Originally posted by politiepet
I thought that was called tri-linear mip mapping, or is that the same?
Not necessarily for those pics, but the principle is similar (check out the link I provided).
If you read the definition of mip-mapping, you will see that different versions of the same texture are used to reduce memory bandwidth requirements.
However, the problem with mip-mapping is that if you focus on a texture and slowly move forwards or backwards in a game, you can often see the moment when the renderer switches between different versions of the texture.
Tri-linear filtering uses bi-linearly filtered textures, but helps resolves the mip-mapping problem by taking samples from TWO versions of the same texture. That way, instead of seeing a sudden \'jump\' as the renderer switches to a lower or higher res texture, you will see a smoother transition instead.
In static screenshots, only a very keen eye will notice any difference between tri/bi-linear filtering. However, in game, the slight improvement of tri-linear filtering over bi-linear filtering is more noticeable.