Getting from raw microscopy data to a 3D mesh that renders well in a browser involves more steps than I expected. Here's the pipeline I landed on.
The first stage is UCSF Chimera: load the volumetric data, adjust voxel sizing, step size, and isovalue to get the surface you want, then export as a single nested .obj file. Chimera's .obj export has one advantage over ChimeraX: it packages everything in one file, which simplifies downstream handling.
Baking ambient occlusion in Cinema4D
Stage two runs in Cinema4D via the Python SDK. The script imports the .obj, merges it with a starter file containing render settings, applies polygon reduction and smoothing via deformers, generates UV maps using spherical projection, then bakes ambient occlusion at 1024×2 resolution with a 3-pixel border. Key settings: Maximum Distortion = 1%, Relaxation Steps = 0.
The last stage loops through the full file list. Polygon count is the main variable that affects processing time — I found that reducing the polygon-reduction percentage from 90% dramatically improved performance without meaningful loss in visual quality at web resolutions. The output is a .obj plus a .png texture that three.js loads directly.