Topology problems....Help!

b_rizzae

So i've been trying to model a Nokia 3310. In my head i'm trying to follow a game asset workflow but i'm really not sure if i'm approaching this correctly. I'm using a sub-d workflow, keeping the base mesh as simple as possible and only adding geometry where absolutely possible (tell me if this is the wrong approach!). The bit i'm really struggling with is the holes and indentation for the earpiece/speaker. 

Please help on how best to approach this problem. I have added more loops but that started affecting the curve and shape of the phone.

Neox

Yeah to me you are working too lowpoly. What do you think you gain for modeling this lowres? So technically what you have, I would use for the base shape. Then subdivide once or twice, then punch holes and do the shape you marked

b_rizzae

Hi Neox, Thanks for the feedback. I was trying to make something with a good level of detail that was low poly enough to use in a game engine. It's not specifically for a game engine, just trying to practice standard workflow for game assets. I'm getting the feeling that this is the wrong approach tho. 

So basically i need to:

• add more geometry (destructively subdivide) to actually get the detail i need
• bake the normals from this mesh
• simplify/use the mesh i have and apply the normal map baked from the high poly model

pior

No, you got it wrong, probably because you are putting the cart before the wheels instead of taking the time to figure out the workflow (doing test bakes). Points 2 and 3 in your list do not make sense.

You need to build a high (using any workflow that works well for your specific case) ; and you need to build a low (also using any workflow that is suited to this case - meaning that you may reuse bits from your high base, or not). And then the surface information of the high gets baked to a texture according to the UVs of the low.

I would suggest to stop everything you're doing with that phone model and instead taking the time to understand the workflow with a simple set of cubes or cylinders (high and low).

sacboi

I'd also advise, avoiding manual editing by hand because there's tell tale signs of non uniform edge / vertex spacing resulting in an undulating surface over parts of that cage so instead learn to utilise mesh - loop tools / modifiers to introduce a predictable efficient process which will usually derive accuracy for a given (hard surface) object.

edit:

again, set aside time to browse through the last 10 or so pages of the subd thread I had previously linked, for further insight into various techniques generating similar content.  

wirrexx

So here's a couple of things to notice.

1. Always build the mesh as it's built in real life.

- If pieces are separate, they should be built separately.

2. Always begin with the main shape, Primary shape means the entire BIG shape of the phone..

- This way, if the shape subdvivide is good, you can always subdivide 1-2 times, to get the share amount of density you need for...

3... Secondary Shapes. Now you have more density, you can maybe add shapes that are smaller than the primary shape.

- Now you either use the existing topology or you add/cut in new edges for your shape.

b_rizzae

pior said:

I would suggest to stop everything you're doing with that phone model and instead taking the time to understand the workflow with a simple set of cubes or cylinders (high and low).

Thanks prio - the problem is i have spent countless hours trawling through the hundreds of threads and videos on how to model.

A: i'm not actually modelling anything
B: i have read so much conflicting information on workflows.(ie. make low poly first then go high poly/make high poly then go low poly/start off with just going straight with a clean high poly model.)

I'm struggling to find a clear, concise and succinct summary of an industry standard workflow for hard surface game asset modelling. 

I understand topology flow, i understand using edge loops and all the tools available to me to actually do the modelling, it's the underlying principle methods that i'm overwhelmed by. 

Neox

there is no one workflow to rule them all. there are times when starting with the low and doing the high after are fine, there are times where you do the highpoly first, then retopo to create the lowpoly. there are times you retopo just for the sake of a cleaner highpoly and you retopo again for the lowpoly. there is no ONE solution

b_rizzae

Thanks Neox. I guess it depends on how the prop/model will be used in game then. if it's a hero asset or just some set dressing w/e.

pior

Heya -

Well, you keep mentionning lowpoly and highpoly, hero assets vs set dressing, the conflicting information you got on that, and so on ... but in a way, none of that really matters until you get your hands dirty with actual baking. I am saying this because some of your statements show that you do not quite understand that aspect at this time, and this is the main reason why you are confused by what to model and how.

So as said my advice would be to stop worrying about how to model this or that, take some low and highs that you already have (either the phone low and high in whichever state they are currently, or some simple test cubes) and focus on practicing baking from high to low.

That's the mistake that so many people do : being led to believe that a process is linear ("do this, then do this, then do that" - because of poorly thought out video tutorials), tackling these things in that specific order, and then being completely stuck at the end because the last step is actually the most important and most complex to understand and should actually be figured out first.

Figure out your full tech stack *FIRST* on test models that don't take more than 5 minutes to create ; spend all the days needed to understand the tech aspects that your goal requires ; and *THEN*, apply all that to your work.

sacboi

b_rizzae said:

[...]
I understand topology flow, i understand using edge loops and all the tools available to me to actually do the modelling,

Well sorry, the condition of your model's geometry reads the complete opposite.

Now I'm in no way trying to be obtuse or argumentative but tying yourself up in knots over preceding pipeline tasks without first gaining at least some basic comprehension on laying out a valid mesh in a subd workflow makes no sense to me, at all and triangulation has yet to enter the picture...

There are many examples on these boards by seasoned hard surface artists that will underscore the sentiment so fundamentally creating whatever realistic biased object of interest is essentially reliant upon the quality of it's topology. 

b_rizzae

sacboi said:

Well sorry, the condition of your model's geometry reads the complete opposite.

Is this not good topology then?

Man, i thought it was pretty legit. even sized faces, good flow, localised where needed (around the button holes etc), only using quads.....

sacboi

Immediately noticeable here is the unnecessarily high polygon density / distribution by applying either a turbosmooth or subsurf (sub-diivsion level) prior too correcting the underlying base edge flow, causing distortion flowing around the boundary of this shape which merely compounds the issue.

edit:

I'm on my way out the door, at the minute however I'll post a lengthy reply describing simple concepts governing subd modeling hard surface content when time permits later this evening [UTC] alongside supporting workflow techniques I'd earlier indicated were already sticky'd and available too explore.