Workflows for games, movies and other environments

rodpv · May 2014

Hi everybody!
I'm a traditional artist transitioning to digital sculpting. So far I love Zbrush and the feel of it and I'm already getting quite used to it. I'm working on my portfolio as character artist.

What I'm struggling with is the theoretical understanding of the modelling/texturing/rendering workflows.

These are my doubts:

1)I can model, retopologize, texture (with this I mean paint and create passes) and render exclusively in Zbrush and Photoshop. Where would this workflow be useful? Concept art and illustration? 3D Printing?

2)How would a workflow in game cinematics differ from what I do? And what about in-game character and movies?

3)Let's suppose I'm a 3d character artist for a game company. What do I have to deliver to the next department? A simple low poly model, a rendered 3D model or I'd have to render it and then someway extract textures and give those separately? What about movies? The render passes are useless in these situations?

4)What would be an ideal workflow which would let me stay the majority of time in Zbrush without jumping back and forth to Maya or Max?

5)The neeed for UV mapping and unwrapping comes into play only when I have to export my sculpt to other packages, am I right?

6)I followed another tutorial where I sculpted a portrait in Zbrush and rendered in Maya, that's great but what should I learn from this? I always end up with an illustration.

My confusion comes from the fact that I see loads of tutorials and worflows, each one with a different twist. Normal maps, Displacement maps, AO maps, bake, don't bake, render don't render... these are all aspects which I can't understand in my personal workflow since my focus is on creating something with the least amount of programs.

I'll be enormously grateful if somebody has the patience to clear my concerns!

Many thanks!

rodpv · May 2014

Hermit wrote: »

What I think has been the most common workflow
1. Base mesh can be done in old school modeling software such as 3DS Max, Maya, Modo, etc. Or could be done directly in Zbrush.
2. Sculpting in Zbrush.
3. UV in 3DS Max, Modo, etc.
4. Baking xNormal + Handplane
5. Photoshop

What I think is the new workflow
1. Base mesh can be done in old school modeling software such as 3DS Max, Maya, Modo, etc. Or could be done directly in Zbrush.
2. Sculpting in Zbrush.
3. UV in 3DS Max, Modo, etc.
4. Substance Designer for baking, masks, setting up nodes, etc. Maybe some photoshop here and there.
5. Substance Painter
6. Tweak substances in game engine if desired or necessary

Thanks Hermit for your workflows, my doubts are a bit more related to the fact that I don't understand theoretically why do I have to do all that.

I can follow a tutorial which tells me to do that but I don't understand, from my limited artistical point of view, why I couldn't do everything in Zbrush for example? What is the practical need in movies and games for UV?

As I said, I can create a 3D model and render it with Zbrush and Photoshop so I'm trying to understand why this process isn't enough in the industry.

Thanks

martinszeme · May 2014

This is a huge question to answer as there are specific roles in games and more importantly films industry devoted to pipeline alone. But to simplify if you want to create an image on your own its all fine by just modeling, texturing it and then rendering in zbrush as there is no need to change it later, give it to another person to work on it or even animate. When creating a still you don't even need to model half of the objects, as you might not see their other side.
When you are working with models for films for example you need to have detailed, nicely modeled object that look real and also can be animated later down the line. To animate something it has to have somewhat limited amount of detail as its easier to control lets say 50k polygons than controlling something like 20 million polygons that zbrush spits out. Also the lower detailed model has to have polygons flowing in a way that is beneficial for animation. For example face has to have proper edge loops around eyes, mouth etc to deform properly. Once you have this 50k model you want to transfer the detail from the super detailed model so you bake out displacement, normal textures. Rigging and modeling for rigging is a form of art on its own IMO.
Texturing also comes on its own as you have to have an understanding how to fake realistic properties of lets say metal on to some virtual polygons.
Maybe you should watch some making of's like LOTR one or some gaming ones.

rodpv · May 2014

martinszeme wrote: »

