PDA

View Full Version : Methods for 'fixing' non-planar polygons


particledecay
03-25-2007, 08:26 PM
Hi all,

I've recently got myself into a bit of a fiddle with a low poly model I'm making for the Torque Game Engine (dts format). Whilst editing a curved surface my polygons have become non-planar. As the model needs to be tripled anyway this didn't bother me until I imported the model and noticed cellulite on the surface shadows and reflections. The texture has smoothing applied and I've rebuilt parts of the model to get the polygon flow as complementary to the form as I can manage. The flaws appear in Layout also which points to my model being flawed. It seems that the most likely culprit is variation in normals of the tris across the 3D curve. It is causing some unsightly shadows and bumpy reflections and I'd like to improve my model so that these are minimized.

Could you share any techniques you've found useful for low poly box modeling without subdivision to 'fix' non-planar polygons once they get out of whack? So far I've tried:

the "flatten" tool which seems to work although it creates more non-planar polys adjacent to the selection.
manually adjusting the points. This seems very inefficient and testing the result of each change in the engine is a very time consuming multi-step process.
adding polygons. This has worked but I fear continuing with this solution will raise my polygon count to an undesirably high number for a game engine.


Here's the model in question (very much a WIP):
http://i48.photobucket.com/albums/f224/gogo_racer_monkey/GG/sportCarDetail.jpg

particledecay
03-26-2007, 12:01 AM
Here's the original before I started messing around, non-planar polygons and all.

http://i48.photobucket.com/albums/f224/gogo_racer_monkey/GG/sportsCarDetailQuad.jpg

Surrealist.
03-26-2007, 12:05 AM
You should not be getting nonplanars from tris. But have you checked your statistics to see if for some reason there are duplicate points that make your "tris" quads?

If they are tris, I have had polygons go weird before. I solve this by just rebuilding them. You could also have some kind of smoothing error for some other reason. Either case, try rebuilding them from scratch one at a time and see if it goes away. Also sometimes point data can just get messed up and I find just deleting the poly is the only answer. Sometimes even cutting and pasting.

If this does not work, go into a background layer and start building these tris in the exact place (with new points) and see if the problem persists. If not, build these up from scratch and try cutting and pasting into the layer and patching it in.