I trained this splat with Spirula Studio but I added a small patch that created a void inside the body of the bee. previous versions of this model added splats inside the reflective torso to help generate the highlights, this version has a more opaque surface.
I captured this scan using 226 focus stacks with a rig I built on a Raspberry Pi. I created software to drive the capture of all the stack positions. The capture took about 20 hours to run.
I used multi-background triangulation matting to generate an alpha channels/masks for each image, every focus position had 3 exposures. About 27000 images were used to create 226 transparent focus stacks. I processed the resulting images using batch processing software I created to remove the backgrounds and sensor dust using flat exposures. The stacking was done using Shine Stacker. A new alpha channel capability has been add to shine stacker that allows me to complete the stacks with transparent images.
The source images can be downloaded here: https://drive.google.com/drive/folders/1wQDvkSBGVv1xOVNjkFD5vVWmrPd9d4LG?usp=sharing
A smaller version of this is here.
https://superspl.at/scene/7b64ea2c