I just was dealing with the same exact issue. I had un-archived an old scene
with a lot of reflection which I did not notice having this problem before.
Please note that it is unsure whether these issues existed before. It was
full up and rendered in 3.0 with out any noticeable issues in the
reflection. Now I am modifying (a lot of it) that same scene and I have had
to add hard edges and more local subdivisions to force the normals in the
right direction. Just for your confirmation.
-Scott
-----Original Message-----
From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of
André Adam
Sent: Friday, June 02, 2006 12:06 PM
To: XSI(at)Softimage.COM
Subject: Re: AW: Reflections..
If you mean the reflection issue, that's rather quickly explained, to
calculate reflection (as well as highlights and shading, btw) the
surface's normals are evaluated to determine the surface's direction,
which of course is needed to find out what's reflecting on this surface
from the current point of view.
Now, since we're dealing with triangles that are only an approximation
of a real surface, the limited information of the vertex normals gets
interpolated across the polygons. The distortions are caused by a too
broad approximation of the modelled topology relative to what it is
meant be.
In Martin's example I guess the hole in the middle of the geometry is
causing the problems where the unified vertex normals around the hole
get bent towards the hole's center, therefore distorting the
interpolated normals on the main surface.
The quickest workaround is to create hard edges to break up the normals
so that all normals contributing to the main-surface's normals point in
the very same direction.
A better fix would be to create more geometry at the right places to get
a better triangle approximation of the real-life surface he is trying to
simulate.
Hope that helps and makes sense! Cheers!
-André
Xsiberger wrote:
>I also would appreciate any link/book which describes the theory behind
this
>phenomenon. is it called surface discontinuity? google didn't deliver
>something usefull to this subject.
>
>
>
>-----Ursprüngliche Nachricht-----
>Von: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] Im Auftrag
von
>Martin Belleau
>Gesendet: Freitag, 2. Juni 2006 16:27
>An: XSI(at)Softimage.COM
>Betreff: RE: Reflections..
>
>Just like few people understand the concept of scene optimization. What's
>clean and what isn't.
>
>Do you have a link so I can read up on this (normals)?
>
>-----Original Message-----
>From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of
>kim aldis
>Sent: Saturday, June 03, 2006 1:06 AM
>To: XSI(at)Softimage.COM
>Subject: RE: Reflections..
>
>It's to do with the way normals are calculated on points by averaging plane
>normals of contributing faces. If you have a think about how this works it
>starts to make sense.
>
>An understanding of how shading works and how normals are built is
>fundamental to good poly modeling and it always surprises me how few people
>understand it.
>
>Chris, you still have your beveled cube thing kicking around?
>
>
>-----Original Message-----
>From: owner-xsi(at)Softimage.COM [mailto:owner-xsi(at)Softimage.COM] On Behalf Of
>Martin Belleau
>Sent: 02 June 2006 14:37
>To: XSI(at)Softimage.COM
>Subject: RE: Reflections..
>
>Oki it works much better... But what happens if I don't want to use hard
>edge, and use a Bevel...
>
>
>
>
>---
>Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
>unsubscribe xsi
>
>
>---
>Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
>unsubscribe xsi
>
>
>
>
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
---
Unsubscribe? Mail Majordomo(at)Softimage.COM with the following text in body:
unsubscribe xsi
