Fri Apr 28, 2017 2:40 pm
Sorry. If you set the Texture2D.FilterMode.GL.HiQ to Trilinear or Anisotropic, Fusion will make MIP maps for your input. It takes a tiny bit longer to upload the texture, but it's roughly the same speed to render.
Trilinear textures will evaluate the screenspace partial derivative of the UV coordinates and sample between MIP levels based on that. If your fragment is far away, small, whatever, it will use the lower resolution texture. If it's close, large, whatever, it will use the higher resolution texture. It blends between them smoothly (which is where the trilinear part comes in). Anisotropic is the similar, it just makes the MIP maps where instead of uniformly scaling the texture down, it does it in X and Y separately. Then it evaluates the partial derivative in separate components, so if your texture is small in U but large in V, you'd sample from a MIP level where the X is scaled down more than the Y. If you're just moving a facing image plane towards and away from the camera, trilinear is fine. If you have things receding into the distance, like a floor or something, then you'd want to use anisotropic filtering.
Chad Capeland
Indicated, LLC
www.floweffects.com