How to keep it quads ?

[drmerman]

THAT, my friend, is the big question :) I think its really down to just methodically constructing your model, as opposed to creating something, and then trying to reduce it down to quads afterwards. My own method is to simply model a bit, check back over it, clean it up, reduce it down to as few polys as possible whilst retaining the quad structure, then model abit more, go back to check again...and repeat until you're done. Its one of the most time consuming things about modelling, and can take a while to fully learn, but in the end it can make rigging/rendering a whole lot easier.

Cheers
Dr Merman

[EvilArcana]

I'm currently working on a model that has some areas that are not quads...BUT...they are spots that come to points...whats the best way to convert those points into quads without messing up the rest of the model or havign to move all the way up your edge loop/ring chain to add another segment?

[safe05]

Can you post an image of the area?

[EvilArcana]

Sure, here ya go...this is the tail on my new spider lady...un subdivided.

Edit: A little more information since its not a full view...If you were to create a regular sphere (not geosphere) and pull the vertex at one of the poles into that position...you'd have what I have there...its basically the point of a sphere's pole...which is the primitive I started with.

[EvilArcana]

I also have an area that makes a triangle on the model half, but its a quad when the symmetry modifier is applied...since I'm working on half the model...is that acceptable?

Also, I noticed that Stahlberg preaches a lot about edge loops and quads, but his models have a lot of polygons that are more then 4 sides if you count the 5 sided ones where an edge loop comes to a halt at a larger 4 sided poly thus making it 5 sided because of that divided edge.

Like this:

[musashidan]

Have a read of THIS (http://forums.cgsociety.org/showthread.php?t=347843)

It's only a few threads below this one.

[mech7]

:argh: But what I tend to have is that when splitting polygons, is that I only move the tri? So is there a sneaky trick in the book that I don't know about

[EvilArcana]

Neither of those posts answered my question about how to bring an edge loop down to a fine point like in my first image post...

[safe05]

Basically, you want to have a quad represent the tip instead of a single point. You could try doing a chamfer on that point and then breaking up the top into quads. Generally speaking though, coming from a sphere with a pulled point from one of its poles is probably not the best way to model this shape since it becomes increasingly difficult to convert it to all quads. What I would suggest is you keep your sphere (or round body) and orient it so there is a quad where the tail meets the body. Then either (a) select the quad and extrude it a few times to get the shape you want, or (b) draw a spline that represents the "path" of the tail and extrude the face along it using "Extrude Along Spline."

As for Steven's modeling, tris/quads/ngons all have their purpose. tris will pinch the subdivided surface, ngons will flatten or dissolve the subdivided surface, and quads will keep everything flowing. If you don't have a "quad only" restriction imposed, tris and ngons can be very useful in getting subtle shapes in your surface without making your object too complex. For example, ngons work great on blending a crease into another area. (like the forearm picture above) However, if you're going to ZBrush or another package that needs quads, you have to do some preplanning of your model and some trickary to get everything flowing.

Hope this helps. :)

[McKertis]

Also, I noticed that Stahlberg preaches a lot about edge loops and quads
You noticed wrong. Steven Stahlberg always said he doesnt care about his models being all quads and all this quad hype.

[musashidan]

Neither of those posts answered my question about how to bring an edge loop down to a fine point like in my first image post...

The link was actually intended for the guy who started the thread...........

[kpi]

It would be good if somebody can make a script to select all the triangles in a mesh... because sometimes, in complex meshes, they are hard to find... and after meshsmooth (in max I mean) you get a mess...


IMHO,

kpi

[oDDity]

Waste of time. Don't worry about triangles, as long as they aren't causing any visible anomalies on the mesh. I'm not sure where this idea of 'all quads = perfect model' comes from. Triangles are very useful if you know how and where to use them.
Some guys talk about triangles like a cancer in their model.

[safe05]

For packages like ZBrush, you often need a model of all (or mostly) quads. This is probably due to their "pixol" methodology. But if you're staying in regular DCC packages, whatever works. Tris/Quads/NGons all have their uses. :)

[pauljs75]

I'm not sure where this idea of 'all quads = perfect model' comes from.

Ummm... Maybe they're a user of edge-loop based topology and want some kind of "nice" flow in any subsequent Catmull-Clark subdivision. Seems to make tweaking easier. But what do I know?

In the end it doesn't really matter what geometry is used if the model looks good when rendered and meets the parameters specified for the given application.

[CGTalk Moderation]

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.