View Full Version : Faceted shadows with normal maps

asier cubero
07 July 2010, 10:15 PM
Hi, I'm a 3d studio max user and I have a litle problem with normal maps.

As you can see in the image, there are more than enough polygons (at least that's what I think), but when I activate the normal map, big enough, shadows appear faceted. The shadow is ray trace, the renderer mental ray and the materil is standard in oren nayar binn. The normal map has been tested in jpg and dds formats and there is no change. And if I increase the resolution of the mesh, facets appear smaller, but I don't want millions of polygons if they are not really necessary

Any idea?

Thanks in advance.

07 July 2010, 12:17 AM
Seems like you export settings of the normal map are wrong. Also be sure to load them with gamma 1.0.

asier cubero
07 July 2010, 09:07 PM
I don't know where those gamma settings are. If they are in preferences it's off. What are you talking about?
The method i used for the normal map was screen, changing to tangent the facets are almost gone, although it looks a bit dark now. It has a strange look depending in the angle.

You are right in one thing, the settings in zbrush can't be OK. Look at the result of every map I make

Borders are horribly done. I started from a square OBJ made of one single square. Then 12 subdivisions and I modeled the rocks. I wanted something tileable, so I didn't manipulate the borders. Just to give them a little of texture. Any idea of the reason?

07 July 2010, 10:54 PM
The gamma settings should be when you load your texture map into your main 3d-package - max, maya etc. It should be per bitmap, not globally changed. Your map looks more like a displacement than normal.

07 July 2010, 09:36 PM
no gamma correction for normals, disps, bumps or any weight maps as far as i know.

07 July 2010, 12:24 PM
I'm a wee bit puzzled by the second image: the one with the lightly superimposed wireframe. Looking from left-to-right about 3/4 of the way down from the top, I see two flat curved-surfaces that do not seem to have any corresponding part in the boulder upon which they are superimposed.

The jaggedness does appear to me to be related to the polygons; perhaps their existing normal-vectors sans the map. Are all of the surface normals of all of the polygons (in the underlying mesh) "facing out?" How does the render look with this normal-map disabled? Or under "plain Jane" (not ray-traced) light/shadow conditions?

What renderer settings are you using? I think I see noticeable differences between the top and the bottom edges of those shadows... and, left vs. right. Could that foreground boulder somehow be interfering? I'm trying to infer "where the error is versus where it is not," and from that to guess why.

I don't think you've provided enough information yet...

asier cubero
07 July 2010, 02:07 AM
Sorry sundialsvc4 for not replying sooner, but I haven't connected during the last days.
I can tell you all normal faces look to the outside.

This is without normal maps

And this with shadow map instead of raytace. Apart from that there is a skylight with the hdri from the background as a source of color.

I don't understand very well what are you refering when you talk about the two flat curved-surfaces. I can tell you there is a plane for the ocean and another one for the grass ground. The rocks are two. The upper one and the lower one that is bigger than the first one. So it behaves as a floor for the upper one.

The settings are the default ones. I activated the final gather and reduced the points to 25, but bigger numbers don't improve the result, just slows the render.

If you need more information, just let me know. And thanks for your time.

07 July 2010, 02:48 AM
In the second image of the first post of this thread, at the lower left-hand quadrant of the image there appears to be a curved wireframe that does not correspond to the geometry of the rock (in the first image). And, if you move straight across from that, almost but not-quite to the right hand edge, you see another one. That is all that I was referring to...

asier cubero
07 July 2010, 02:29 AM
This is a link to the 3ds file of that part of the scene. I think with that you will be able to see perfectly the distribution. It's in low resolution.

faceted (

Any idea?

CGTalk Moderation
07 July 2010, 02:29 AM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.