I’ve not gone to school officially for CG/3D, and am self taught (I assume like the majority of us). Because of this, certain ‘industry standard techniques’ are probably not implemented…
One thing I’ve always wondered…
I know that when creating models for gaming, you want to remove all non-essential geometry…to save on both memory and UV space…however, that being said, how many of YOU actually go through every bit of mesh and remove all hidden faces?
As in this example here, would I go through and remove ever hidden face?
I will answer my own question…yes…yes I do. But a second part to my question is…is this standard…do all ‘professional’ modelers remove ALL hidden faces on ALL projects?
Yes they do - almost always. Especially for models which will be cloned hundreds of times. It’s not only that we (I am game dev) remove faces that are fully invisible, we make models with as few intersections as possibe. So if only part of the face would be invisible, we would cut the mesh so only this invisible part could be deleted. Of course everything has to stay within reason. Spending unreasonable amount of time on saving few polys is plain stupid.
On your model, you’re not even ready to remove hidden geometry. Are you doing a 1mm flyby of that house? Are there any dynamic lights you expect to approach that wall? No? Then all of those details should be texture, not mesh. The facing wall of the house should be three triangles, no more.
I’m not a game dev, but I got contracted once for some level design work, maybe 20 years ago? 25? I remember being so proud that I fit so many pipes into one scene with careful geometry. My goal, back then, was something like 500 tris max. Lotsa parallax. Mmm, pipes and crates. Such shit in hindsight.
If you’re not part of a team, though, you do what you can. You don’t have to do any more than that. You can’t, or else you’re not doing what you can. You say, I’ve got to optimize both AI and meshes. Realistically, I’m not going to get both done before I burn out. Which is more important? You deal with whichever is more important. Not burned out yet, despite what you anticipated? You deal with the next most important thing. Whatever that is.
On your model, you’re not even ready to remove hidden geometry. Are you doing a 1mm flyby of that house? Are there any dynamic lights you expect to approach that wall? No? Then all of those details should be texture, not mesh. The facing wall of the house should be three triangles, no more.
Thanks Bandages. No, this is low poly.
The whole house is 456 faces and 896 tris. That doesn’t seem bad for a house you might walk near if not up to in a game. I certainly wouldn’t make the inner details normal maps…wouldn’t that be more for an isometric mobile game or a house in the distance? I’m also following along with a tutorial for game modeling, so I assume the guy (Grant Abbit) knows what he’s doing.
Anything that you’re doing has to be done in an understanding of what your game is. Where is your camera going to be? That’s the most important question. Where are your dynamic lights? That’s the next most important question. You can’t make your models without even vague answers to those questions. What kind of game did Abbit say the tutorial was for?
Try it out. Bake to a really low poly and compare the look. From expected camera angles and distances. In engine. That’s what will tell you whether you should be using texture or mesh for your details.
After that, put together a scene, including some characters. You can’t judge vert counts in a vaccuum. It’s not about how many verts are in your house. It’s about how many verts are in the camera (well, camera plus). Does your prototype have good performance? With anticipated shaders, iIs it fill rate limited or transform limited? Who knows, you might not be using as many verts as you could get away with.
But vert counts are always a tradeoff. A vert saved on your house is a vert you can use on your character. 100 verts on 20 models are 2000 verts you can use on your character.
Thank you. I understand this. I know I’m running a basic tutorial here, but it’s more to see how Abbit does it…I do understand a bit about games. I know my initial question seemed very amateur, but I wanted to know how other working pros handle hidden faces. I haven’t worked as a modeler on a dev team. Thanks for the reply!
I know my initial question seemed very amateur, but I wanted to know how other working pros handle hidden faces. I haven’t worked as a modeler on a dev team. Thanks for the reply!