Lost in the Weeds...Hi Poly to Lo Poly - ZBrush to Maya Workflow

TROGLODYTE

Hi guys,

I think this may be my first post to Polycount ever.  I'm trying to bridge the gap between 2D concept art into 3D so I'm struggling with the learning curve as I graduate into texturing.
I have learned tons about modeling hi detail in ZBrush.  I learned how to retopologize in ZBrush into lo poly models but then realized I need to bring it into something like Maya (full spectrum 3D program) to create normals maps, texture maps, create actual renders, etc...so I learned retopology in Maya as well.  But I'm stuck now....
The characters and hard surface models I tend to make have items coming off them that either present a nightmare for low poly and unwrapping like pipe frame off a submarine or the protruding belt loops off a character's trousers.  Do I try to work belt loops and shoelaces into a low poly model?  Don't such fine protrusions completely go AGAINST the whole point of creating a lo poly model in the first place?

*** Below I have placed a link to a video I just uploaded to YouTube showing my hi poly submarine model with steel frames and antennae sticking out so you guys have a better idea of what I intend.  Oh...and these are for game assets.

https://youtu.be/0cjslH91zoQ

Deqa

I guess you're asking if you should make the belt loops and really tiny details super high fidelity in the low poly? I'm assuming you would just make them low fidelity, as much as you can, or try some type of auto-retopo tools for super complex belt loops in a low poly (maybe zRemesher) and try to use auto-unwrapping for those complex belt loop details in your low poly.

TROGLODYTE

Well, let me be more specific via these two attachments...

Prime8

The level of detail depends on how you want to use the low poly model.
Are you planning to make a game where belt loops play a vital role and take up quite a bit of screen space? Than you probably want to model them. If your character is seen only from far away, you would probably not model it at all and just add the whole belt into the texture.

On the submarine, just model the safety frame, don't "bridge" that gap. 

Ghogiel

It's a non deforming mesh, in this mesh you won't lose anything or cause any issues what so ever making the rails and those antenna's as separate pieces of low poly non connected to the main hull geo (they aren't even connected on the high poly, and are just rammed into the main hull there too)

They can all be attached together into a single low poly object and all fit on the same UV as well even if they are just intersecting pieces of separate geo.

Alex_J

As this is your first forray into 3d, I'd recommend putting this high poly model aside for the time being, and learn how to do the opposite workflow first. That is, build a clean, low poly mesh with block modeling techniques in Maya. Then UV layout that and texture it. Very managaeable workflow that will get you comfy with the basics.

Next, add on more step, but it's quite complex and will be challenging. Blockout a new model, but this time use subD techniques to create a hi-poly model, do your UV's, and then your baking. Finally, end with texturing.

After doing two models that way, beginning in zbrush and working the bass-ackwards way will make a lot more sense. You won't be swimming upstream so much.

Although hi-to-low is what you see so much of on youtube, I do not believe it is the most commomn production workflow. Certainly not for me anyway. It is a powerful workflow for concepting and has it's merits, but it is probably the hardest workflow you could start with because it requires really solid understanding of 3d fundamentals and program familiarity.

Plenty of tutorials out there to learn this workflow. I got started with Justin Marshalls "intro to maya" courses on pluralsight, which are pretty decent to learn subD modeling.

TROGLODYTE

Alex.....THANK YOU.  I have been struggling for over a year now because of exactly the amount of tutorials out there that have made me think high to low is just 'the way it is'.  Your explanation was clear and I see now exactly what you mean!  Also, I always had this nagging question that NO ONE on the web answers...and that is "is it normal for pieces of a model to be 'separate' from each other?"  You answered that as well.  Since I have never worked in a pipeline, I am still blind as to how the 'combined' sections of the final unified model get inserted into an engine.  There is so much on the web about the obvious 'meat' of 3D process but so little on the web about the simpler application logistics that people like me need to know to see the bigger picture and therefore know how the smaller pieces need to be managed.

In fact, I had dismissed even the notion that such a reverse workflow was legit.  You see, this is a good case for the argument of going to art school.  You always hear "Oh you can learn everything you need on the web".  The problem is, one has no way of knowing what one needs to filter out the 'less qualified' (for lack of a better term) of the massive sea of information on the web.

Again, thank you!


Jonathan Acosta
https://www.artstation.com/jacosta

Alex_J

I was having a lot of trouble making sense from tutorials for awhile as well. But I tend to be a pretty irreverent jackass, so I thought I could do things better on my own. Turns out, 3d isn't so complicated if you just learn it the more old fashioned way -- that is starting with block modeling and saving baking and all that until you've got the basics down.

@pior also gave me some good advice early on which saved me a lot of headache and helped me to realize that beginning in zbrush and going backwards isn't the only way -- and many times not the best way -- to work.

So yeah, just keep on keeping on and don't be afraid to slow and try doing things totally opposiite if you feel like you're hhitting your head against a brick wall. About having objects be separate or combined, just totally depends on your game or render goal. Easiest way to learn is just make a game -- or like a simple demo with the basic components of a game. There is no right way to do things, but by challenging yourself this way you'll naturally learn what most of the common practices are.

TROGLODYTE

Cool!

Can I ask you another related question?

I am doing character modeling with the mindset of maximizing their use for multiple future uses, from static models for concept paintings to possible use for future deformations for posing or possibly even rigging for animation (as a higher goal), just to appeal to companies hiring concept artists.  Currently building sitting male pilot just to design seat, cockpits and ships around.

I have found that based on my preferred workflow, it does not help me to build lo poly models where new edge loops added at the feet (for example) travel all the way up the trunk and to the head, thereby unnecessarily complicating facial edge flow in face.  Same issue goes from left side to right side.  So I'm thinking of pursuing low poly foundations where head and limbs have self contained edge loops and edge flow that end where they connect to the trunk.  I figure this might hamper seamless texturing at the joints a bit but it will help me immensely in detaching and replacing entire limbs and heads on-the-fly, plus speed up working on limbs separately rather than on entire new bodies every time.

My question is:
Would you say that approaching character modeling on a 6-detachable-part approach (2 arms, 2 legs, 1 trunk, 1 head) to later be unified in an engine or in film animation a bad workflow mindset?  I'm trying to save a huge chunk of time by working from a modular mindset to speed up my process.  Would this be considered a taboo approach?

Alex_J

This should answer most your question:

http://wiki.polycount.com/wiki/SkankerzeroModularCharacterSystem

My advice, just make regular characters for now and leave the complicated stuff alone. Nothing you make now will be something you want to reuse a year from now, so better to learn the basics without added complexity.

TROGLODYTE

Wow thank you.