Mali T760 INCOMPLETE_ATTACHMENT when attach a texture with floating point format to Framebuffer

Does the driver in the Galaxy s6 support int or unsigned int render target?

or it supports only unsigned byte.

Thanks

Yes - both signed and unsigned int targets are supported (they are a standard part of the OpenGL ES 3.0 spec). Just watch out for the bandwidth requirements - they require a lot of data, so will be slower and more power hungry than the narrower color formats.

HTH,
Pete

after I changes format type to integer 32 bit, I failed to read integer data from frame buffer returned data all zeroes.

while I type data in shader the shader (hard coded) - commented line - the data returned by the values.

I checked if there is any error but no errors found.

Can anyone help me to solve this problem?

/ *********** frame buffer ********** /

GLuint texture;

glGenTextures(1, &texture);

glBindTexture(GL_TEXTURE_2D, texture);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_NEAREST);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);

glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA32I, width, height, 0, GL_RGBA_INTEGER, GL_INT, NULL);

GLuint depthbf;

glGenRenderbuffers(1, &depthbf);

glBindRenderbuffer(GL_RENDERBUFFER, depthbf);

glRenderbufferStorage(GL_RENDERBUFFER, GL_DEPTH_COMPONENT16, width, height);

GLuint framebf;

glGenFramebuffers(1, &framebf);

glBindFramebuffer(GL_FRAMEBUFFER, framebf);

glFramebufferTexture2D(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_TEXTURE_2D, texture, 0);

glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_DEPTH_ATTACHMENT, GL_RENDERBUFFER, depthbf);

/********* texture ********/

int32_t * data = (int32_t*) malloc(size * sizeof(int32_t));

// for loop to fill data

  for (int i = 0; i < size ; i = 4+i) {

            data[i] =  1;

            data[i+1] =  0;

            data[i+2] =  0;

            data[i+3] = 0;

}

glGenTextures(1,&_Mat);

glActiveTexture(GL_TEXTURE0);

glBindTexture(GL_TEXTURE_2D,_Mat);

glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA32I, width, height, 0, GL_RGBA_INTEGER, GL_INT, data);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_CLAMP_TO_EDGE);

glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_CLAMP_TO_EDGE);

glBindTexture(GL_TEXTURE_2D, 0);

// after run the kernel I read the output buffer by this line

glReadPixels(0, 0, width, height,GL_RGBA_INTEGER, GL_INT, outData);

//  *****  fragment Shader *******//

uniform sampler2D myMat;

void main()

    {      

        // just for testing I read the input data

    gl_FragColor = texture2D(myMat,vec2(0.5,0.5));

    // gl_FragColor = vec4(1,0,0,0);

    };

Thanks in advance,

Your sampler type is wrong in the fragment shader if you are reading from an integer texture. Signed and unsigned integer formats have a specialist sampler type (isampler2D for signed integer and usampler2D for unsigned integer), so it's likely just returning black.

Also, you are setting up bilinear filtering (GL_LINEAR) for your texturing filter, which isn't allowed for integer formats. They are not filterable in the standard, and only support GL_NEAREST.

HTH,

Pete

Actually I changed sampler2D to isampler2D after I asked the question.

I removed texturing filter but the output still the same all channels  = zeros except alpha channel =1

Any suggestions?