« OpenGL ES specs on the iPhone | Main | Rapid Eye: Every Day a picture from any point on earth - made in Brandenburg/Germany »

Sigraph Paper: Capture Normal Maps with Camera Flashlight

In my ongoing series about absolutely stunning developments from Siggraph this year comes a techdemo /white paper that is so simple yet so powerful. Every serious 3D artist knows that it is hard to capture real life textures and replicate them with lots of details and try to make them appear haptic. Tricks like making fake bump/displacement maps by extracting a range of values in a certain color channel only get you so far and touch up is time expensive.

Now Mashhuda Glencross and Gregory Ward from the University of Manchester in the UK and Dolby Canada in Vancouver developed a dead simple - everyone can do it way to extract the normal map of any photo that is shot twice - one time with normal exposure and one time with the flash turned on. From that the get the depth data of the surface. Look at the video of how simple this is - no commercial product yet but believe me that this takes about two weeks to surface (if it isnīt patent locked in some closet).

Watch the video to see how it works and be amazed:

Original Article over at the NewScientist

And I found this tutorial in the comments over there on how to fake this with 4 photos and a lightsource other then camera flash.

TrackBack

TrackBack URL for this entry:
http://prototypen.com/cgi-bin/mt/mt-tb.cgi/2249

Post a comment