Alt Docs

Alt Docs - Week 2: Volumetric Capture

For volumetric capture, I chose photogrammetry to scan my toy Tidehunter via photoScan and Meshlab.

Tidehunter, a hero from game DOTA 2

Tidehunter, a hero from game DOTA 2

photoScan photogrammetry

photoScan photogrammetry

meshlab editing

meshlab editing

sketchfab editing

sketchfab editing


  • For photogrammetry, when capturing the object, we need to make sure:
  1. No overexposed highlight area in photos.
  2. Adjust the lighting condition to make the highlight on the object as little as possible, baby powder can help sometimes if we need only need the geometry.
  3. Try to reduce the reflection and shadow from the base, a rough textile will help
  4. Usually, over 70 captures and over 80% photo aligned should be fine.
  5. The positions of capturing are like rings of different ranges. It doesn't have to be linear.
  • For photoScan and Meshlab usage:
  1. Follow the workflow, and do it step by step. Uncheck Generic preselection and lower the accuracy will give us the better photo-aligned result.
  2. Duplicate chunk after each workflow if the process takes too much time. 
  3. Export as the OBJ format and then it can be imported to Meshlab.
  4. In Meshlab, select vertexes and delete them to do mesh editing. Wireframe view is helpful.
  5. OBJ exported from Meshlab doesn't include the material. But it can be found from the photoScan one. Compress OBJ + MTL + JPG(or other format materials) into a ZIP file, which will be sketchfab ready.
  • For sketchfab:
  1. We can play with some basic 3D settings on sketchfab to give us a better result. I think Shadeless looking is better then LIt in my case.
  2. Sketchfab can allow us to embed 3D models via iframe: