c++openglglsltexture-mappingopengl-4

Failing to map a simple unsigned byte rgb texture to a quad:


I have a very simple program that maps a dummy red texture to a quad.

Here is the texture definition in C++:

struct DummyRGB8Texture2d
{
    uint8_t data[3*4];
    int width;
    int height;

};

DummyRGB8Texture2d myTexture
{
    { 
        255,0,0,
        255,0,0,
        255,0,0,
        255,0,0
    },
    2u,
    2u
};

This is how I setup the texture:

void SetupTexture()
{
    // allocate a texture on the default texture unit (GL_TEXTURE0): 

    GL_CHECK(glCreateTextures(GL_TEXTURE_2D, 1, &m_texture));

    // allocate texture:
    GL_CHECK(glTextureStorage2D(m_texture, 1, GL_RGB8, myTexture.width, myTexture.height));

    GL_CHECK(glTextureParameteri(m_texture, GL_TEXTURE_WRAP_S, GL_REPEAT));
    GL_CHECK(glTextureParameteri(m_texture, GL_TEXTURE_WRAP_T, GL_REPEAT));
    GL_CHECK(glTextureParameteri(m_texture, GL_TEXTURE_MAG_FILTER, GL_NEAREST));
    GL_CHECK(glTextureParameteri(m_texture, GL_TEXTURE_MIN_FILTER, GL_NEAREST));

    // tell the shader that the sampler2d uniform uses the default texture unit (GL_TEXTURE0) 
    GL_CHECK(glProgramUniform1i(m_program->Id(), /* location in shader */ 3, /* texture unit index */ 0));

    // bind the created texture to the specified target. this is necessary even in dsa
    GL_CHECK(glBindTexture(GL_TEXTURE_2D, m_texture));

    GL_CHECK(glGenerateMipmap(GL_TEXTURE_2D));
}

This is how I draw the texture to the quad:

void Draw()
{
    m_target->ClearTargetBuffers();
    m_program->MakeCurrent();

    // load the texture to the GPU:
    GL_CHECK(glTextureSubImage2D(m_texture, 0, 0, 0, myTexture.width, myTexture.height,
            GL_RGB, GL_UNSIGNED_BYTE, myTexture.data));

    GL_CHECK(glBindVertexArray(m_vao));     

    GL_CHECK(glDrawElements(GL_TRIANGLES, static_cast<GLsizei>(VideoQuadElementArray.size()), GL_UNSIGNED_INT, 0));
        m_target->SwapTargetBuffers();
}

The Result:

enter image description here

I can't figure out why this texture won't appear Red. Also, if I change the texture internal format to RGBA / RGBA8 and the texture data array to have another element in each row, I get a nice red texture.


In case its relevant, here are my vertex attributes and my (very simple) shaders:

struct VideoQuadVertex
{
    glm::vec3 vertex;
    glm::vec2 uv;
};

std::array<VideoQuadVertex, 4> VideoQuadInterleavedArray
{
    /* vec3 */ VideoQuadVertex{ glm::vec3{ -0.25f, -0.25f, 0.5f }, /* vec2 */  glm::vec2{ 0.0f, 0.0f } },
    /* vec3 */ VideoQuadVertex{ glm::vec3{ 0.25f, -0.25f, 0.5f }, /* vec2 */   glm::vec2{ 1.0f, 0.0f } },
    /* vec3 */ VideoQuadVertex{ glm::vec3{ 0.25f, 0.25f, 0.5f }, /* vec2 */    glm::vec2{ 1.0f, 1.0f } },
    /* vec3 */ VideoQuadVertex{ glm::vec3{ -0.25f, 0.25f, 0.5f }, /* vec2 */ glm::vec2{ 0.0f, 1.0f } }


};

vertex setup:

void SetupVertexData()
{


// create a VAO to hold all node rendering states, no need for binding:
    GL_CHECK(glCreateVertexArrays(1, &m_vao));

