Scientists from University of Manchester and Dolby Canada in Vancouver have figured out a very simple method to capture textures and assign them with depth-information. You will need this to model rooms, walls or characters with specific height information, in order to set height-maps, depth-maps or albedo-maps. Capturing them was a time consuming task that had to be done by hand or that required special expensive equipment like laser-scanner. Now we got this:
If the video does not start correctly, move the slider somewhere in the middle of the video.
You just take pictures of the desired texture. One with flash, one without. Some post-processing on the colors (that differ by the color of the flashlight and specific shadows) result in the height map, we are ready to use. This technique should not give *that* accurate results, but at least it works perfectly, in relation to the cost. So, happy hacking and modeling! More details on Gizmodo and NewScientist.
Realy cool finding,
I don’t know if you have visited my blog recentely, but I have made a post about Microsoft’s new tool Photosynth.
3D restitution from plain pictures seems a very active research field.
http://beart.wordpress.com/2008/08/23/new-3d-modelisation-tool-by-microsoft/
Posted on August 28th, 2008 at 15:25