DirectX Computational Graphics: Textures Tutorials

Textures in DirectX

Textures are fundamental to creating visually rich and realistic 3D graphics. They are 2D images that can be mapped onto surfaces of 3D models to add detail, color, and visual properties. This section provides a comprehensive guide to understanding and implementing textures within DirectX.

What are Textures?

In computer graphics, a texture is a bitmap image that is applied to the surface of a polygon or other geometric primitive. This process is called texture mapping. Textures can represent a wide variety of surface properties, including:

Color: The most common use, providing the base color of a surface.
Surface Details: Bump maps and normal maps can simulate surface irregularities without adding more geometry.
Material Properties: Specular maps, roughness maps, and metallic maps control how light interacts with the surface.
Transparency: Alpha channels in textures can define areas that are transparent or semi-transparent.

Texture Formats

DirectX supports various texture formats, each optimized for different use cases and hardware capabilities. Common formats include:

Uncompressed: DXGI_FORMAT_R8G8B8A8_UNORM (32-bit RGBA), DXGI_FORMAT_B8G8R8A8_UNORM (32-bit BGRA).
Compressed: BCn formats (e.g., BC1, BC3, BC7) offer good quality with reduced memory footprint and faster loading times.
Floating-Point: DXGI_FORMAT_R32G32B32A32_FLOAT for high-precision data.
Depth-Stencil: Used for depth buffers and stencil buffers.

Loading and Creating Textures

Textures can be loaded from image files (like .dds, .png, .jpg) or created programmatically. The process typically involves:

Defining the texture properties (dimensions, format, mipmap levels).
Allocating GPU memory for the texture.
Uploading pixel data from a file or generated data to the GPU memory.

The DirectXTex library is a popular and powerful tool for loading, saving, and manipulating textures, including conversion between formats and generation of mipmaps.

Sampling Textures

Once a texture is on the GPU, it needs to be sampled in shaders to retrieve pixel color (texel) data. This involves:

Creating a Shader Resource View (SRV) to bind the texture to a shader stage.
Using sampler states to define how texture coordinates are interpreted (e.g., filtering, wrapping, addressing modes).
Writing shader code to access texture data using functions like Sample() or Texture2D.Sample().

Mipmaps

Mipmaps are pre-calculated, downscaled versions of a texture. They are crucial for performance and visual quality by allowing the GPU to select the most appropriate resolution of the texture based on the object's distance from the camera. This reduces aliasing artifacts and memory bandwidth.