An approximation style is the general subdivision scheme mental ray for Maya uses to break the surface into triangles. Some approximation styles work by repeatedly cutting the entire surface from end to end, while others are capable of adding triangles in a more localized fashion.
mental ray for Maya provides a few standard approximation styles (Grid, Tree, and Delaunay) and the Fine approximation style.
The standard approximation styles use as few triangles as possible to approximate a surface to achieve the quality you define in the approximation settings.
Fine approximation subdivides very complex surfaces (especially detailed displaced surfaces) into a large number of roughly uniformly-sized small triangles in order to guarantee a smooth result. To deal with the large number of triangles mental ray for Maya breaks the surface up into independent sub-objects that are each tessellated and cached separately, generating the triangles without consuming excessive amounts of memory.
This approximation style only supports the Spatial approximation method, which specifies the size of triangles to be generated.
Fine approximation cannot be used together with merging and connections (that is, when surfaces are stitched, there may be holes along the stitch).
| Geometry type | Can I use Fine approximation? |
|---|---|
| polygon displacement | yes |
| NURBS surface displacement | yes |
| subdivision surface displacement | yes |
| NURBS surface approximations | yes |
| curves | no (because they are not tessellated to triangles) |
To allow for Fine approximation, the granularity of mental ray for Maya’s cache manager is reduced, whereby smaller units are formed by splitting objects into smaller sets. These sets can be individually tessellated without excessive memory requirements.
Fine approximations support a small subset of approximation techniques since the other styles exist only to trade off triangle counts vs. quality, which is no longer a problem for fine approximations.
Except where otherwise noted, this work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 3.0 Unported License