Acropora

I am using regions (3-dimensional volumes) and masks (for 2D scoping of the volume, much like your rules) to limit the effect of modifiers and operators. One could image a large tree structure containing multitudes of regions and masks that essentially derive the landscape or object. In the end you can generate a script that can be loaded into a shader (as Ryan Geiss did) along with LOD controls. On the other hand CG tools like MAX have no problem handling the large meshes. Quite often I optimise the meshes for faster performance. Anyway, it is a work in progress.

Hi Andrew, Nice to hear from you. Yes, exactly, this is what I had in mind, but to expand on the idea on procedural terrain/geometry. If you look at the examples on the net (Procedural terrain) you can see parallels with Acropora. The problem with this kind of application is the difficulty in designing the density function required to create an infinite landscape. Tools are limited if non-existent. Ryan Geiss programmed his shaders to do this, but tweaking such a landscape is difficult and impractical. By the way, excellant software, Andrew.

I'll have a look and get back, though I have a feeling we are talking apples and oranges here: procedural mesh generation vs modelling. All the meshes in the images provided on the web-site took a very short amount of time (typically an hour or less). However, that being said, the results were semi-deterministic. Thanks for the comments. PS: Take a look at the links sections on the web-sites. You can get an idea where the inspiration for Acropora came from...

Speed is always an issue when voxels are concerned. The code to employ the GPU is written (via CUDA) and shortly will be included. Processing times are expected to be in the 20-30x range over multithreading (as it is now). Remember, to create voxels from a large density field requires a lot of processing. Acropora divides the workload into blocks to expedite processing.

There are several artifacts usually associated with voxel mesh generation (typically a result of marching cubes): - numerical debris: small variations in the density field function that result in isolated polygons. - slivers: thin polygons that cause unsightly flicker and poor topologies To remove disconnected polygons and polygon groups, Acropora provides a sweeper function. The amount of removal is user controllable. Retopologizing (say this ten times fast :-)) takes care of slivers and other unsightly polygons.