *objFile, char *vert, char *frag);
int loadObjCopyShader(struct obj_t *obj,const char *objFile, struct obj_t *sdrobj);
these are basically the same as their createObj counterparts except the OBJ is loaded from a compiled binary wavefront object instead of from embedded data.
void initSprite(int w, int h);
void drawSprite(float x, float y, float w, float h, float a, int tex);
like glPrinf the sprite subsystem must be initialised before use, pass the dimensions of the screen.
when drawing a sprite you specify where you want it (x & y) the size of the sprite (w & h) the rotation (a) and which texture to use (tex)
void setMouseRelative(bool mode);
if mode is true the mouse will report relative position changes only, this is handy for mouse look where you dont want the mouse constrained by the window. By default absolute mouse position is reported
struct joystick_t *getJoystick(int j);
void updateJoystick(struct joystick_t *js);
void releaseJoystick(struct joystick_t *js);
to get a pointer to a joystick call getJoystick with the index of the joystick 0-7 call this once only
once a frame call updateJoystick you will then have (in the joystick structure)
js->axis[0..7] upto 8 axes per joystick (signed short) js->buttons long - each bit represents a button
when finished with a joystick you should call releaseJoystick to close its file handle and free the structures memory.
void initPointClouds(const char* vertS, const char* fragS, float pntSize);
struct pointCloud_t* createPointCloud(int size);
void drawPointCloud(struct pointCloud_t* pntC,kmMat4* m);
void freePointCloud(struct pointCloud_t* pntC);
initPointClouds is used initialise the common shader used by the point clouds and set the size of the individual points (this can be changed on the fly later by changing a shader uniform)
createPointCloud reserves space for the position and velocity components of each individual point in a cloud
when drawing a point cloud you must pass the combined model/view/projection matrix in a similar manner to drawing obj shapes, note you must update the individual point positions optionally using the supplied velocity value for each point
While it is ok to keep a point cloud around without drawing it for later use. When the resources used by the cloud need to be released call freePoint cloud this frees the point cloud structure itself and the associated points data
void reProjectGlPrint(int w, int h) void reProjectSprites(int w, int h)
When glfw calls your window resize callback you should run these functions if you are using the apropriate facilities to inform them of the screen changes
void resizePointCloudSprites(float s)
This allows you to change the size of the point sprite's used - you would usually do this in the screen resize callback.