I have made a lot of mistakes in the last four days and learned valuable lessons about AgiSoft StereoScan, the free program I am using to convert stereo photographs into 3-D meshes. I noticed a problem that when I blended the photos in Photoshop first the floor often melded with the material and StereoScan treated it as a floor plane. Download the 3-D model.
After a lot of trial and error, I found a way to blend multiple photographs and build a 3D model in StereoScan. I ended up manually selecting and deleting everything except the illuminated material in each layer of the sequence. Then StereoScan could map the volumes of the multi-frame packet.
This is the method for converting stereo pairs of photographs into a 3D model:
- Record synchronized stereo photographs.
- Manually delete everything in the frame except the illuminated material.
- Load the blended left and right images into StereoScan
- Create a 3-D model
The models may, of course, be imported with other programs. I used Rhino with V-Ray to make a quick render of one of the simplified surfaces.