Hi every one, i'm new on this site. But appear that is the best xsi site, so i can talk with all XSI user finally and try understand why ULTIMAPPER still buggy from xsi 3.5 until today. Possible that SOFTIMAGE before, and AUTODESK today never correct a tragic bug in the NORMAL MAP tangent space...?
I start to explain. The bug happen only when you generate a NORMAL MAP TANGENT SPACE.
The bug consist in a WRONG interpolation of the vertex color for the TANGENT CALCULATION on the VERTEX where the UV polygons are separated in another cluster.
So take al look to the image attached: image bug_1.jpg : a basic cylinder, with a very simple UV, the 2 cap are separated islands in the UV editor.
Image Bug_2.jpg show ho ULTIMAPPER generate WRONG vertex color for tangents in the MESH in some vertex where the polygons ISLANDS UV connect together.
Image Bug_3.jpg Show the Totally wrong distorsion of the normal map in tangent space and also in BENT mode.There are some points like "DROP of color". The result in shading realtime is like a bulge.
XSI ultimapper Historical MEGA BUG
-
yashugan
- Posts: 115
- Joined: 30 May 2012, 18:58
XSI ultimapper Historical MEGA BUG
You do not have the required permissions to view the files attached to this post.
-
yashugan
- Posts: 115
- Joined: 30 May 2012, 18:58
Re: XSI ultimapper Historical MEGA BUG
Me too, i continue to explain:
I maded a test in OBJECT SPACE, and the render i PERFECT.
See image bug_4.jpg.
Than i changed UV and i maded another test. The bulge change too seeking the new UV setup.
I maded a test in OBJECT SPACE, and the render i PERFECT.
See image bug_4.jpg.
Than i changed UV and i maded another test. The bulge change too seeking the new UV setup.
You do not have the required permissions to view the files attached to this post.
-
yashugan
- Posts: 115
- Joined: 30 May 2012, 18:58
Re: XSI ultimapper Historical MEGA BUG
And finally i tryed with X normal (a freeware very cool tool) and the result is this: PERFECT.
The tragic question is that EVERY time that you have a ISLAND in the UV the vertex COLOR is DAMAGED and the TANGENT calculation toon. I work perfct with OBJECT SPACE ALBEDO, DEPTH, MATCOLOR, OCCLUSION. But Tangent Space and Bent are always damaged in some part where the normal map mut be attached.
Some one have a solution?
thank for you attention!
Yashugan.
The tragic question is that EVERY time that you have a ISLAND in the UV the vertex COLOR is DAMAGED and the TANGENT calculation toon. I work perfct with OBJECT SPACE ALBEDO, DEPTH, MATCOLOR, OCCLUSION. But Tangent Space and Bent are always damaged in some part where the normal map mut be attached.
Some one have a solution?
thank for you attention!
Yashugan.
You do not have the required permissions to view the files attached to this post.
-
rray
- Moderator
- Posts: 1810
- Joined: 26 Sep 2009, 13:51
- Location: Bonn, Germany
Re: XSI ultimapper Historical MEGA BUG
Hi, yashugan
Could it be that the smoothing value is too high? I think there should be no smoothing across different islands. The map will look less smooth then + have discontinuities in its color, but that should be OK because the same discontinuities are in the tangent map which is used as the base vector orientation. This should let the render look smooth.
(edited...) xnormal uses a custom tangent space (base vector) which is smooth.. but the method must be implemented in the software (source code here) for xnormal's tangents to render correctly.
Could it be that the smoothing value is too high? I think there should be no smoothing across different islands. The map will look less smooth then + have discontinuities in its color, but that should be OK because the same discontinuities are in the tangent map which is used as the base vector orientation. This should let the render look smooth.
(edited...) xnormal uses a custom tangent space (base vector) which is smooth.. but the method must be implemented in the software (source code here) for xnormal's tangents to render correctly.
Last edited by rray on 09 Jul 2012, 08:50, edited 2 times in total.
Reason: I wrote rubbish
Reason: I wrote rubbish
softimage resources section updated Sep 26th 2026
-
yashugan
- Posts: 115
- Joined: 30 May 2012, 18:58
Re: XSI ultimapper Historical MEGA BUG
The error is REAL. IS simply to demostrate: If i re-do the normal map in OBJECT space, the bulge don't exist. That because the tangent are not calculated. And also for a simple REASON : The bulge don't EXIST in the OBJECT 3d. Xnormal use the same operator to calculate the normal map. I can compare XSI ultimapper in some cases (where the UV are unvrapped and unified) and xnormal , and the result is the same. It change only WHEN the ISLAND of UV are separated.
The most simple stuff is try! The errors happen.
I use ultimapper strongly from year 2005 and i always haved this problem. But during these 7 years i also try ATI normal mapper (use standard tangent operator) and in some cases the normal map where similar (execpt that XSI create the ERRORS)
This can be tryed also by yourself. I'll prepare another test to explain better what happen...
Thank!!!!!!!
The most simple stuff is try! The errors happen.
I use ultimapper strongly from year 2005 and i always haved this problem. But during these 7 years i also try ATI normal mapper (use standard tangent operator) and in some cases the normal map where similar (execpt that XSI create the ERRORS)
This can be tryed also by yourself. I'll prepare another test to explain better what happen...
Thank!!!!!!!
-
rray
- Moderator
- Posts: 1810
- Joined: 26 Sep 2009, 13:51
- Location: Bonn, Germany
Re: XSI ultimapper Historical MEGA BUG
xNormal uses a different tangent space generation (see link: http://wiki.blender.org/index.php/Dev:S ... ormal_Maps)
This results in a smoother looking normal map, but in the rendered result it shouldn't make a difference whether the normal map looks smooth, has bulges or not. In theory at least ;)
This results in a smoother looking normal map, but in the rendered result it shouldn't make a difference whether the normal map looks smooth, has bulges or not. In theory at least ;)
softimage resources section updated Sep 26th 2026
-
yashugan
- Posts: 115
- Joined: 30 May 2012, 18:58
Re: XSI ultimapper Historical MEGA BUG
Here a test to explain better. I also compared some x-normal and ultimapper results and seem that in the end x-normal is not more smoothed than xsi.
But this is not the problem. My problem is discovert how to avoid the BUG of the bulge in xsi, because the ultimapper is integrated and is very usefull.
Please if some one have discovered how to avoid this problem should be nice! Thank for you help guys!
Please if some one have discovered how to avoid this problem should be nice! Thank for you help guys!
You do not have the required permissions to view the files attached to this post.
-
ActionArt
- Posts: 853
- Joined: 25 Nov 2010, 17:23
- Location: Canada
Re: XSI ultimapper Historical MEGA BUG
It certainly seems there's something wrong there. I don't know the answer but you might want to post this on the XSI mailing list. There's a lot of experience people on there that might be able to help.