RenderMan Quality Settings

The quality settings for RmF are certainly fewer than those for Mental Ray, but there are a couple basics in the Render Globals under the Quality tab that are both easy to manipulate and also very effective. Shading Rate essentially refers to the antialiasing quality of bitmap textures and procedural shaders, such as Ramps. The default value is 5 and typically yields satisfactory results. However, when rendering something as simple as a Surface shader whose color is mapped to a Ramp with no interpolation (e.g. a V Ramp that has straight bands of blue, green and red), choppy stairstepping can be seen between the colors.





















A render with the Shading Rate set to 5.0. There is noticeable stairstepping between the green and red. Note however that the profile of the sphere (which is being subdivided at render time) is perfectly smooth and sharp.

















When the Shading Rate is set to 1, the aliasing smooths out to a crisp delineation between colors. We sacrifice render time for quality in this case.

















Pixel Samples refers to a more general antialiasing quality. By default, it's at 3. This will generally give good results, and is almost always completely sufficient for a still-rendered image, but I've found that it needs to be increased for animated scenes, especially those that involve motion blur.

A still render with a Shading Rate of 1.0 and Pixel Samples at (1.000, 1.000.) The color banding is even worse than that in the first example, and the profile is severely degraded.

















Here are three examples of how Pixel Sampling affects motion blur. From left to right, Pixel Samples are set to (1.000,1.000), (3.000,3.000) and (6.000,6.000); The Shading Rate is set to 1.0 for each image.








The Motion Blur settings are as follows:





















Everything is at its default setting except for the Shutter Opening's first value, which I set to .9 (the default is 0.0.) Simply stated, with this at .9 and the second value at 1.0, the blur will be derived mostly from the previous frame to the present frame, creating a more realistic effect. If we were to leave the initial value at 0.0, the blur would be derived equally from the previous frame and from the next frame.