Theoretical SUB-D

#621

a while ago, someone on 3Dbuzz asked for help on modeling a perforated sheet. i remembered the technique pointed out here by celticdog for creating a hole in a quad, and build on that using array. here’s the 3Dbuzz thread, showing what i came up with.

just wanted to give a heads up, credit where credit is due, you know… and it might be a nice addition to this wonderful thread.

#622

Are all points on a poly in a technically poly based model supposed to be coplanar with each for each poly they make up?

#623

No. The interior edge that divides every quad into two triangles will add extra curvature to the model. If surface curvature is important in my model, for example I’m modeling lower-res for games, then I pay attention to those mostly-hidden interior edges, turning them as needed.

#624

low poly modeling is generally the only time you’ll have to worry about that too

#625

yup

#626

posm,

But there shouldn’t be any hidden edges, so those two triangles making up that quad you were t alking about should actually just be two triangles with their edges visible if it’s a true poly object right?
In that case the verts would still be coplanar with the poly’s they are defining. It would just be two triangles, not a four sided poly with an invisible edge

#627

We’re talking about 3ds max, right? Every model in 3ds max is made of triangles, Editable Poly, NURBS, Patches, whatever. As I understand it the quad/edgeloop thing is just an interface on top of triangles.

The four verts of a quad can be coplanar but don’t have to be.

Here’s a pic that might help explain.

The verts for both are in the same positions, but the interior edge is flipped. No smoothing, but it is still a quad.

Does this make sense?

This link might help. The info is a bit old, needs fleshing out. Maybe someday.
[/edit]

#628

posm,

Oh yeah I already know what you are talking about I just mean like in a real poly based system do are all verts in a poly coplanar. Like in a true poly system there are only poly faces, like topologically there are just visible edges and the vertices that make them up and that’s it. I’m just trying to figure out if in a true poly system all verts in each poly face are coplanar because it seems like they would have to be, otherwise it’s seems like it’s more of a mimicking type of system.
It seems like a true poly modelling system wouldn’t have anything but coplanar faces, so like if you did have a quad polygon in Max with vertices that are skewed so they didn’t form a flat face (like in your exmaple); instead of it looking like a contorted polygon, wherever a bend occurred across the poly face across hidden edge, it would actuallly require a visible true edge to be there for it to bend like that otherwise.

#629

Can you point me to this “true poly” system? I’d like to see it.

As far as I can tell, only very exotic rare hardware handles pure polygons. All the rest use just triangles. All games use triangles. All modeling programs convert to triangles at some point, to display on your screen.

Sounds like kind of an esoteric discussion to me.

There are other modeling systems that don’t use polygons, voxels, wavlets, etc., but mostly they convert to triangles to display. Unless you have a bizarre unusual 3d monitor display.

Cool topic.

#630

Yeah, well, my background is from mathematics and physics so I guess I’m trying to understand something about 3d modelling that is more abstract or pure than how it’s practically implimented in 3D programs for making stuff (like in Max). I guess that’s where/how I’m getting mixed up about things.

#631

From a practical perspective, coplanar quads would be kinda awkward to use yes? Kind of limiting if I am understanding things right. You’d need more polys to define the shapes you want than if they can be non-coplanar. Or maybe I’m just too drunk to understand properly…

#632

yeah, it would be impossible to do a twisted loop of true flat quad polys, right? <edit>uh oh, this is possible, when the quads are trapezoid or parallellogram …</edit>

I just try to keep them as flat as possible, otherwise add detail or if i see a diagonal of the quad goes the wrong way, i’d make it go the other way with edit triangulation.

How important is the triangulation anyhow?

it gives a nice feel to the mesh if all triangulated quads run the same way, so the hidden edges form loops too. and i guess it’s faster for a videocard to display them as it can be computed as tristrips or wafers and all that low level technical 3d stuff.

#633

The triangulation isn’t important at all if you put a meshsmooth on it. It only uses the visible edges for the tesselation. How important the triangulation is for the gfx card I have no idea

#634

This is an interesting mathematical question: Does a polygon (any number of sides) cease to be a true polygon once one or more of it’s edges are no longer coplanar with the rest of it’s edges?

edit: as a side note I’ve been refering to more than just quad poly’s. If you bent a quad poly face at it’s invisible edge the idea is that to keep the mesh made of true poly’s the bend woudn’t really be able to occur unless that hiddenedge was visible i.e. the quad poly would either change into to two triangular polys or upon skewing one or more vertices of the quad poly the program would automatically make the appropriate edges visible to keep the mesh composed of actual polygonal faces.

See the idea is that a true polygon mesh has no hidden edges to manipulate because there shouldn’t be hidden edges in the first place in a true polygonal mesh. Yes the shape may get converted into triangles in the end for rendering but that’s not what I’m talking about.

#635

real edit:

Sorry if this is off topic it’s just that I think the whole discussion here about edge loops and stuff is topologically cool (with subd modelling) and I thought maybe this thread would be the most pertinent for the question I’ve been having, like someone else is most likely to know here.

#636

Having the program automatically create edges to keep the polygon coplanar would be very bad from a workflow point of view. You need to be able to “bend” polys to keep modeling without too much hassle

#637

we’re such nerds …

#638

Just want to quickly thank everyone who’s contributed to this thread especially 3DZ, Gnarly and Urg.

I’ve learned so much more on this thread than with all the Max manuals put together.

Thanks and keep up the good work!!!

#639

Awwwwwwwwgeee, here I’ve been outta touch dealing with no-comp issues, and somebody still remembers me. I feel all smart!!

I have a few more questions and ideas to toss around… I’m gonna take a bit to review the thread and then probly bring that stuff in, if it ain’t too late… bah, I might have to wait till after siggraph though, mighty busy this week.

#640

I did read all this threat and wow, great job guys. I’ve learned a lot of tecnical things :buttrock:
Hope you arent tired talking about subdiv

So… I’m starting to make my models prepared to animation and I’m trying to create edges loops for it. But in some cases, I get confused… when making the model, I think there will be an edge loop and doesn’t, visa-versa. Then, I’m starting to study more its principies to make wanted results in one shot. Hope you got it, so moving on…

Heres an model example:
http://www.redpixel.hpg.com.br/edgeloop01.jpg

With this structure, I have an edge loop and here it is:
http://www.redpixel.hpg.com.br/edgeloop02.jpg

BUT, why the edge loop is breaked when I create this under poligon? It shouldnt get going?
http://www.redpixel.hpg.com.br/edgeloop03.jpg

Fine, weld the vertex and back with the edge loop:
http://www.redpixel.hpg.com.br/edgeloop04.jpg

And here’s another way that edges loop continues but dont know why:
http://www.redpixel.hpg.com.br/edgeloop05.jpg

So the point is, what principies EDGE LOOP folow? How i’m going to know theres going to be an complete edge loop?