My name is Nicolas Morel and I am a French 3D Artist based in London.
I studied at ESMA in Toulouse and right after my graduation, in 2017, I moved to London where I am now working. I freelanced for 1 year either as a 3D Modeler or 3D Generalist, but always with a specialization in creature/character modeling. I then worked at MPC London for a bit more than 1 year as a Creature Modeler. There I worked on Call of the Wild, Kong vs Godzilla and Clifford.
I am now working at One of Us as a Creature Specialist where I am in charge of concept, modeling, sculpting, muscle creation, blendshapes, textures, and sometimes lookdev of the assets. Part of my job also includes Pipeline development. I work closely with riggers and my Creature Supervisor in order to develop their Creature Department, which is still fairly new.
Sailor Face Project: How It Started
I first started this project 1 year ago when I wanted to use the new multi-channel displacement map from texturing XYZ. I've always been interested in portraits and especially of old people as I find their faces much more expressive. I did a few of them back in school, a few years ago. But as I knew way less about facial anatomy, I couldn't manage to get any close to what I wanted to reach in terms of quality.
For the last two and a half years, I have been studying organic modeling through my job and my personal projects. For me, they are a fantastic way to keep learning more about anatomy and CG workflows. And in my last few projects, I started to go beyond the lookdev part by just presenting my assets in a neutral environment. For example, in the frame of my project “The Elephorse”, I wrote an article for TexturingXYZ on how to push the creature workflow and integrate an asset into a 2D environment.
And working with Madhav Shyam allowed me to create fantastic animations. We already worked on my Grizzly project where he did an incredible muscle and skin simulation, bringing the asset to a whole new level.
Back to the sailor:
At first, I wanted this project to be quick. These maps were a perfect excuse to start it. Knowing that those maps would bring me the last 10/15% needed to make a portrait believable (even if, in my opinion, I am not entirely there yet), I was able to work more on the sculpt only focusing on the primary and secondary shapes
Another good thing is that you can get a nice result pretty quickly, but if you want to push the details, you still have the flexibility to do it.
Sculpting in ZBrush
For the sculpting part, 95% of the work took place in ZBrush. My main brushes are really basic: Standard, Dam Standard, Move Topology, and Smooth. And that’s about it!
A more challenging part was that I decided to go for an original design. And if you are not careful with how you use your references, your character can quickly become a “Frankenstein”. Being able to choose attributes from different personalities, mix them organically, and make it look natural can be really tricky. And that’s the essence of concept design.
Also, during the whole project, I kept in mind that I wanted to animate the character. So I had to question myself and really think about the wrinkles in terms of motion. And what I find interesting about blendshapes is that it is all about movement, skin sliding, feeling the bones structure underneath the loose skin.
I blocked the shape of the face pretty quickly and spent a lot of time on getting the anatomy right. For this part, use as many references as you can. And if possible, not 3D ones.
When I started this project more than 1 year ago, I already wanted to do some blendshapes. So I anticipated that by doing some retopology over my Zremesher. I made some adjustments to ensure that I was able to make the blendshapes properly afterward (for example making sure my poles are in the right position or my edge flow is correct so that I can get a correct deformation when animating). For that part, I used the basic tool from Maya. I also worked on a really dense mesh, about 120k polygon, as I needed to visualize all the wrinkles in my viewport and relied on the displacement only to bring the micro skin details.
So when I picked that project up again a few weeks ago, I was able to start working straight away without modifying the topology as I already took care of that part 1 year ago.
Texturing the Skin
Texturing went pretty fast. As I said, I used the new multi-channel maps from texturing XYZ with Photoshop. I bought a 20s pack as I wanted as little wrinkles information as possible because I already had them in ZBrush. I first used Photoshop because it allows you to project both the displacement and diffuse on your face so they match perfectly.
So, thanks to these maps I was able to focus on my sculpt and also visualize it with a nice shader without spending 2 weeks hand-painting a nice diffuse and sculpting the skin micro bumps in ZBrush.
And when I later decided to make him a sailor, add tattoos and the burnt skin, and push the texturing, it became necessary to move to Mari. So I simply exported my maps from Photoshop and kept working in Mari.
Speaking of micro skin details, there are 2 different techniques to sculpt and use them to refine the primary ones with the help of Displace from Photoshop/Mari.
