16.4.3 Optimizing Texture Mapping

Follow these guidelines when rendering textured objects:

• Avoid frequent switching between texture maps. If you have many small textures, consider combining them into a single larger, tiled texture. Rather than switching to a new texture before drawing a textured polygon choose texture coordinates that select the appropriate small texture tile within the large texture.
• Use texture objects to encapsulate texture data. Place all the glTexImage\*() calls (including mipmaps) required to completely specify a texture and the associated glTexParameter\*() calls (which set texture properties) into a texture object and bind this texture object to the rendering context. This allows the implementation to compile the texture into a format that is optimal for rendering and, if the system accelerates texturing, to efficiently manage textures on the graphics adapter.
• If possible, use glTexSubImage\*D() to replace all or part of an existing texture image rather than the more costly operations of deleting and creating an entire new image.
• Call glAreTexturesResident() to make sure that all your textures are resident during rendering. (On systems where texturing is done on the host, glAreTexturesResident() always returns GL_TRUE.) If necessary, reduce the size or internal format resolution of your textures until they all fit into memory. If such a reduction creates intolerably fuzzy textured objects, you may give some textures lower priority.
• Avoid expensive texture filter modes. On some systems, trilinear filtering is much more expensive than point sampling or bilinear filtering.