This is a huge question to answer as there are specific roles in games and more importantly films industry devoted to pipeline alone. But to simplify if you want to create an image on your own its all fine by just modeling, texturing it and then rendering in zbrush as there is no need to change it later, give it to another person to work on it or even animate. When creating a still you don't even need to model half of the objects, as you might not see their other side.
When you are working with models for films for example you need to have detailed, nicely modeled object that look real and also can be animated later down the line. To animate something it has to have somewhat limited amount of detail as its easier to control lets say 50k polygons than controlling something like 20 million polygons that zbrush spits out. Also the lower detailed model has to have polygons flowing in a way that is beneficial for animation. For example face has to have proper edge loops around eyes, mouth etc to deform properly. Once you have this 50k model you want to transfer the detail from the super detailed model so you bake out displacement, normal textures. Rigging and modeling for rigging is a form of art on its own IMO.
Texturing also comes on its own as you have to have an understanding how to fake realistic properties of lets say metal on to some virtual polygons.
Maybe you should watch some making of's like LOTR one or some gaming ones.

Thanks Martin, those are concepts I was quite familiar with but you put them in quite a direct way.

So that is the game pipeline more or less. What about the film? Is the unique requirement a clean topology with good edgeflow?

For example I'm working on this sculpture right now

Let's suppose I'm working on a film. Now, apart from refining details and finishing sculpt... my worfkflow was to paint it in Zbrush and then render this in Maya or Photoshop.

What would you have to do in a movie instead?

thanks

martinszeme · May 2014

Its actually very similar to games workflow nowadays, just more specific roles, more specialized people and the end product has different use. In your case somebody would model that head in a neutral pose in fairly high detail. After that they would remodel with proper topology (for animation) do couple sets of UVs, animation TDs would test the hell out of it, adding muscle, fat layers (they might skip some of the steps) after that few guys would do couple hundred morph targets. While all of this is going other people would test and do textures, shading, hair/fur (eyelashes, eyebrows, facial hair). After that rendering department would render out several render passes, masks etc and comping department would do their magic for final reel. Again I am simplifying it all and some of the stuff depending on the company pipeline will happen at the same time or will happen before/after.
For example most texturing for films happens in Mari not zbrush as zbrush just can't handle it properly. While Mari can do several 8k-16k textures sets at the same time.

rodpv · May 2014

martinszeme wrote: »

Its actually very similar to games workflow nowadays, just more specific roles, more specialized people and the end product has different use. In your case somebody would model that head in a neutral pose in fairly high detail. After that they would remodel with proper topology (for animation) do couple sets of UVs, animation TDs would test the hell out of it, adding muscle, fat layers (they might skip some of the steps) after that few guys would do couple hundred morph targets. While all of this is going other people would test and do textures, shading, hair/fur (eyelashes, eyebrows, facial hair). After that rendering department would render out several render passes, masks etc and comping department would do their magic for final reel. Again I am simplifying it all and some of the stuff depending on the company pipeline will happen at the same time or will happen before/after.
For example most texturing for films happens in Mari not zbrush as zbrush just can't handle it properly. While Mari can do several 8k-16k textures sets at the same time.

I see so there's an extreme segmentation of roles!

That's quite interesting.

So in my case, since I want to continue develop my portfolio and my passion lies in creating compelling characters with stories (whether a portrait, or a figure, or a creature), should I continue what I am doing or should I also focus on laying out UVs correctly, outputting texture maps, etc?

martinszeme · May 2014

Really depends on industry and even the company you want to apply for. For example Blur and Digic Pictures and pretty much all advert companies will want fairly decent generalists. So people who can do models, UVs, textures, shading, lighting and even animation. Not a lot of those kind of people around so they get paid well.
On the other hand companies like Framestore, Weta, ILM will want pure specialists. They have people who sculpt stuff in clay and then other person cleans it up, retopos, hands it to UVs/Texture department and so on down the chain.
It depends how you want to work. If you want to work with smaller group of people with different backgrounds, skills I'd go with gaming, small advertising, vfx houses as you can directly influence the final result. If you are ok working on very specific tasks with very limited effect on the end result creatively then films is the way to go as you meet the most skilled people in their own field and you might get a chance to work on something great (like LOTR or Star Wars). Again, to each his own.
Small warning about film industry is that it might be bad for family as you have to relocate from gig to gig. Not the best time for VFX houses at the moment.

