Yoda becomes 3D from one single image!
Recently there are a bunch of work done on generating 3D models from one single image. Previously I 3D printed and painted a Yoda model. Seeing it sitting on my desk, I’m thinking maybe it’s time to take a photo of it and see what the 3D model generated looks like!
Input image (Left) Generated Depth Map (Middle) Generated Surface Normal (Right)
Generated result from MCC, quite interesting the network can understand the head is a smooth oval shape
Generated result from Stable DreamFusion, looks more realistic, but yoda’s jaw looks like punched in!
The two methods I tried, they both can imagine the back head of Yoda, but you can tell the detail of face is not precise enough. As the blog is written, there are many more new research being done in this field. Cannot wait to see the potential of this in game modeling, interior design, architecture, etc!