View Full Version : Quads in archtecture
11-24-2003, 02:55 AM
Just a question. I know that when it comes to organic modeling, especially if the model is intended to go to sub-d's, one should try and use only quads throughout the model. I am wondering about architecture, though.. say large flat walls with i.e. windows, details and stuff in 'em.. is it ideal in any way at all? most walls will serve no purpsoe being converted to sub-d's and can exist happily as a polygon or 4. Here's my example:
Click for Picture (http://members.shaw.ca/zoidberg/cgtalk/door1.gif) (door1.gif)
This is a door for a building I'm working on right now.. the outside kind of border of the door can be simplified a lot more than that, I'm sure.. like I could make the polygons based on the pieces of wood that might have been used to make the door. I quickly hacked it in photoshop to show what I mean:
Click for Picture (http://members.shaw.ca/zoidberg/cgtalk/door2.gif) (door2.gif)
So, is there any reason to not do this? You see the lines I removed retain polygons which are quadrangular in shape, but have various connection vertices along their inside edges. Should I feel free to go ahead and make my life easy? :D Will I be ridiculed by my peers? ;)
11-25-2003, 06:04 PM
Honestly, even in sub-d, quads don't really matter much when you are talking about a planer surface. If you ever need to sub-d the thing, of course there will be problems.
There are a few things of note. Your specular highlights may have problems with ngons, tris, and poles. If you have broad highlights, just make sure to check your speculars before rendering.
Also, the major thing to think about, is if the building needs to ever come apart...violently. Then quads and uniformity can matter. Also if your building needs to distort, ie, bend, bulge, twist, etc, you will need to model with sub-d in mind.
Any type of dynamics simulation you may wish to do(ie, a ball bouncing of a wall, or a piece of cloth drapes blowing against a window) can be affected by lack of geometry. This can be fixed with increasing sub-frame sampling, but again, it can be problematic sometimes. Of course, this can be fixed by simply tesselating the wall/door/etc. that needs dynamic influence.
Photon-based radiosity and GI lighting can also be somewhat affected by your choice of geometry. In the latter case, the software ususally adds the detail it needs. In the former, there are some situations where you may need to increase the geometry in order to keep render times down. This will come into play if you use something like 3DS Max's Lighttracer, where the samples are filtered a certain way. Of course, this can be worked around with great ease. The software noadays is pretty smart.
Having said all that, none of those situations could really be seen as a factor of note. They all are only slight considerations as the software is usually intellegent enough to fix all that for you. I've never know a single person who has made a building using all quads unless the above mentioned situations were a factor. It's rather pointless, and somewhat of a waste of polys.
Sorry that was so long, but those are my thoughts on the matter.
11-25-2003, 06:30 PM
Cool, thanks 3dZ, that was helpful. I know when it comes to sub-d's, one thing matters to me at this stage- my primary renderer, mental ray, does not do non-quad-faced sub-d's. It simply won't render the object. Eeep!
11-25-2003, 06:36 PM
It certainly must be able to render tris though, right?
11-25-2003, 06:43 PM
Nope. Quads only. It gives a warning too.. something like "non-quad face in sub divisional object. ignoring."
11-25-2003, 06:52 PM
Ah! Ok. Makes sense now. It's probably trying to make your sub-ds into surfaces. If it uses higher-level math like renderman, it will be attempting to subdivide at the pixel level. If its that smart, I'm impressed. If not, I'm rather displeased. Catmull-clark's algorthym makes all quads. If from ngons and tris. If it's too stupid to use clark's method, I'm not sure I want anything to do with it. Of course, that is only if it's actually doing that math it's self. Strange.
01-16-2006, 06:00 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.