Re: Ultimapper- Normal map issues

Date : Thu, 31 May 2007 14:23:58 -0500
To : XSI(at)Softimage.COM
From : Rares Halmagean <rares(at)janimation.com>
Subject : Re: Ultimapper- Normal map issues
I see what you're saying. I'm not sure the uv's have anything to do with this particular problem as they are very clean with minimal distortions. I unwrapped the object using Roadkill. See for yourselves;



I also applied the texture onto the object itself to see where on the topology the pink is exposed;



I just tried your method Pedram and it looks like it did the trick which is strange because i did the same thing yesterday and it didn't seem to work. The pink covering the entire front side is no longer there now though which what i expect a 'normal' normal map to look like :-) .

Now this is what I expect a 'normal' normal map to look like.

Thanks guys for your suggestions. Really helped a lot.


Rares Halmagean
2D / 3D Digital Artist





André Adam wrote:
The tangent operator, unless accidentially frozen, is active all the time nonetheless. Adjusting the operator's smoothing value will have great effect though. That's for compensating on the aforementioned uv-space distortions, meaning it will work less well for heavily distorted uv spaces. It really comes all back to creating a good texture projection. The more you distort a texture projection to wrap around an object, the less good the normal map results will be. Finally, cracks in the texture projection are a no-no, they are a guarantee for cracks in the normal map and rendering result as well.
Normal mapping is a very unprecise technique with lots and lots of variables influencing the quality of the result...

   -André


Pedram Karimfazli wrote:
remove the tangentmap from the object, then tryfeezing your transforms and reopen the utlimapper pp so it generates a new tangentmap and try generating a new normalmap.

On 5/31/07, * André Adam* <a_adam(at)49games.de <mailto:a_adam(at)49games.de>> wrote:

    Nope, the tangent vector information encoded as a basis in the normal
    map completely depends on the texture projection. If your texture
    projections shows distortions relative to the polygon structure's
    shape
    (very likely), you will see additional colour information in the
    normal
    map to compensate for that. To say it as clearly as possible, the
    straight and artifact-free normal maps one can generate with photoshop
    plugins are plain wrong when it comes to 3d, which is the source
    for a
    general misunderstanding of the technology for many people...

    Does your map look wrong when rendered on the object?

        -André


    Rares Halmagean wrote:
    > Basically the pink banding you see there should not be present.
    As a
    > normal map from tangent space overall color present in the map
    should
    > be blue with variations in color present only in the wrinkles.  I
    > don't know whats creating the pink color but it's not supposed
    to be
    > there.
    >
    > *Rares Halmagean
    > 2D / 3D Digital Artist *
    > <http://www.janimation.com>
    >
    >
    >
    >
    > André Adam wrote:
    >> Basically, what's wrong with the tangent map you posted? Looks
    ok to
    >> me on first sight...
    >>
    >>    -André
    >>
    >>
    >> Rares Halmagean wrote:
    >>> I'm trying to export a normal from tangent map from an imported
    >>> zbrush mesh using ultimapper but the resulting map doesn't
    look right;
    >>>
    >>>
    >>>
    >>> So I extracted polygons from the zbrush mesh thinking there
    may be
    >>> something wrong with the imported zbrush mesh but results remained
    >>> the same. I then extracted a new mesh from original xsi mesh and
    >>> then exported a normal seeing if that would yield better results.
    >>> Instead I got this;
    >>>
    >>>
    >>>
    >>> Heres the settings I used;
    >>>
    >>>
    >>> --
    >>>
    >>> *Rares Halmagean
    >>> 2D / 3D Digital Artist *
    >>> <http://www.janimation.com>


---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi


Search the XSI List archives here or use the advanced search form to search across mailing lists. Searching help is available.
This site supposedly brought to you by Benjamin Grosser and the Imaging Technology Group.