Hi,
I have been always fascinated with the idea to create normalmaps+heightmaps from photos. Especially now that we have DirectX 11 and Tesselation and Paralax Occlusion mapping. The usual way to take a photo and convert its brightness values with some filter combinations into a Heighmap might work for normal maps but isnt suitable for displacement mapping. Therefore I want to write a program that uses stereo photopairs to compute real depth values. There is a program out there called 'PhotoSculp' that uses exactly this approach. But since I am interested in computer vision I'd like to make my own 'Converter'. So now I wanted to ask all the artists out there what features they would like to see in such a program and what you are missing in existing ones. I just want to collect some ideas for the project, so if you are interested to share your ideas I would highly appreciate it.:)
Replies
-Combine with diffuse bump/emboss for added detail w/slider & preview
-Save/load modifier stack
-Preview as tessellation/normal map/parallax map with or without diffuse/spec
-Batch
I'd ask for automatic generation of ao and spec but that is getting a little out of scope.
I think extracting details from the diffuse map is a great idea.
Unfortunately I dont know what you mean by "Save/load modifier stack"
and I am also unsure what a "profile curve" is. Maybe you can explain that in more detail.