[vtkusers] (yet another) Opacity Issue (hmm)

Marius S Giurgi rendezvous at dreamxplosion.com
Thu Jan 13 20:07:23 EST 2005


I know, I know. there's a lot of feedback regarding the opacity of 
overlapping geometry issue, but to my disappointment, after spending a 
loooong long time reading through the VTK archives, I still couldn't 
find a satisfactory answer to this (rather annoying) problem. By the 
way, most of the 3D software (I know) out there have this issue already 
solved (3DS Max, Alias|wavefront Maya etc). I wonder how comes VTK (4.4 
and counting) still has issues with opacity? I'm sure there should be 
some smart developers out there to figure this out and have it history 
once and for all. Those VTK developers should probably take a peek and 
see how the smart guys at SGI (Alias|Wavefront: MAYA) approached this 
issue.

(the riddle of the intersecting transparent polygons).
In Maya (Alias), for instance, I create two perpendicular planes 
(intersecting each other), give them two different colors and set their 
transparency to 0.5 (let's say). They display fine both in the 
interactive window and in the rendered scene without any artifacts. How 
do they do it? Magic.

Yes, I came across the vtkDepthSortPolyData "cute" idea, which is not 
only slowing down the whole pipeline but also fails to work on the 
simple scenario of two intersecting polygons (both having opacity < 1). 
Besides I'd like to see this "sorting" be part of the rendering engine 
rather than have it as an extra element in the pipeline (via some 
DepthSort class of sort).

Also I noticed a similar (opacity<1) issue while I was rendering a 
scene containing a 3D volume (with opacity<1) combined with some 
polydata (also with opacity<1). If the volume is in front of the 
PolyData, the volume fails to hide the PolyData. Again, some sort of 
ordering issue.


God speed our VTK developers' way in getting this issue solved!

marius

PS: if there IS a solution tho this riddle, please let me know. Thanks!




More information about the vtkusers mailing list