rodpv · May 2014

martinszeme wrote: »

Really depends on industry and even the company you want to apply for. For example Blur and Digic Pictures and pretty much all advert companies will want fairly decent generalists. So people who can do models, UVs, textures, shading, lighting and even animation. Not a lot of those kind of people around so they get paid well.
On the other hand companies like Framestore, Weta, ILM will want pure specialists. They have people who sculpt stuff in clay and then other person cleans it up, retopos, hands it to UVs/Texture department and so on down the chain.
It depends how you want to work. If you want to work with smaller group of people with different backgrounds, skills I'd go with gaming, small advertising, vfx houses as you can directly influence the final result. If you are ok working on very specific tasks with very limited effect on the end result creatively then films is the way to go as you meet the most skilled people in their own field and you might get a chance to work on something great (like LOTR or Star Wars). Again, to each his own.
Small warning about film industry is that it might be bad for family as you have to relocate from gig to gig. Not the best time for VFX houses at the moment.

Thanks again Martin,
I'd probably be also very happy in a pre-production phase, I'd be glad to do some concept work as Scott Patton with his company. I've seen that they are starting to use Zbrush for concepting right now.

But then again, I would also want to be fairly employable. So would this workflow be something acceptable for me to learn well?

OPTION A - 3D PRINTING
1) concept, base mesh and sculpt: zbrush
2) retopology and UV layout/unwrapping: Zbrush (Zremesher and UVMaster)
3) decimate model and send to print

OPTION B - ILLUSTRATION/CONCEPT ART
1) concept, base mesh and sculpt: zbrush
2) retopology and UV layout/unwrapping: Zbrush (Zremesher and UVMaster)
3) polypaint, extract maps and compositing/rendering: Zbrush+Photoshop
OR Maya/Max

OPTION C - FILMS or GAMES
1) concept, base mesh and sculpt: zbrush
2) retopology and UV layout/unwrapping: Zbrush (Zremesher and UVMaster)
3) extract normal and displacement map - zbrush ?
4) texture painting and outputting - Mari
5a) render high poly in Max/Maya (would I still have to bake normal/displacement in movies?)
OR
5b) low poly creation and rendering in Max/Maya

Obscura · May 2014

rodpv wrote: »

Thanks again Martin,
I'd probably be also very happy in a pre-production phase, I'd be glad to do some concept work as Scott Patton with his company. I've seen that they are starting to use Zbrush for concepting right now.

But then again, I would also want to be fairly employable. So would this workflow be something acceptable for me to learn well?

OPTION A - 3D PRINTING
1) concept, base mesh and sculpt: zbrush
2) retopology and UV layout/unwrapping: Zbrush (Zremesher and UVMaster)
3) decimate model and send to print

OPTION B - ILLUSTRATION/CONCEPT ART
1) concept, base mesh and sculpt: zbrush
2) retopology and UV layout/unwrapping: Zbrush (Zremesher and UVMaster)
3) polypaint, extract maps and compositing/rendering: Zbrush+Photoshop
OR Maya/Max

OPTION C - FILMS or GAMES
1) concept, base mesh and sculpt: zbrush
2) retopology and UV layout/unwrapping: Zbrush (Zremesher and UVMaster)
3) extract normal and displacement map - zbrush ?
4) texture painting and outputting - Mari
5a) render high poly in Max/Maya (would I still have to bake normal/displacement in movies?)
OR
5b) low poly creation and rendering in Max/Maya

Option A: you don't need to retopo and unwrap, because you aren't going to texture it.

Option C:

2) I'm not sure about that the Zbrush unwrapping is enough good in every cases, and I'm sure, Zremesher is not enough when we are talking about games. But it makes good result to use as a base lowpoly, that you can optimize down. But in the case of films, the Zremesher is probably enough.

3) Probably there are better tools for baking normalmaps. Displacement maps are rarely used in games, but they are used in movies.

4) You can polypaint in Zbrush, or you can simply use PS, I think Mari is used more in the movie industry.

5) You don't need any maps if you do simple clay renders

6) you should render the lowpoly in a game engine or in Toolbag, depending on what it is. I'd recommend a game engine if its an environment.

rodpv · May 2014

Obscura wrote: »

Option A: you don't need to retopo and unwrap, because you aren't going to texture it.

Option C:

2) I'm not sure about that the Zbrush unwrapping is enough good in every cases, and I'm sure, Zremesher is not enough when we are talking about games. But it makes good result to use as a base lowpoly, that you can optimize down. But in the case of films, the Zremesher is probably enough.

3) Probably there are better tools for baking normalmaps. Displacement maps are rarely used in games, but they are used in movies.

4) You can polypaint in Zbrush, or you can simply use PS, I think Mari is used more in the movie industry.

5) You don't need any maps if you do simple clay renders

6) you should render the lowpoly in a game engine or in Toolbag, depending on what it is. I'd recommend a game engine if its an environment.

Thanks buddy, gonna try again basing it on your suggestions:

OPTION A - 3D PRINTING
1) concept, base mesh and sculpt: zbrush
2) optional: export render passes for quick clay render in Zbrush or Keyshot
3) decimate model and send to print

OPTION B - ILLUSTRATION/CONCEPT ART
1) concept, base mesh and sculpt: Zbrush
2) retopology and UV layout/unwrapping: Zbrush or Topogun+UVLayout
3) polypaint, extract maps and passes, compositing/rendering: Zbrush+Photoshop
OR Maya/Max

OPTION C - FILMS or GAMES
1) concept, base mesh and sculpt: Zbrush
1a) optional: export render passes for quick clay render in Zbrush or Keyshot
2) retopology and UV layout/unwrapping: Zbrush or Topogun+UVLayout
3) texture painting and output - Zbruh or Mari
4) extract normal and displacement map - Zbrush or xNormal
5a) render high poly in Max/Maya
OR
5b) low poly creation and rendering in Marmoset or Max/Maya

Now here comes another big confusion of mine:
I speak about render passes that I generate in Zbrush like Specular, AO, Depth but are these the same as textures? So when I create render passes and I export them, I'm creating and exporting the textures?

Obscura · May 2014

"I speak about render passes that I generate in Zbrush like Specular, AO, Depth but are these the same as textures? So when I create render passes and I export them, I'm creating and exporting the textures?"

No. Textures are wrapping around your mesh, and your render passes are just images which doesn't contain all parts of your mesh, so they can't be used as texture. However you can bake out the matcap's look from Zbrush, which can be used as a base of your texture.

There is a Wiki section here at polycount, on the very top of the page, and it has very useful informations about almost every aspects of an asset creation, but they are mostly focused on game art pipelines, but still worth to take a look.

martinszeme · May 2014

I would really suggest looking through these two http://www.thegnomonworkshop.com/store/product/396/Character-Modeling-for-Production http://www.thegnomonworkshop.com/store/product/425/Character-Texturing-for-Production#.U3zWJ_ldV8E They're a bit old, but at the time they blew my mind about production.
Though loads of Gnomon stuff is worth looking at.

EDIT: Forgot these two by Bulgarov: http://www.thegnomonworkshop.com/store/product/544/Character-Design-and-Modeling-for-Next-Gen-Games http://www.thegnomonworkshop.com/store/product/949/3D-Design-for-Production#.U3zXgvldV8E

rodpv · May 2014

Obscura wrote: »

"I speak about render passes that I generate in Zbrush like Specular, AO, Depth but are these the same as textures? So when I create render passes and I export them, I'm creating and exporting the textures?"

No. Textures are wrapping around your mesh, and your render passes are just images which doesn't contain all parts of your mesh, so they can't be used as texture. However you can bake out the matcap's look from Zbrush, which can be used as a base of your texture.

There is a Wiki section here at polycount, on the very top of the page, and it has very useful informations about almost every aspects of an asset creation, but they are mostly focused on game art pipelines, but still worth to take a look.

