OpenGL (ES) Image Processing C++ - c++

Question

I am currently trying to do image processing using OpenGL ES. I am trying to do basic image effects like blurring switching color space and so on.
I want to build the simplest program to do the following things:
Image loading
Image processing (using shader)
Image saving
I have managed to build the following setup :
An OpenGL context
The image I want to do effects on loaded using DevIL.
Two shaders (one vertex shaders and one fragment shaders)
I am now stuck at using the image I loaded to send data to fragment shader. What I am trying to do is to send the image as a sampler2D to the fragment shader and apply treatment on it.
I have multiple questions such as:
Do I need a vertex shader if all I want to do is pure 2D image processing ?
If I do, what should be done in this vertex shader as I have no vertices at all. Should I create quad vertices (like (0,0) (1, 0) (0, 1) (1, 1)) ? If so, why ?
Do I need to use things like VBO (which seems to be related to the vertex shader), FBO or other thing like that ?
Can't I just load my image into the texture and wait for the fragment shader to do everything I want on this texture ?
Can someone provides some simple piece of "clean" code that could help me understand (without any fancy classes that makes the understanding so complicated) ?
Here is what my fragment shader looks like for simple color swapping:
uniform int width;
uniform int height;
uniform sampler2D texture;
void main() {
vec2 texcoord = vec2(gl_FragCoord.x/width, gl_FragCoord.y/height);
vec4 texture_value = texture2D(texture, texcoord);
gl_FragColor = texture_value.bgra;
}
and my main.cpp :
int main(int argc, char** argv)
{
if (argc != 4) {
std::cerr << "Usage: " << argv[0] << " <vertex shader path> <fragment shader path> <image path>" << std::endl;
return EXIT_FAILURE;
}
// Get an EGL valid display
EGLDisplay display;
display = eglGetDisplay(EGL_DEFAULT_DISPLAY);
if (display == EGL_NO_DISPLAY) {
std::cerr << "Failed to get EGL Display" << std::endl
<< "Error: " << eglGetError() << std::endl;
return EXIT_FAILURE;
}
else {
std::cerr << "Successfully get EGL Display." << std::endl;
}
// Create a connection to the display
int minor, major;
if (eglInitialize(display, &minor, &major) == EGL_FALSE) {
std::cerr << "Failed to initialize EGL Display" << std::endl
<< "Error: " << eglGetError() << std::endl;
eglTerminate(display);
return EXIT_FAILURE;
}
else {
std::cerr << "Successfully intialized display (OpenGL ES version " << minor << "." << major << ")." << std::endl;
}
// OpenGL ES Config are used to specify things like multi sampling, channel size, stencil buffer usage, & more
// See the doc: https://www.khronos.org/registry/EGL/sdk/docs/man/html/eglChooseConfig.xhtml for more informations
EGLConfig config;
EGLint num_configs;
if (!eglChooseConfig(display, configAttribs, &config, 1, &num_configs)) {
std::cerr << "Failed to choose EGL Config" << std::endl
<< "Error: " << eglGetError() << std::endl;
eglTerminate(display);
return EXIT_FAILURE;
}
else {
std::cerr << "Successfully choose OpenGL ES Config ("<< num_configs << ")." << std::endl;
}
// Creating an OpenGL Render Surface with surface attributes defined above.
EGLSurface surface = eglCreatePbufferSurface(display, config, pbufferAttribs);
if (surface == EGL_NO_SURFACE) {
std::cerr << "Failed to create EGL Surface." << std::endl
<< "Error: " << eglGetError() << std::endl;
}
else {
std::cerr << "Successfully created OpenGL ES Surface." << std::endl;
}
eglBindAPI(EGL_OPENGL_API);
EGLContext context = eglCreateContext(display, config, EGL_NO_CONTEXT, contextAttribs);
if (context == EGL_NO_CONTEXT) {
std::cerr << "Failed to create EGL Context." << std::endl
<< "Error: " << eglGetError() << std::endl;
}
else {
std::cerr << "Successfully created OpenGL ES Context." << std::endl;
}
//Bind context to surface
eglMakeCurrent(display, surface, surface, context);
// Create viewport and check if it has been created correctly
glViewport(0, 0, WIDTH, HEIGHT);
GLint viewport[4];
glGetIntegerv(GL_VIEWPORT, viewport);
if (viewport[2] != WIDTH || viewport[3] != HEIGHT) {
std::cerr << "Failed to create the viewport. Size does not match (glViewport/glGetIntegerv not working)." << std::endl
<< "OpenGL ES might be faulty!" << std::endl
<< "If you are on Raspberry Pi, you should not updated EGL as it will install fake EGL." << std::endl;
eglTerminate(display);
return EXIT_FAILURE;
}
// Clear buffer and get ready to draw some things
glClearColor(0.0f, 0.0f, 0.0f, 1.0f);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
// Create a shader program
GLuint program = load_shaders(std::string(argv[1]), std::string(argv[2]));
if (program == -1)
{
std::cerr << "Failed to create a shader program. See above for more details." << std::endl;
eglTerminate(display);
return EXIT_FAILURE;
}
/* Initialization of DevIL */
if (ilGetInteger(IL_VERSION_NUM) < IL_VERSION) {
std::cerr << "Failed to use DevIL: Wrong version." << std::endl;
return EXIT_FAILURE;
}
ilInit();
ILuint image = load_image(argv[3]);
GLuint texId;
glGenTextures(1, &texId); /* Texture name generation */
glBindTexture(GL_TEXTURE_2D, texId); /* Binding of texture name */
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR); /* We will use linear interpolation for magnification filter */
glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR); /* We will use linear interpolation for minifying filter */
glTexImage2D(GL_TEXTURE_2D, 0, ilGetInteger(IL_IMAGE_BPP), ilGetInteger(IL_IMAGE_WIDTH), ilGetInteger(IL_IMAGE_HEIGHT),
0, ilGetInteger(IL_IMAGE_FORMAT), GL_UNSIGNED_BYTE, ilGetData()); /* Texture specification */
Thanks.

Answer 1

Do I need a vertex shader as all I want to do is pure 2D image processing ?
Using vertex and fragment shaders is mandatory in OpenGL ES 2.
If I do, what should be done in this vertex shader as I have no vertices at all. Should I create quad vertices (like (0,0) (1, 0) (0, 1) (1, 1)) ? If so, why ?
Yes. Because that's how OpenGL ES 2 works. Otherwise you would need to use something like computer shaders (supported in OpenGL ES 3.1+) or OpenCL.
Do I need to use things like VBO (which seems to be related to the vertex shader), FBO or other thing like that ?
Using VBO/IBO won't make practically any difference for you since you only have 4 vertices and 2 primitives. You may want to render to texture, depending on your needs.
Can't I just load my image into the texture and wait for the fragment shader to do everything I want on this texture ?
No.