Generating a Float Texture and usage in Shader

Question

I have problems generating a Float texture with depth data in it.

Since I think its more an OpenGL ES (OES 2.0) problem instead of the the SDK i am using, I try to get help here considering the OpenGL ES part.

The error i am getting is just that my app crashes, this is why I think, that the error is a wrong Texture and/or wrong usage of the float Texture.

Just a few annotations about the data i am using:

-The depthFrame I am getting is in Float
-With the use of depthInMillimeters, it is possible to get actual mm values in Float 
-Since i am working on an iPhone 5 GL_HALF_FLOAT_OES should be available to use
-If someone wants to know, the data comes from the StructureSensor

This is my function to generate the Texture from the streamed depth data. I am using nearly the same function to generate RGB Textures from images, but with other data input and values.

    -(GLuint) generateDepthTexture: (STDepthFrame*)depthFrame{

    //data from depth in mm
    NSData *data = [NSData dataWithBytes:depthFrame.depthInMillimeters length:depthFrame.width*depthFrame.height*4];

    //data to texture
    UIImage* depthImage = [[UIImage alloc] initWithData:data];

    CGImageRef spriteImage = [depthImage CGImage];
    if(!spriteImage){
        NSLog(@"Failed to load image depth");
        exit(1);
    }

    size_t width = CGImageGetWidth(spriteImage);
    size_t height = CGImageGetHeight(spriteImage);

    GLubyte * spriteData = (GLubyte *) calloc(width*height*4, sizeof(GLubyte)); // *4


    CGContextRef spriteContext = CGBitmapContextCreate(spriteData, width, height, 8, width*4,
                                                   CGImageGetColorSpace(spriteImage), kCGImageAlphaPremultipliedLast);

    CGContextDrawImage(spriteContext, CGRectMake(0, 0, width, height), spriteImage);
    CGContextRelease(spriteContext);


    GLuint texName;

    glGenTextures(1, &texName);
    glBindTexture(GL_TEXTURE_2D, texName);

    glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_NEAREST);

    glTexImage2D(GL_TEXTURE_2D, 0, GL_FLOAT, (GLsizei)width,(GLsizei)height, 0, GL_FLOAT, GL_HALF_FLOAT_OES, spriteData);

    free(spriteData);

    return texName;


    }

This is the important part of my render method:

    //depth
    //depthFrame is given every call
    _depthTexture = [self generateDepthTexture:depthFrame];

    glActiveTexture(GL_TEXTURE2);
    glBindTexture(GL_TEXTURE_2D, _depthTexture);
    glUniform1i(_depthUniform, 2);

I am using it in the shader with this:

    //texture 
    uniform sampler2D DepthTexture;

    //usage of the Texture
    mediump vec4 depthV = texture2D(DepthTexture,TexCoordsIn);

I am not sure about the right usage here, because i would rather get the values in a float, than in a vec4.

And again, i would like to get some help considering the OpenGL ES part with the float texture. If its completely right, the problem must be with the data I am getting. If thats the case, I will ask the people working with the sensor.

The error might be only a wrong texture size or memory.

I really hope, someone has an answer to this.

Reto Koradi · Accepted Answer · 2016-07-03 00:30:40Z

The OP seems to be aware of this already, but just to send it ahead for clarity: float textures are not a standard ES 2.0 feature. They are only supported if the OES_texture_float and/or OES_texture_half_float extensions are present. This is fairly obvious for the half-float case, since the extension introduces the GL_HALF_FLOAT_OES value, but extension support is also required for the float case, even though the extension does not introduce new enum values.

Your attempt at using the extension has a number of issues:

glTexImage2D(GL_TEXTURE_2D, 0, GL_FLOAT, (GLsizei)width, (GLsizei)height, 0,
             GL_FLOAT, GL_HALF_FLOAT_OES, spriteData);

You're trying to pass GL_FLOAT/GL_HALF_FLOAT values to the internalformat, format and type arguments of glTexImage2D(). But the extension only introduces new values for the type argument. So the correct call to allocate a 1-component float texture is:

glTexImage2D(GL_TEXTURE_2D, 0, GL_LUMINANCE, (GLsizei)width, (GLsizei)height, 0,
             GL_LUMINANCE, GL_FLOAT, spriteData);

Note that ES 2.0, unlike newer version of OpenGL, still uses GL_LUMINANCE or GL_ALPHA for the internalformat and format of 1-component textures.

For a half-float texture, the call is:

glTexImage2D(GL_TEXTURE_2D, 0, GL_LUMINANCE, (GLsizei)width, (GLsizei)height, 0,
             GL_LUMINANCE, GL_HALF_FLOAT_OES, spriteData);

For this half-float case, it's important to understand that you need to pass half-float (16-bit per value) data into the call. If your original data is in float format, you'll have to write (or find) code that converts it into half-float format.

This all looks different in ES 3.0 and later, where the sized internal formats from more modern OpenGL versions were introduced. There, the call to allocate a 1-component float texture would be:

glTexImage2D(GL_TEXTURE_2D, 0, GL_R32F, (GLsizei)width, (GLsizei)height, 0,
             GL_RED, GL_FLOAT, spriteData);

Collectives™ on Stack Overflow

Generating a Float Texture and usage in Shader

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related