Damn this is really confusing. I read a bunch of stuff and yet I wonderf if I'll ever grasp all this technical jargon: this guy here talks about render passes and he shows a pic of em http://www.cgarena.com/freestuff/tutorials/zbrush/monster/monster4.html

I see AO,spec, depth, oil.... so these are not texture? They are not MAPS but render passes?

I've seen the section but it's so difficult to make sense of all of this in an actual pipeline

martinszeme · May 2014

Rendering in passes means that instead of getting one image that 3d application gives you, you get the same image split up in layers. It make stuff more complicated, but at the same time it gives you ability to change things. For example if you rendered a character on a stripy background you can now get the background pass or the mask for the background and make the background whatever you want. Same applies for the character. If you want you can make characters hairs red instead of black, make his sword golden instead of plain steel. You could do the same with just a single render with all flattened, but it would take more time and effort. It especially comes in handy with animations/rendered footage.

Textures are the images you put on models to make them more believable. Like a brick wall needs brick texture to look real, to have those reddish, yellow bricks and to also have different reflectivity (grout would be less reflective than smooth bricks etc).

Obscura · May 2014

"They are not MAPS but render passes?" Yes. These can be used as later added layers of your rendered image, so this works mostly when you make still images, or any pre-rendered stuff (including movies). They are added together with different blend modes later to get the desired result.

I took a look again at your link, and there is a diffuse pass. The diffuse texture is visible in it, but it can't be used as a texture when its just a render pass, because then it contains only that parts of your object which are visible from the current camera angle.

Other thing, you don't need to understand every possible situations, a general understanding in things is enough. You need to decide what you want to do, like movie characters, game characters, movie environments, game environments, animations, highpoly sculpts for still images, etc, and you need to be really good just in the choosen topic.

rodpv · May 2014

Obscura wrote: »

The diffuse texture is visible in it, but it can't be used as a texture when its just a render pass, because then it contains only that parts of your object which are visible from the current camera angle.

This sentence totally clarified my misunderstanding.

thanks a lot guys!!

vargatom · Jun 2014

martinszeme wrote: »

For example Blur and Digic Pictures and pretty much all advert companies will want fairly decent generalists. So people who can do models, UVs, textures, shading, lighting and even animation. Not a lot of those kind of people around so they get paid well.

That's true for Blur where "character modelers" are also doing texturing, shading and even hair work.

But here at Digic we work with specialists - sculptors, modelers, texture painters, shading artists, hair TDs. Compositors usually also do the lighting though, and give feedback and even adjustment values for the shading artists.
It's been like this from the start as we've always had a distant goal of moving past game cinematics CG work. So the entire company is built around such principles.

We do have some generalists though as there are always going to be cases where it's a lot more efficient to have a single person working on an asset who's capable of combining different approaches. But it's usually for the set and environment stuff.

vargatom · Jun 2014

martinszeme wrote: »

To animate something it has to have somewhat limited amount of detail as its easier to control lets say 50k polygons than controlling something like 20 million polygons that zbrush spits out. Also the lower detailed model has to have polygons flowing in a way that is beneficial for animation. For example face has to have proper edge loops around eyes, mouth etc to deform properly.

Hehh

we're already breaking our existing records on our current projects, some of the characters are close to 1 million quads before subdivision. Previous champion was Halo 4 Master Chief at 700K. Our current heads are between 25 to 40K quads, we need that detail to sculpt facial wrinkles for the expressions.

I can hardly wait for computers to get fast enough so that we can easily manage raw meshes from sculpting apps without retopology, as it can take a lot of time, especially for hard surface stuff. I mean at 500K quads you can sculpt almost anything independent of the topology - but at this point we still need to pay attention to it and it can get incredibly time consuming.

I imagine this next step will happen in something like 5-10 years and I really can't wait to get there.

vargatom · Jun 2014

Yeah Vitaly is also a nice example of current limitations. He can model super amazing looking stuff that needs no subdivision but it's got horrible topology that you can't subdivide or UV map properly. Then again he's so good that it makes sense to hire 3-5 guys just to retopo his stuff

But, if your pipeline could handle the raw stuff, you could skip hiring those guys, you could have the results immediately in the shots and you could do very quick revisions as well...

Workflows for games, movies and other environments

Replies