I have been experimenting with photogrammetry lately : a computational method for producing 3D models based on photographs taken from many angles.
This stone bear figure from New Mexico is about 2.5 inches long and is a great semi-complex small object to practice photogrammetry. I have been using a Canon 6D with a 50 mm macro lens to capture images. To begin, you simply need to take a series of photos from many angles and load them into the software. Overall, the software works really well automatically, but I have noticed that some optimization is definitely required to create a final accurate model.
Beyond making sure you have covered all angles of the subject (and that all photos are similarly exposed, etc.), it is important to consider the surface the object rests on. Newspaper or something similar provides reference points that make it easier for the software to stitch all of the photographs together. I was also able to make models of small objects by placing them across terrain lines on a map.
After loading your images, it takes a few minutes for the program to run and build the model. If all went well (it is possible to encounter errors if there is a problem with the photos), the model is loaded and can be manipulated and refined in 3D dimensions. If you find any obvious problems you can manually stitch points in the photos together at this point as well.
This is a fascinating technique. I have a feeling this bear is just the beginning of quite a bit of photogrammetry I will be doing in the future.