Horde3D

When mapping a texture as a stream, you get the format from the TextureElem, but then to actually get a pointer to the data you use the ImageElem, and likewise you get the size from the ImageElem. But if it has multiple slices, you get the number of slices from the TextureElem. This is really confusing. Why not just fold the properties of the TextureElem into the ImageElem?

Well, there is a system to it. A texture can have subresources (mip maps or cube map faces) which are the image elements. The format must be the same for all subresources, that's why it is stored in the texture element. The dimensions (width, height) can be different for each subresource (e.g. mip maps). The number of slices belongs of course to the container (which is the texture element).