1

For research purpose, is there any existing algorithm that can transform 2d image to 3d models? For example, transforming a circle into sphere or transforming an image of apple into 3d model of apple. Simple algorithm or imperfect algorithm is fine too. If there isn’t, what are the fields or areas that I should take a look into to know more about it?

Han Lim
  • 661
  • 7
  • 17

2 Answers2

2

It's worth searching here http://make3d.cs.cornell.edu/code.html There are the sources of 2D-to-3D propcessing algorithm based on the conception of neural networks. This project was introduced by prof. Andrew Ng team.

Video demo of conversion: https://www.youtube.com/watch?feature=player_embedded&amp=&v=GWWIn29ZV4Q

Alexey R.
  • 6,081
  • 1
  • 8
  • 25
-1

Very funny idea, as it is obvious that there isn't enough information in a 2D image for precise 3D reconstruction. Exploiting some a priori knowledge of the real world, you can do something, but this is far-reaching.

Much simpler: see the image intensity as a depth information and you have a so-called range image, which is 3D (but the surfaces will be quite unnatural)

Yves Daoust
  • 53,540
  • 8
  • 41
  • 94