Tools which render a z-depth image?

Community Forums/Developer Stations/Tools which render a z-depth image?

Matty(Posted 2006) [#1]
Hi, I've been working on a simple isometric engine that uses pre rendered scenes as a simple background engine. However it requires that I know the depth of each pixel relative to the camera viewpoint to make the image. This is a 2d engine not a 3d one.

I read that 3dsMAX can render a 'z-Depth Map' as a render element. However I cannot afford 3dsMAx so have to look elsewhere. Currently I use blitz3d and simply do a camera pick on every pixel in the scene to generate the depth map. However blitz3d cannot handle the high poly meshes that I would like to use as scenes. (Nice looking scene in poser - 300,000 polygons all in a single mesh goes over directx/blitz3d limit of polygons/vertices per surface meaning I cannot generate z map for these).

After that longwinded explanation of mine - are there any tools other than 3dsMAX which can render or generate a depth map for a scene? Preferably those which can read either poser files, or obj files.

EDIT - I actually just found a way to do it in poser - using depth cueing and changing object colors to black and background to white (as found on a tutorial on the web somewhere).

gellyware(Posted 2006) [#2]
XSI can do z-depth.

XSI Foundation is $495 (USD)

jfk EO-11110(Posted 2006) [#3]
You can even do it in blitz3d, using fog and white fullbright meshes on black background. Limited to 256 shades of grey tho.

Matty(Posted 2006) [#4]
Hi Jfk - The only problem with doing it in blitz3d is that the meshes I need to import into b3d have too many vertices/polygons to render (MAV at Renderworld). Does fog work though in orthographic camera projection mode?

boomboom(Posted 2006) [#5]
you can do this pretty much with any 3d package. Try blender

jfk EO-11110(Posted 2006) [#6]
>>Does fog work though in orthographic camera projection mode?<<

I guess yes. Tho not tested.

If your model has too many vertices or triangles per surface you may write a function that will clone the mesh and split the surfaces into smaller ones. The color doesn't matter since they all will be painted white. (where you need to prevent automatic surface count optimation I think, probably by using some individual brush parameters that have no affect on the visuals).