2D to 3D Conversions

Dan Jacob takes 2D images that secretly want to be 3D and converts them.

Dan Jacobs

When M.C. Escher created many of his illusionary images it is likely in his mind he saw them in 3D. A great example of this is when the 2 hands above draw each other off the flat paper into real dimensional hands. If Escher were alive today I think he would be very pleased to see many of his works transformed into the 3rd dimension. This is no small task and the people like Dan who do 3D conversions are true craftsmen.

There are a number of ways to go about doing 2D to 3D conversions. Some artists use what is called a “depth map” which is a grayscale image that matches the shapes in the original image and uses shades of gray to define the depth of each area in the image. The gray shades are set up so white is closest to you and black is furthest away from you in the distance. For example, if your image was of a simple landscape such as a desert, then the depth map would be a blend from white at the bottom to black at the top. The depth map image is processed by a program that manipulates the image based on the depth map inputs to create the second viewpoint. There is more detailed information on the depth map approach here.

Dan uses a more manual approach of slicing up the image and moving things around to simulate what a true stereo pair would be like and which creates the 3D effect. This approach requires a lot of detailed photo editing work. For example, when something is moved usually the background must be recreated to match what would have been there in a real view. An intricate scene will take a lot of work to mask each item so it can be moved to work as an alternate viewpoint to complete the stereo pair.

In a few of the samples below, Dan has created what is known as a Phantogram, it is a special type of 3D image meant to be viewed at a 45 degree angle so it looks like the object is sitting up off the plane of the paper or screen. If you are viewing this on a laptop, simply angle the screen down to achieve this effect. For best results, print it to view it correctly at an angle. You can still enjoy the image at any angle but you will notice an exaggerated stretching of the image at other angles.

Take a trip with Dan below as he converts the masterful work of M.C. Escher into the dimensional illusions the artist had in his mind.

Dan is active on Flickr

Dan Jacobs

Dan Jacobs

Dan Jacobs

Dan Jacobs

5 Responses to 2D to 3D Conversions

  1. David Kesner says:

    Oh how I wish I had Dan’s ability and patience so I could (re)create such works of art.

  2. But, from the depth map, how do you create the other image? It’s not just a shift of pixels, is it?

  3. neilnathanson says:

    Yes it is a shift of pixels, The second image is made by moving pixels based on how far back they are in space as defined by the gray values. Of course it is not perfect and some areas that would have been seen correctly are now gaps that get filled in by stretching pixels, that is the main limit of this approach. A skilled conversion artist will manually recreate these gap areas but a depth map shifting program can only guess at this by stretching adjacent pixels.

  4. Thanks for the answer. Any thoughts on getting intermediate frames from a stereo pair or getting a depth map from a stereo pair? Is there a proven method that’s superior to all others or is it pretty much all heuristics?

  5. neilnathanson says:

    There is software that will do intermediate frames like a morph but I have not tried them to say how well they work. Interestingly there is a website http://www.start3D.com that does this in order to make smoother wiggle type 3D (they call it piku-piku).

    Also this discussion talked about making a depth map from a stereo pair.
    I think the general consensus was it did not work well.

Comments are closed.