<%@ include file="../../check-login.jsp" %> Subdivision.org: Re: Loop Adaptive Subdivision Problem <% session.setAttribute ( "messageNum", new Integer ( 1711 ) ); %>

Re: Loop Adaptive Subdivision Problem

Posted by ati3414 on March 16, 2005 at 10:40:27:

In Reply to: Re: Loop Adaptive Subdivision Problem posted by Settgast on March 14, 2005 at 06:20:06:

: : In my work, I noticed that while the tesselation changes based on the distance to the viewer the normals have to be attached to points on the surface, any other combination produces a "shimmer" effect destrying lighting under animation. By contrast, subdivision surfaces subjected to subdivision (either fixed or adaptive) should exhibit this shimmering artifact since there is no symbolic representation of the final surface, each approximation producing new vertices, none of which residing on the same surface.

: We used the subdivision rules by Catmull Clark and Loop. For those there are so called limit point and limit normal rules available.
: (Biermann, Levin, Zorin: "Piecewise Smooth Subdivision Surfaces with Normal Control" e.g.)
: There is no shimmer effect and all the points are on the surface.

I read the Zorin paper: only the limit points and the limit normals are on the the limit surfaces. Subdivision surfaces represent a series of polygonal approximations, each instance DIFFERENT from the previous one. If you did a simple animation of flying an object along the z-axis, as you traverse levels of details, the coarseness of the approximation keeps changing and so DO THE NORMALS AND THE VERTICES. This is why I am so concerned that the lighting and the texturing will jitter (shimmer) if one implements adaptive subdivision.
By contrast, when I applied my method to parametric surfaces ALL THE NORMALS AND THE VERTICES belong to the limit surface at ALL TIMES. I belive that this is the reason why there absolutely no jittering for adaptive subdivision under animation.

Follow Ups:

Post a Followup

Name:
E-Mail:

Subject:

Comments:
: : : In my work, I noticed that while the tesselation changes based on the distance to the viewer the normals have to be attached to points on the surface, any other combination produces a "shimmer" effect destrying lighting under animation. By contrast, subdivision surfaces subjected to subdivision (either fixed or adaptive) should exhibit this shimmering artifact since there is no symbolic representation of the final surface, each approximation producing new vertices, none of which residing on the same surface. : : We used the subdivision rules by Catmull Clark and Loop. For those there are so called limit point and limit normal rules available. : : (Biermann, Levin, Zorin: "Piecewise Smooth Subdivision Surfaces with Normal Control" e.g.) : : There is no shimmer effect and all the points are on the surface. : I read the Zorin paper: only the limit points and the limit normals are on the the limit surfaces. Subdivision surfaces represent a series of polygonal approximations, each instance DIFFERENT from the previous one. If you did a simple animation of flying an object along the z-axis, as you traverse levels of details, the coarseness of the approximation keeps changing and so DO THE NORMALS AND THE VERTICES. This is why I am so concerned that the lighting and the texturing will jitter (shimmer) if one implements adaptive subdivision. : By contrast, when I applied my method to parametric surfaces ALL THE NORMALS AND THE VERTICES belong to the limit surface at ALL TIMES. I belive that this is the reason why there absolutely no jittering for adaptive subdivision under animation.

Optional Link URL:
Link Title:
Optional Image URL: