A little help from the experienced! Improving workflow for 3D characters

miketzet

Hello guys,

I am a 3D character artist student in my first year of studies. As I don't have so much experience in creating characters--I have created 1 full creature, posed and rendered in marmoset and a bust of the same, also rendered in marmoset. I am having a very hard time trying to orient my workflow. At this moment, I am creating a female warrior that I started sculpting in zbrush and she is equipped with all sorts of things (armour, leather straps, trousers, boots, a belt, fabrics, that count down from her waist, etc. etc.). In zbrush I sculpted the base mesh body and with extracts and using marvelous, I created clothes and armor. After that, I retopologised only the bust of the girl, regarding the rest I zremeshed and made them into low-poly objects ready for uv unwrapping. I textured, gave pose through zbrush, maintained all the objects separated and....its's actually a mess. I am trying so hard to organize everything and have total control over my character and I'm getting lost. My goal is to create a character that is not presented as a soulless tpose model, but as a model, ready to pose and be "photographed".

I know that everyone has their workflow and it is something that depends to their way of thinking and organising, but I really wanted to know what is it would be a good way to orient my workflow so that I can get better and better.

Thank you in advance

pior

"At this moment, I am creating a female warrior that I started sculpting in zbrush and she is equipped with all sorts of things (armour, leather straps, trousers, boots, a belt, fabrics, that count down from her waist, etc. etc.)."

Think outside the box and work the other way around : Once your concept sheet is in a solid enough state (capturing most of what you want out of this character/project), start by making a simple, lightweight body model (or even just a rough blockout) and begin working on its skeleton structure right away. Then check that it can hit all the lively poses that you want. You will be effectively previewing your final end goal ... just a few days into the project. That's extremely valuable.

Once that is done and firmly established, then you can start thinking about details, sculpting, and so on.

99% of CG character artists don't do that (because they think that making a skeleton and doing some skin weighting is so boring, and playing around in Zeebrush is so fun), and are fine with slaving away at a character for months. Tackling things differently will allow you to be much faster, and will allow for much easier iterations.

throttlekitty

You gave a list of your major steps, but your complaint about a soulless model has nothing to do with workflow or organization. Where do you feel you went wrong there? I guess my question is what is it about the current organization of your character and subobjects that's getting in your way?

miketzet

throttlekitty said:

You gave a list of your major steps, but your complaint about a soulless model has nothing to do with workflow or organization. Where do you feel you went wrong there? I guess my question is what is it about the current organization of your character and subobjects that's getting in your way?

The most important thing that I think gets in my way is that I have high expectations of my work. That's not bad, sometimes, but I just entered a field in which I haven't experienced and tried many things yet. I am used to think as a sculptor (I was in fine arts) creating things with clay. The thing is that technical errors come on my way in the process. At first I am thinking "keep your objects separated, maintain control" and then comes the "how are you supposed to pose this?". It's been an almost 3 months struggle with this particular project. My desire would be to rig this character and play with poses. But how to rig a character that has separate objects? (that was just rhitorical).

carvuliero

You could use zbrush simple transpose or zsphere rig

If you know how to rig [joint rig in maya for example ]you can use that if not can use one of many autorig scripts

After that bind the body correct weights if you have only one pose , just for that pose and either parent constrain accessories that wont bend or wrap deform the one that will bend

You could also check mixamo not sure if its free now they keep changing it

You could use zbrush mannequins to find suitable pose 

Alex_J

Not that Pior needs a songbird, but I really have to second their advice to consider working on your game res model first, and doing the detail sculpting second. Especially for a beginner, going from hi to lo can really be tough and confusing. I really believe its a workflow best suited for those with enough experience to really understand all the in's and out's of their 3d software. 

As a beginner I was having a lot of trouble doing the hi-to-lo workflow. Finally I said enough, it shouldn't be this difficult. So I tried things the opposite way, and it really worked for me. Now I make a character that looks better than some of my earlier characters in literally 1/10th of the time. And it's a more enjoyable way for me to work as I don't need to do any retopo. 

So, if you are having too much trouble with your current workflow, run some experiments with alternatives others have tried, and of course don't be afraid to try anything out from your own imagination as well. 

miketzet

What I don't understand is do I need to keep doing what I am doing, that is keeping my model into submeshes or do I have to retopologise my model with the rest of the equipment? After the main high Res model is created, what do I do next?

I am sorry if I sound like desperate, but I have so many questions that I would like someone to answer me. It's good to hear from someone experienced what works and what not, in order to learn it and adapt it later to your workflow, but the most important is to have a good start.

carvuliero

That will depend on what your end goal is .Do you just  want a beauty render for your portfolio not using other programs but zbrush or you want game ready model or something else

miketzet

carvuliero said:

That will depend on what your end goal is .Do you just  want a beauty render for your portfolio not using other programs but zbrush or you want game ready model or something else

I tend to create game ready models and importing them in marmoset, which I get pretty good results. But if I wanted a beauty render, what is the difference?

carvuliero

If its game model then yes you have to do the whole pipeline retopo uv texture rigging posing and what not if you want something fast and dirty you could pose in zbrush using transpose master render bunch of different materials comp them in photoshop add some photoshop magic and paint over and you have beauty render - illustration

miketzet

carvuliero said:

If its game model then yes you have to do the whole pipeline retopo uv texture rigging posing and what not if you want something fast and dirty you could pose in zbrush using transpose master render bunch of different materials comp them in photoshop add some photoshop magic and paint over and you have beauty render - illustration

So, then, a more general question. You have to be creative and free avoiding being slave to the programs, but, in order to "work", you have to cope with the way the programs work. Can the "slavery" be inevitable in any way?

Alex_J

Sounds like you need to define what your goal is very specifically. 

Do you want to make a model for games? For movies? Just to enjoy on artstation?

Once you know that, then you can find plenty of tutorials and articles to take you step by step through the process.

carvuliero

What do you mean been slave to a program , programs are just tools you chose which program to use your life doesnt depend on using specific program or workflow for example sculpting itself you can done zbrush mudbox blender 3d coat even maya and probably bunch other programs same thing is true for the rest of steps in the pipeline if you dont like how certain program work just pick one that work for the way you want

throttlekitty

miketzet said:

throttlekitty said:

You gave a list of your major steps, but your complaint about a soulless model has nothing to do with workflow or organization. Where do you feel you went wrong there? I guess my question is what is it about the current organization of your character and subobjects that's getting in your way?

The most important thing that I think gets in my way is that I have high expectations of my work. That's not bad, sometimes, but I just entered a field in which I haven't experienced and tried many things yet. I am used to think as a sculptor (I was in fine arts) creating things with clay. The thing is that technical errors come on my way in the process. At first I am thinking "keep your objects separated, maintain control" and then comes the "how are you supposed to pose this?". It's been an almost 3 months struggle with this particular project. My desire would be to rig this character and play with poses. But how to rig a character that has separate objects? (that was just rhitorical).

That and the other replies explains your situation better, cool. Yeah, there's a technical aspect to digital/3d that's simply unavoidable, but it gets better, and will continue to do so. But we still need to get through the learning and experience parts.

I'm not a big ZBrush user, but I know there's some nifty pose tool that you can use like a simple(?) skeleton to pose a character on the quick. But that's about all i know on the topic, it seemed like the easiest way to get to a posed character while avoiding "doing it for real". The other option would be to jump into another program, create a skeleton, or use an autorigger to assist in that step. Weight painting can be copied to other objects, which is a huge help

Specifically to having multiple objects on a posed character, it's totally doable, but not always the norm for realtime games. Unless there's lots of character customization/part swapping going on, clothing is typically a single thing, or broken down into something like Boots, Pants, Shirt. Some dude with suspenders and an armband might exist as separate things in the ZB sculpt, but the low poly game model would have them modeled into the shirt, or even baked on without special geometry if they're flat enough. Another common one is that a long shirt sleeve ends and the arm begins from there, rather than having a full body under the clothing. Sorry this doesn't explicitly answer your question, I'm just rambling out some things to think about without seeing what you're working on or knowing what your end goal and quality are exactly.

kanga

What pior had to say about starting at the end is really useful, and would be a great time saver. Apart from that, the way I deal with a complex model is to break it up into as many pats as I need for sculpting. The model below is made up of 16 subtools.

Im only interested in game models so I have to make a low poly version. With texturing in mind I group parts together so they will bake properly. There is no real reason to do this because the lowpoly could easily be one mesh. I just prefer to separate things for more control, and a better end result (like the skin so if I need an sss shader I can do that more easily plus painting the body diffuse in ZB is quicker for me). Breaking the figure into parts making sure none are juxtaposed for baking I group the subttols into a few groups. Here I used 3 groups. I export the groups using fbx and naming conventions for substance painter. Then I copy and decimate the groups and send them to a 3d app for retopo/creating low poly cages. I create the lowpolys and uv them, then export them with the same naming convention system. At this stage I'm not interested in posing the character yet. However if you follow pior's advice stuff like weapons and baggage or any character assets wont get in the way of the end animations.

I then import the low models into substance painter and the groups are identified as texture sets.

The  hipoly model body already has a diffuse so that bakes fine. The hard surface groups have been poly painted in ZB with different 'clown' colors so they can be isolated for making textures. On the left is the assembled model on the right are it's three parts. Once I have extracted the maps I will use the UE4 skeleton by importing it into whatever 3d app I'm using, then skin the parts to it staking the skin modifies for the groups onto the skeleton. Export the rigged/textured model to UE4 and have some fun. Using the game skeleton lets me use the animation sets from the engine. That is how I take a 16 part model and merge it into one asset.

Hope that helps.

miketzet

kanga said:

What pior had to say about starting at the end is really useful, and would be a great time saver. Apart from that, the way I deal with a complex model is to break it up into as many pats as I need for sculpting. The model below is made up of 16 subtools.

Im only interested in game models so I have to make a low poly version. With texturing in mind I group parts together so they will bake properly. There is no real reason to do this because the lowpoly could easily be one mesh. I just prefer to separate things for more control, and a better end result (like the skin so if I need an sss shader I can do that more easily plus painting the body diffuse in ZB is quicker for me). Breaking the figure into parts making sure none are juxtaposed for baking I group the subttols into a few groups. Here I used 3 groups. I export the groups using fbx and naming conventions for substance painter. Then I copy and decimate the groups and send them to a 3d app for retopo/creating low poly cages. I create the lowpolys and uv them, then export them with the same naming convention system. At this stage I'm not interested in posing the character yet. However if you follow pior's advice stuff like weapons and baggage or any character assets wont get in the way of the end animations.

I then import the low models into substance painter and the groups are identified as texture sets.

The  hipoly model body already has a diffuse so that bakes fine. The hard surface groups have been poly painted in ZB with different 'clown' colors so they can be isolated for making textures. On the left is the assembled model on the right are it's three parts. Once I have extracted the maps I will use the UE4 skeleton by importing it into whatever 3d app I'm using, then skin the parts to it staking the skin modifies for the groups onto the skeleton. Export the rigged/textured model to UE4 and have some fun. Using the game skeleton lets me use the animation sets from the engine. That is how I take a 16 part model and merge it into one asset.

Hope that helps.

You, sir, have been very helpful. I appreciate that you showed me your workflow. So, “controllable” workflows are totally fine, as I understand.

kanga

Hi miketzet. Im not sure what you mean by 'controllable workflow'. The above is just how I develop a character right now. As far as assets go, the character above has a bird, so that is a separate model. If the character had a gun, that would be a separate model if it was a large asset, or if it was small (like a pistol) I would include it in the groups above.