Hi all,
My idea is to use Seith's MD5 importer to render models in mental ray with maximum quality

I've imported a model in Maya, changed the default lambert textures to blinn, assigned specular maps to specular color and normal maps to normal camera, the diffuse map being assigned by default to material color.
The problem is i don't know how to assign the bump maps (*_h.tga) because material bump is already in use with normal map

I don't want to do displacement mapping because changing the number of polys in the model will alter the UV mapping...
Any idea will be welcome, thanks in advance

LordBelial is right. The ONE bumpmap that is actually used in Doom3 for the rendering is generated from the bumpmap(s) and the heightmap(s) as defined in the material shader's bumpmap stage, using the "addnormals(bumpmap, bumpmap)" command and the "heightmap(heightmap, scalefactor)" command that turns a heightmap into a bumpmap to be used as an argument for the addnormals command.
I don't know much about Photoshop or the nvidia plugin that most people are using, but there's probably a way to add normalmaps onto each other, too. Turning heightmaps (grey) into bumpmaps (blueish) is what most people use it for but it can do more than that I think.
Thanks a lot for your help guys

You gave me some ideas : i will try to blend the heightmap and the normal map textures in Maya before assigning the resut to normal camera channel, like it's done in Doom3.
Perhaps i will need the nVidia plugin to add the two layers in Photoshop if i can't do the blending in Maya, but i don't know if it's possible to add normalmaps like that, because a normalmap RGB information is a representation of normal vectors and adding color information in Photoshop may produce bad resulting vectors...
By the way Der Ton, do you remember the name of the nVidia plugin ? I think i've already heard of it but can't put a name on it.