I'm trying to render a skybox with a cubemap on it and to do so I'm using DDS Texture Loader from DirectXTex library. I use texassemble to generate the cubemap (texture array of 6 textures) into a DDS file that I load at runtime. I generated a cube "dome" and sample the texture using the position vector of the vertex as the sample coordinates (so far so good), but I always get the same face of the cubemap mapped on the sky. As I look around I always get the same face (and it wobbles a bit if I move the camera). My code:


    Texture::Texture(const wchar_t *textureFilePath, const std::string &textureType) : mType(textureType)
        //CreateDDSTextureFromFile(Game::GetInstance()->GetDevice(), Game::GetInstance()->GetDeviceContext(), textureFilePath, &mResource, &mShaderResourceView);
        CreateDDSTextureFromFileEx(Game::GetInstance()->GetDevice(), Game::GetInstance()->GetDeviceContext(), textureFilePath, 0, D3D11_USAGE_DEFAULT, D3D11_BIND_SHADER_RESOURCE, 0, D3D11_RESOURCE_MISC_TEXTURECUBE, false, &mResource, &mShaderResourceView);

// SkyBox.cpp:

void SkyBox::Draw()
    // set cube map
    ID3D11ShaderResourceView *resource = mTexture.GetResource();
    Game::GetInstance()->GetDeviceContext()->PSSetShaderResources(0, 1, &resource);

    // set primitive topology


// Vertex Shader:

cbuffer Transform : register(b0)
    float4x4 viewProjectionMatrix;

float4 main(inout float3 pos : POSITION) : SV_POSITION
    return mul(float4(pos, 1.0f), viewProjectionMatrix);

// Pixel Shader:

SamplerState cubeSampler;
TextureCube cubeMap;

float4 main(in float3 pos : POSITION) : SV_TARGET
    float4 color = cubeMap.Sample(cubeSampler, pos.xyz);
    return color;

I tried both functions grom DDS loader but I keep getting the same result. All results I found on the web are about the old SDK toolkits, but I'm using the new DirectXTex lib.

  • \$\begingroup\$ Your comment says "vertex shader" but you've actually posted your pixel shader; can you post your actual vertex shader please? There is most likely an issue with the position you pass from it to the pixel shader. \$\endgroup\$ – Maximus Minimus Apr 12 '18 at 9:29
  • \$\begingroup\$ @MaximusMinimus Sorry, my mistake, I edited the post \$\endgroup\$ – Luca Apr 12 '18 at 9:35

I found my error. The vertex shader outputs the homogeneous clip coordinates with SV_POSITION semantic before the model space position with POSITION semantic. The Pixel shader reads the input position from the first register so I was sampling the texture cube map using clip space coordinates, hence I was always facing the same direction. I was ignoring the warning from the compiler.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.