Modeling questions that bug me

A few questions I have regarding basic modelling.
I’m modeling a house for a game.

As a simple game object(for phone) how many faces is accepted? <1k 5k 10k !?

Subsurface modifier: When do you use it?
1- Start of the project ?
2- Not at all because of adding to much faces?

Quads are good. Is it a problem to have tris / n-gons / 6+ poles?

If I have a wall then add a plane. I model the plane in a window and move it up to the wall. Is it ok if it is not ‘connected’ to my wall? Or should all objects be 1 ?

Thanks

If you’re modelling for a game, then yes you should keep subdivisions to a minimum. Subdivs multiply verts/edges/faces which adds to the poly-count, and the aim in a game is usually to keep that as low as possible in order to utilise resources effectively.

For gaming purposes, I don’t think you should ever have nGons in your models. Games usually convert quads into triangles as well so for modelling in Blender, you should keep to triangles / quads and then convert all into triangles when you’re finished (as far as I know anyway).

Yes, you can have separate objects. You could join them all into one or even model objects as one object if you’re skilled enough, but more often than not it’s not usually necessary and just makes texturising a bit harder.

Good modelling means your model remains functional and ‘looks right’ with subdivision on or off. so it doesn’t matter when you apply it. You can turn it on at the start, half-way, or at the end of the process depending on your own preferences. You’ll just have to experiment and find out when your ideal implementation is. Personally, I like to turn it on at the start then turn it off (using the eye button in the modifier stack) when I need to make general shape changes without distractions, and then on again when I need it on.

Thanks allot Kurtis! Really helpful and fast

You say you are deploying to a mobile platform (phone)? In that case the subsurf question is simple: DO NOT USE IT! haha!

phones, especially ones with weak CPUs like any iPhone or iPad cannot handle very much in terms of graphics. The rule of thumb for polycount is more a question of how much will be visible at one time. For iPhone, shoot for 30-50k tris visible at once, for android, you might be able to push that up to 100k, and you need to seriously optimize your texture atlases and group static objects together to take advantage of draw-call batching. So really, when designing each object, use JUST enough geometry to get the silhouette correct, and paint in the rest of the detail in your texture.

What game engine are you using? That might help narrow down tips more, I’m quite familiar with unity3d, and might be able to add some further details if thats what your using =)

Thanks Dave that clears allot. I am not developing myself and there is no pre-made engine used. The game is pretty far up the pipe line but lacks resources. My goal is to learn blender, and so I asked some concept art so I could just fiddle around with it. I do want it to be great with a correct mesh and stuff like that.

I have looked at unity for private projects, but figured I needed models. And so I found blender and got sucked into this amazing world =)