I am modelling this cornice bracket for a modular building, and I don’t know how to cap the side without ending up with a huge n-gon or a multitude of long tris. I would like to do this with quads only…
Well, come to think of it… It will end up in UE4, so yes it will definitely be triangles at the end of the day. It’s just that all the material I am using to learn modeling hammer the holy dogma of ‘quads only, tris are bad, m’kay?’. I just thought I’d save myself some trouble when entering the UV mapping phase, which I’m dreading very much.
Are you saying it’s actually okay to leave n-gons and let the game engine handle the triangulation at the time of .fbx import…? (For hard surface meshes obviously…)
There are very little resources available on modeling for game with Blender, almost everything is oriented toward other disciplines. I wish it wasn’t so…
I have little to nothing experience with game engines except the lots of reads that triangles are even hardcoded into any decent video card now a days. I hope ngon wont do any harm or cause complications during unwrap but if there is you will see this right here in blender.
You can after triangulate mesh Ctrl-T and check how it feels, then export. Leave some backup before triangulate.
It is important to triangulate manually if you do some record low poly object - automated process left alone might change shape in not desired ways. For bigger things it’s not so critical. But it’s triangles in the end. It’s just that modeling is easier with quads and ngons.
When it’s for a game engine, it will not matter much if you simply convert your planar ngon into triangles, as eppo said, in a game everything will end into triangles.
The advantage of having quads is that a lot of Blender tools are designed to work with quads or at least give their best results with them.
Additionally, some game engine keep the shading you had with a quad based mesh, even when internally converted to triangles, it can matter.
Depending on the game engine. N-gon just means N-sided polygon. Where N = the number of sides (remember Algebra class? Haha!). Triangle = 3 sided polygon, “quad” = 4 sided polygon, or quadrangle, etc etc. It is true that all video cards convert all topology to triangles at runtime. However, not all game engine editors know how to handle N-gons, though all the ones I’ve worked with do. The method I would use is the one shown by eppo, just because its a bit faster to carry out, and it makes sure the edges of the profile stay nice and crisp no matter what happens to the large N-gon in the middle. Just remember that those N-gons will still be split into many triangles at runtime, and not always in the most efficient way. So it can add significantly to your runtime tris count.
Also, there is little difference between modeling for a game in blender, and modeling for a game in any other software. Keep the polycount as low as you can while keeping your chosen level of detail (modern equipment can handle 2-5 million tris on screen without noticeable frame rate drop, so its not as important as it used to be, but little things do add up). Use a texture atlas for as many individual items as you can apply the same material to. And simplify the armature of your character rigs as much as you can get away with while still keeping the quality of animation you want. The less the CPU and GPU have to do for each object in the scene, the better off you will be.
Yeah, I generally have no trouble following game modeling tutorial with Blender, although there are still tools in 3DS Max that are lacking from Blender. Bevel doesn’t work exactly like Chamfer in Max, align tools are easier to use, there’s no Relax UV feature in Blender that I know of, no Quick Planar either, etc.
That’s about it though, I really hate having to do everything with the mouse in Max or Maya, I know just about all shortcuts in Blender and I made some more (I got a Logitech G15).
I guess specifically what is lacking in the material available to learn modeling for game are clear principles on topology like the ones you can find for “normal” modeling. I have been trying to study actual models from AAA games but as they are extracted from their engine directly they’re already triangulated so there’s no telling what their original topology was.
If the relax UV thing is similar to the minimize stretch feature in blender, then there is! Try it out next time you unwrap a face. Select the face island and go to the UV menu, hit minimize stretch and roll your mouse wheel.
Anyway, my point was more about the techniques involved rather than the tools used in the workflow. The only thing I do differently when modeling a character for a game instead of animation, is avoid subdivision surface modifier, and opt for adding extra loops + the smooth tool where needed to get my shapes without excessive extra geometry.
Yes, topology is very important, and very rough to get the hang of. Dont be afraid to model in a method your comfortable with, and cut into the model later to rebuild certain loops to flow the right way. Eventually you’ll learn from all your “surgery” on otherwise finished models, and start working the correct loops into your design from the start. Also remember to consider what the models final application will be. You don’t need to spend 5 hours shaping the perfect hips and butt, if the character is always gonna be in a full Victorian gown, etc.