Some beginner's question

urekmazino_

Suppose I'm modeling a house, any simple house.
I know that I want to do is to model that house, but the topology of it is what I don't know what I should do.
Should I always aim for quads?
When is ok to not have quads?
Is n-gon always a bad thing?
Should I model everything as a single object, many objects or many objects and join them at the end/as I go modeling?
When I apply subdivision surface?
Do I always apply subdivision surface?
Bevel a corner or subdivision surface and then fix that corner?
Why sometimes Edge Loops doesn't goes around the faces and how I fix it?
As a beginner, should I aim to model something first and then care about topology later on?
If so, when I start to care about topology, should I care about it as soon as I start modeling something?

Udjani

First you should block out the big shapes then start to apply details. Topology is something that should be be considered from the start, but don't worry too much about, that's something that you will learn with time and practice.

You don't nee to keep it as a single object, you can copy the reference as it is if want, but most of the time you  can simplify it and merge objects, like paneling areas. 

You should only ''apply the subdvision'' if you want to get more resolution or export to another program for baking or something, And in that case i think that in most of the programs you don't need to manually apply too, it will be done whe you export it. 

''Bevel a corner or subdivision surface and then fix that corner?''
I don't understand. 

The edge loops only works when you have a quand loop, if you have an ngon or a triangle in that loop it will not work correctly.

Those are very minor things and there are a lot more of them for you to learn, you maybe are already doing it, but if not, i reccoment you to chek this cahnnel (https://www.youtube.com/channel/UCSLLdTBwLMfTKWS56tOiQpw) and go through the basics tutorials. 

Burpee

urekmazino_ said:

Suppose I'm modeling a house, any simple house.
I know that I want to do is to model that house, but the topology of it is what I don't know what I should do.
Should I always aim for quads?
When is ok to not have quads?
Is n-gon always a bad thing?
Should I model everything as a single object, many objects or many objects and join them at the end/as I go modeling?
When I apply subdivision surface?
Do I always apply subdivision surface?
Bevel a corner or subdivision surface and then fix that corner?
Why sometimes Edge Loops doesn't goes around the faces and how I fix it?
As a beginner, should I aim to model something first and then care about topology later on?
If so, when I start to care about topology, should I care about it as soon as I start modeling something?

Hey ! Ujani already gave you some answers and here's mine to complete,

It heavily depend if you model for film or game, mine would be more oriented toward pre rendered film / cinematic
Should I always aim for quads?
Basically yes, but the real answer 'd be " if it bend / distord you should have proper quad / edge flow otherwise fuck it ! :d 
When is ok to not have quads? 
As I said above if it don't bend you don't really need to care about your topology
Is n-gon always a bad thing? 
I'd say yes because most of render engine don't like them, but if you use it only for baking purposing you can go crazy, also they don't subdived well.
When I apply subdivision surface?
In film most of the object are subdiv at render time, you should avoid doing " hard " subdiv on your mesh inside of Maya for exemple
Bevel a corner or subdivision surface and then fix that corner?
As I said in film, bevel your corner and they it'll be subdiv and smoothed during the render, if you mean " bevel it and subdivided it until it gets clean and smooth " that's not a good way to do it ( but again depend on what you do with it )
Why sometimes Edge Loops doesn't goes around the faces and how I fix it?
Edge flow ain't right ! 
As a beginner, should I aim to model something first and then care about topology later on? 
You should care about topology as soon as possible since it's the worst thing on earth, eat that frog now and you'll be fine later
If so, when I start to care about topology, should I care about it as soon as I start modeling something?
Maybe not at the beginning, for exemple character creature can be retopo after you've done a clean high rez mesh, but for exemple a car that'd be a pain to retopo afterward

Mark Dygert

1st off don't be afraid to experiment, try things and to find ways that don't work. Finding out that something doesn't work, is one step closer to finding a way that does work. It's only a failure if you give up.

Should I always aim for quads?

When it is for your benefit to do so, but don't be dogmatic about it and cause more problems by trying to fry your brain with a way to keep it all quads. Use ngons and triangles to terminate loops and keep your model workable. In the end render engines and realtime engines don't care about quads, they see triangles. So really, loops and rings are there for you to model triangles quickly and easily. 

When is ok to not have quads? 

This is very situational and there aren't any hard and fast rules other than "don't make your life harder than it needs to be". Through trial and error you'll learn where sticking to all quads will make messy nightmares and where triangles will cause headaches with edge selection, relax functions and baking problems.

Is n-gon always a bad thing? 

Hell no, they are incredibly helpful. Use whatever you need, when you need it. There are a lot of different ways to get to the end goal. Someone may have a better, quicker path and it might be good to make a note and try it out next time but that doesn't mean what you did was wrong.

Should I model everything as a single object, many objects or many objects and join them at the end/as I go modeling?

Break apart what you need when it makes sense to do so. When it comes to high poly modeling, you'll want to break apart a lot of things so they are easy to work on by themselves without having to worry about the rest of the model. 

A very common mistake that beginners tend to make when modeling high poly objects, is model everything in one object, like a clip and the main body of a weapon. Or a flip in a light switch . If the real world counterpart has separate pieces, it's probably going to be better to model it as separate.

For low poly modeling it makes sense to combine objects as elements of a single object, because more objects means more draw calls so combining saves processing power. However quite a few engines will combine assets automatically on import so it really depends, some engines you can feed them separate pieces and they handle it the best way possible.

When I apply subdivision surface?
Do I always apply subdivision surface?

When you want to smooth your model for rendering, but you don't want to work with all of those loops. You work on a lower poly version and run subdivide modifiers or functions just before you render.
In games you almost never smooth your final low poly model, but you might subdivide your high poly that you use to bake textures.
In film, TV, arch-vis you subdivide anytime polygons start to show and it could break the illusion. You're still thinking about how long it will take to render but you're not concerned on the same level as in games that are trying to run in realtime.

Bevel a corner or subdivision surface and then fix that corner?

Whatever makes sense for your workflow and the object you are modeling. If you are working on a subdivided model you might add extra edges to support the corner, even if you bevel it. If you bevel and it shades poorly and adding loops makes it loop better, add them. Just following a rule because someone else said "always do this..." will cause problems.

Why sometimes Edge Loops doesn't goes around the faces and how I fix it?

This sounds like a very specific problem but we're not looking at the example so it's hard to say how it should be handled.

As a beginner, should I aim to model something first and then care about topology later on? 

Think about it, try to make it good topology, but it shouldn't paralyze you. Keep moving forward and you'll find problems, solve them and those solutions will inform how you approach similar things in the future.

If so, when I start to care about topology, should I care about it as soon as I start modeling something?

You should always care about topology when it's important. If you're making static objects that will never animate, like rocks or debris, then it might be fine to make a triangulated nightmare because no one else will need to edit it. If you pass off a triangulated mesh to be rigged and animated, there is a really good chance it will come back to you.

So yea, mess around, make mistakes and learn.