// create vertex buffer objects for data and indices and initialize them:
GL_CHECK(glCreateBuffers(static_cast<GLsizei>(m_vbo.size()), m_vbo.data()));

// allocate memory for interleaved vertex attributes and transfer them to the GPU:
GL_CHECK(glNamedBufferData(m_vbo[EVbo::Data], VideoQuadInterleavedArray.size() * sizeof(VideoQuadVertex), VideoQuadInterle

GL_CHECK(glVertexArrayAttribBinding(m_vao, 0, 0));
GL_CHECK(glVertexArrayVertexBuffer(m_vao, 0, m_vbo[EVbo::Data], 0, sizeof(VideoQuadVertex)));


// setup the indices array:
GL_CHECK(glNamedBufferData(m_vbo[EVbo::Element], VideoQuadElementArray.size() * sizeof(GLuint), VideoQuadElementArray.data
GL_CHECK(glVertexArrayElementBuffer(m_vao, m_vbo[EVbo::Element]));

// enable the relevant attributes for this VAO and
// specify their format and binding point:

// vertices:
GL_CHECK(glEnableVertexArrayAttrib(m_vao, 0 /* location in shader*/));
GL_CHECK(glVertexArrayAttribFormat(
    m_vao,
    0,                  // attribute location
    3,                  // number of components in each data member
    GL_FLOAT,           // type of each component
    GL_FALSE,           // should normalize
    offsetof(VideoQuadVertex, vertex)   // offset from the begining of the buffer
));

// uvs:
GL_CHECK(glEnableVertexArrayAttrib(m_vao, 1 /* location in shader*/));
GL_CHECK(glVertexAttribFormat(
    1,                                  // attribute location
    2,                                  // number of components in each data member
    GL_FLOAT,                           // type of each component
    GL_FALSE,                           // should normalize
    offsetof(VideoQuadVertex, uv)   // offset from the begining of the buffer
));

GL_CHECK(glVertexArrayAttribBinding(m_vao, 1, 0));

}

vertex shader:

layout(location = 0) in vec3 position;
layout(location = 1) in vec2 texture_coordinate;


out FragmentData
{
    vec2 uv;
} toFragment;


void main(void)
{   
    toFragment.uv = texture_coordinate;
    gl_Position = vec4 (position, 1.0f);
}

fragment shader:

in FragmentData
{
    vec2 uv;
} data;

out vec4 color;

layout (location = 3) uniform sampler2D tex_object;

void main()
{
    color = texture(tex_object, data.uv);   
}

Solution

  • GL_UNPACK_ALIGNMENT specifies the alignment requirements for the start of each pixel row in memory. By default GL_UNPACK_ALIGNMENT is set to 4. This means each row of the texture is supposed to have a length of 4*N bytes.

    You specify a 2*2 texture with the data: 255, 0, 0, 255, 0, 0, 255, 0, 0, 255, 0, 0

    With GL_UNPACK_ALIGNMENT set to 4 this is interpreted as

             column 1          column 2              alignment
    row 1:   255, 0,   0,      255, 0,     0,        255, 0,
    row 2:   0,   255, 0,      0,   undef, undef
    

    So the texture is read as

              column 1   olumn 2  
    row 1:    red,       red,
    row 2:    green,     RGB(0, ?, ?)
    

    You have to set glPixelStorei(GL_UNPACK_ALIGNMENT, 1); before glTextureSubImage2D, for reading a tight packed texture.

    If you do not want to change GL_UNPACK_ALIGNMENT (the alignment remains set to 4), you must adjust the data as follows:

    struct DummyRGB8Texture2d
    {
        uint8_t data[8*2];
        int width;
        int height;
    };
    
    DummyRGB8Texture2d myTexture
    {
        { 
            255, 0, 0, 255, 0, 0, // row 1
            0, 0,                 // 2 bytes alignment
            255, 0, 0, 255, 0, 0, // row 2
            0, 0                  // 2 bytes alignment
        },
        2u,
        2u
    };
    

    See further: