PDA

View Full Version : quad [face splitting] problem


j00ey
08-13-2003, 01:04 PM
hi all

i wonder if anyone can help me

i have a low-ish poly model which has a couple of five sided polys in it. i'm trying to stick to just quads and i can't for the life of me figure out how to split the polys without having to propogate the split to the edge of the model. all i want to do is split the five sided in to two 4-sided ones and then somehow terminate the splitting without generating any 3 or 5 sided ones.

can it be that this is a mathematical impossibility? i have drawn it out on paper and tried loads of differenty methods but to no avail.
ideas, anyone?

see attachment for puzzxle template
right hand pic is the propogate to the edge solution

:banghead:

j00ey

lostpencil
08-13-2003, 04:33 PM
Hey j00ey,

If you don't want to propagate to the edge you will be always stuck with a triangle.

The way I look at it is you have your pentagon and the simplest solution to get a quad is to split it like step 1. This leaves a triangle (step 2). Now in order to break down the triangle we have two options we can add an even number of vertices or an odd number of vertices. The simplest of the odd is one vertex (step 3). The simplest of the even is two vertices (step 4). All other number of vertices will ultimately reduce to these two. Therefore you will always end up with at least one triangle.

http://www.lostpencil.com/temp/tri.jpg

Not a mathematical proof per se, but I think it holds.

j00ey
08-13-2003, 04:42 PM
thanks pencil. i'm beginning to think you're right.

so... if that's the case what do YOU do in such a case? live with the pentagon? it's no great tragedy, i was just trying to stick to quads and keep to the minimum also, i.e. not propogate a split all the way and end up with unnecessary detail where it's not needed.

just in case you have any ideas about this ... another question to which no-one can give me a satisfactory answer is this : can mankind melt stone? if so in what would it be melted?

sorry mods for veering off topic, just sprung [back] to mind

j00ey

lostpencil
08-13-2003, 04:52 PM
What I do in such a case depends a lot on the place where it occurs and the requirements of the model. If I don't plan right and get a triangle where I don't want it then I often just add extra geometry and propagate the split - because more often than not the model isn't a 'real time' model and I don't need to worry about poly counts.

CGTalk Moderation
01-15-2006, 08:00 PM
This thread has been automatically closed as it remained inactive for 12 months. If you wish to continue the discussion, please create a new thread in the appropriate forum.