The Cuttlefish library provides utilities to load images, perform operations on the images, convert to a texture, and save the final result. The texture may be saved in a common format (DDS, KTX, or PVR) or the raw data may be extracted to be used in another manner.
The Image class represents an input image. It may either be loaded from a file, a data buffer, or generated programatically. The data may be accessed directly to be manipulated by the program. Some common operations are provided by the Image class. See the Image class for a full list of operations that are supported.
Once the source images have been created, the Texture class is used to create the final texture. After initialization, the images are set through Texture::setImage(). This must be called for all surfaces, such as cube faces, depth slices, and array indices.
When generating a mipmapped texture, you can avoid having to set each mip level manually by calling Texture::generateMipmaps(). When calling this, you should initialize the texture with one mip level and set the images for mip level 0 before calling Texture::generateMipmaps().
Once all images have been set on the texture, Texture::convert() may be called to convert to a texture usable by the GPU. The parameters include:
- The format. This is either the bit depth per channel a compressed format.
- The type of data stored for each channel.
- The quality used for texture compression, which also affects conversion speed.
- The color space of the image. This may be used to flag the GPU to convert from sRGB to linear during texture sampling.
- The alpha type and color mask. This may be used for compressed formats to adjust pixel weights.
- The number of threads to use during conversion. If 1 is provided, no threads will be spawned.
After conversion, Texture::save() may be called to save to a DDS, KTX, or PVR texture. Alternatively, the Texture::data() and Texture::dataSize() accessors may be used to access the raw data for each surface.