The first one, - which is well-known but also very heavy and non-flexible, - is to bring your maps from Mari/Photoshop, project them in ZBrush, sculpt on top of them, and then export the result from ZBrush to use it as a single Displace in Maya. But if you want to keep all your details, you will need to split your model into 5 or 6 subtools. Which means it will be really difficult to make any major change in your project.
The second, more flexible technique which I used is the following one:
Import the microDisplace from Mari into ZBrush but instead of baking it, keep it on a separate layer that I don't touch.
Tip: To do so, you can use Jake Harrel UDIM Importer for ZBrush which is a big time saver. Especially if you have a lot of UDIMs.
- Use this layer as a guide. It is extremely important to use this map not only for secondary shapes but also to refine your primary shapes. A common mistake is to zoom in and focus only on small wrinkles without thinking of the overall shapes.
- When you are happy with your sculpt, you can export your Displace (make sure to hide the layer with the maps before exporting).
- In Maya, you can then use your Displace coming from ZBrush and Mari/Photoshop and manage your 4 channels separately.
A fantastic thing about this workflow is that I keep complete control and flexibility over the intensity of all 3 LODs of the Displace independently of your sculpt coming from ZBrush (the 3 levels being in your RGB channel).
Tip: You can start everything somewhat at the same time going back and forth between sculpting, texturing, and lookdev. One will feed another. And it personally helps me a lot!
Cloth & Hair
For the cloth, I used Marvelous Designer. This guy actually has a complete body. I did the shirt properly in case I wanted to make a full body later. I don't think I will, but at least I know it is there waiting for me! And I used Marvelous Designer because I have always found it hard to get nice realistic folds, plus I wanted to make the fabric with realistic seams. And when it comes to getting realistic seams, this software is perfect with its way of creating UVs based on actual clothes.
Finally, for the hair, I used XGen because I wanted to use something simple and perfectly adapted for Maya/Arnold. Plus you can find a lot of nice tutorials online fo it. I especially recommend those by Jesus Fernandez as they are very user-friendly and well-explained.
During his free time, Madhav developed a motion capture tool using his phone. When I saw the result on one of his characters, I offered him to use my Sailor. We already worked together on the Grizzly project, and as I said, he did an incredible job. So I knew I wouldn’t be disappointed by the result. Plus it was the kick I needed to finally get started with this facial animation project.
I briefed Madhav over some expressions I wanted to implement, like the Sailor staring at the sun, and he did the expression range of motion to test his plug-in. Unfortunately, I won’t be able to tell you more about the recording process. He is still developing his tool and he wants to keep it secret for now.
Refining the recorded data was surprisingly quick because it was extremely clean. But motion capture doesn’t create skin sliding or mass transfer, so I had to manually add them making sure the loose skin was sliding on the bone structure and that we felt it.
When it comes to blendshapes, you can’t really cheat. It either works or it doesn’t. If a movement doesn’t feel right, it is probably because it is not right. And the motion capture obviously helped for that part.
I also added some expressions to make the acting more intense (like the brow raised when he looks at the sun).
Having a wrinkly character can be problematic because when the face moves, these wrinkles either unfold or shrink. Here, I had to use references of old people to make sure I was getting this part right (which was the real challenge of the project and the reason I wanted to do it in the first place). Upper eyelids were probably the trickiest part. The way they completely unfold when he raises his brows and the way they shrink when he closes his eyes are so unique. It is a difficult but such an interesting motion exercise. And it gave me the chance to see if I got my wrinkles flow right!
But what really helps sell this animation is the combination of the 2 methods. The blendshapes brought really nice detail. The motion capture added slight eye movements and facial micro-movements that make the whole animation look really natural.
Setting Up the Renders
I wanted a dry, strong, and contrasting lighting. I simply used an HDRI from HDRIHaven.com which provides high-quality HDRIs for free and combined it with an area light behaving as a sun.
There is no proper post-production on this render as I always try to get the beauty shot as clean as possible. Plus I used the ACEScg colorspace which provides a really contrasting and dynamic render. I only used Nuke to add depth and simulate an F2.8 Aperture. It works really well on portraits combined with a 90mm lens.
That's more or less everything. This workflow is quite simple and nice because the technical part doesn't take too much time and I can focus more on the creative side as well as experiment. I know it can be improved. But even now it is flexible enough so I can freely let my project evolve as I get inspired.