Thread by @delaneykingrox

Okay lemme just kill some myths about #gamedev polycounts. (My upcoming book is packed full of this stuff BTW).

A polygon is not a useful measure.

I repeat, the polygon is NOT a useful measure.

A quad is actually two triangles with a hidden edge. You can display this hidden edge in all 3D aps. It usually shows as a dotted line.

Your 3D card sees your model as a bunch of triangles, not quads.

We use quads when modelling because of two reasons. They subdivide evenly, and they make selecting easier (loops and rings)

There is a myth that students are fed: Never model with triangles. Always use quads.

This is, I am sorry to say, complete bullshit.

Learn to model in quads, because of the reasons above, but understand that before your model goes into your renderer, that you can use triangles. In fact, in very low polygon situations, you really have to to get control over deformations.

In maya, you can find the "display triangles" setting in the shape node of the mesh.

The hidden edge is usually calculated between the furthest apart verts in the quad. If you picture the hidden edge as a fold in the paper, that dramatically changes how it folds.

But connecting or cutting across the quad, you force the program to fold the way you want it to.

When we say "convert to triangles" what is actually happening is all the hidden edges are made into explicit edges like this. Only you don't control how that happens. And so if you want control, you have to do this part manually.

So, to recap... "polys" means jack shit- quads are sneaky triangles and triangles can be your friend when you know how to use them.

Stay tuned for my book folks ;)

And there I will explain why triangle count is also not the best measure of system resources either!

;)

So because I have other commitments, I am squeezing writing the book in between things rather than setting a fixed release date and then dying from overload.

I have a lot of material from over the years and I have collated this, and roughed in the book on paper...

...what I have to do now is turn all the notes and material into clear, consistent diagrams that print well.

This will be a handbook style book with the focus being on communicating principles using clear visual diagrams first, text second and anecdotes and tips in text boxes.

And it is looking like several volumes of material, so I am breaking it up so the first book is a real solid foundation for generalists, then I go into more specific areas in later volumes.

I will also cross over the material into other disciplines, such as level design, motion capture acting and game design so game artists can learn how to collaborate to make their stuff really useful.

Aaaanyway, back to how triangles can really help you make great game assets take a look at how I use triangles to up the roundness of a convex silhouette.

You actually can get away with far less geo on most game models than people tend to think is needed. Normal maps do a lot of heavy lifting on non-VR games.

(VR has binocular vision so the illusion doesn't work anywhere near as well, and so you need lots of triangles or make use of parralax occlusion mapping to sell the illusion to our clever brains.

"Only model in quads" gives you nightmares like this clock asset. The flat surfaces are subdivided like crazy.

And if you learned something from my tutes and wanna see more, toss me a coffee or sixty nine.

Sixty nine coffees... not... well, maybe but... uhm. Maybe just the tip. I mean... shit! Uh... no... no shit... *dies*

ko-fi.com/dellak

I am gonna use some of the money from the book to get a triangle and verts tattooed on my hand. :)

And a fly swat for dudes who mansplain 3d to me.