Character Artist Manuel Arroyo Arrebola first blew away the Substance 3D team with works such as Doctor and Doll, which showed remarkable skill with both the Substance toolset and Marvelous Designer, as well as a knack for capturing the spirit of Rockwell-esque early twenty-first century art. When we began searching for an artist who could test out and showcase the abilities of the Substance in Marvelous Designer plugin, Manuel was a natural choice.
My beginnings in the world of 3D go back to a very specific moment in my childhood. When I was 3 years old, living in Torredelcampo in the south of Spain, my parents gave me a VHS tape of the film Toy Story. It was the first time that I’d seen a 3D animation film and I instantly fell in love with it. I watched it over and over again, rewinding it back to my favorite scenes and wondering how such a movie could be made.
This persistent sense of wonder was one element that caused me, many years later, to sign up for a ZBrush summer course. I enjoyed it so much that I knew that I wanted to dedicate my life to 3D, and to spend as much time as possible learning to improve my skills.
Fitted for a Suit, a final render from the conclusion of the project.
1: Illustration details
Why Leyendecker or Rockwell?
With any illustration or concept art, I first notice its ability to transmit something, and evoke feelings with just one look. I search for characters that convey their own emotions. And I look for overall composition, as well.
The first time I saw a Joseph Christian Leyendecker illustration I knew that I had to represent his work in 3D. I began exploring the work of both Leyendecker and Norman Rockwell, whose work was massively influenced by Leyendecker, and I fell in love with both artists. I loved the way they represent everyday life in their painting, as well as their use of art as a protest against the social injustices of that time. I even loved their own personal stories.
For this particular project, I ultimately settled on a 3D representation of one of Leyendecker’s works, one of his iconic covers of The Saturday Evening Post. I was conscious that I’d need to use Marvelous Designer, so I set about learning how this software could help me to create a piece of artwork that was a suitable homage to Leyendecker’s image. And I was also already considering how the tailor in the image might provide a little link to the world of sewing, both in real life and in 3D.
JC Leyendecker’s original magazine cover, published in April 1916.
The piece consists of two main characters, connected by the tailor’s arm. The challenging part when it comes to translating the illustration into 3D it is to provide the continuity to the scene between the connection of the two characters, and also to credibly represent in 3D some of the characteristic elements of Leyendecker’s work, such as the wrinkles in the clothing or his characters’ expressiveness.
The perspective in this scene presents some difficulty as well. Here, Leyendecker fakes some elements of the perspective in order to get that final look, such as the perspective of the tailor’s right arm, behind the boy, or some of the anatomical proportions. When you’re viewing a scene in 3D, potentially from a range of different angles, you can’t really get away with this sort of visual sleight of hand; any visual trickery is immediately much more obvious. And so it’s here that you need to find a balance between a completely faithful adaptation of the original work, and recognizing the parts you have to sacrifice to remain as true as possible to its overall concept.
2: Modeling Steps
When I start the modeling phase, first I carry out a general analysis of the illustration, to identify which element of the picture presents the easiest starting point. In this case I choose to start with the boy; his pose is more static, and there’s less of a time investment in adapting and modifying this.
I always start characters the same way: in ZBrush, I start from a sphere and then I continue creating a general form. When I have a simple blocking of the proportions and the anatomy, I keep adding details to the face, and then to the body. I constantly check the upper left window in the ZBrush UI, which shows the negative version of the model, the alpha. That’s a quick way of keeping an eye on the silhouette of the sculpt, allowing me to compare it with the original illustration. When the sculpt is done, and detailed, I start placing subtools, to give me an idea of the project’s final aspect.
Once I’ve finished this stage in ZBrush, I transfer all the models into Maya, where I start working on the topology and UVs, making sure the model is correctly laid out across multiple UV tiles (UDIMs). During this step I also model the rest of the props that give form to the scene, like the boy’s hat or the tailor’s glasses.
Then, with these changes made in Maya, I bring everything back to ZBrush and project this onto the previous ZBrush version of the model; this has the effect of adding in all the necessary details.
Clothing, and the Substance in Marvelous Designer plugin
Now it’s time to start clothing our characters. Before I start creating clothes in Marvelous Designer, I carry out an exhaustive search for the patterns in various image search engines. Finding exact patterns is quite complicated, because they’re really scarce, and because they tend to vary a lot depending on the age or physiognomy of the character. For instance, children typically have shorter arms and legs than adults, so the patterns on their clothes tend to be correspondingly smaller. It’s tricky to find the exact pattern you need.
Once I’ve chosen the most neutral and most appropriate pattern for the characters, I upload it into Marvelous Designer as an avatar, and then draw on top of these patterns over the avatar in the 2D windows. It’s frequently necessary to have to adapt these patterns to the avatar shape of the body, and it almost always happens when the character has cartoon proportions.
While I’m creating the different parts of the outfit, I distribute them by layers; this helps to avoid any intersection or flickering problems. Once the outfit is complete, I use the Substance in Marvelous Designer plugin to rapidly view different textures over the cloth adapted to the character. For this, I use my personal predefined materials in Painter and adapt the scale and roughness. This allows me to move ahead in the workflow, and makes the texturing process in Painter easier.
When I’ve finished the clothing, and I’m happy with the final result, I use the Create and Edit Topology tool; I find this adds a lot of speed to the Marvelous Designer workflow. This allows me to create the topology directly, and automatically adapt it into the simulated clothes. Again, I find it’s easier to adapt, create, edit, and visualize the topology this way rather than by using other software. Once the topology is done, I add final touches in Maya. Then I add the clothes to the rest of the project, organize all the UVs, and then I transfer everything into ZBrush once again.
Back in ZBrush, I try to achieve the same wrinkle style as Leyendecker’s, but in 3D. This was undoubtedly the most complicated part of this project. To make this easier, I use the Transpose Master tool; this allows you to unify all the subtools at once, so that we can move, modify, and pose the model as needed with masks . While doing this, I also add wrinkles, and detail them. Finally, when the high-poly model is finished, it’s time to export two versions of the scene, one low-poly, another high-poly with decimation carried out. This allows us to have two different versions to bake in Painter.
Materials and textures
With the low-poly and the high-poly model exported from ZBrush, I create a new project in Painter, then import the mesh, and bake it with the resolution that seems most appropriate. My workflow usually requires the option to work between different UDIMs, which provides more freedom and ease when it comes to texturing. When I have to choose materials and Smart Materials, I usually do a preliminary study of what kind of material each asset is composed of, which can be really tricky sometimes. In the case of the fabrics for this project, I have to investigate which ones were used at that time (Leyendecker’s image was first published in 1916) and the exact type of material each piece of clothing was made of in order to represent the 3D modeling as faithfully and authentically as possible to the original work.
I create the distinctive pattern of the boy’s vest with a predefined sample file, ‘Tiling material,’ that comes by default in Painter. It’s a sample that I’m used to working with in order to create tileable textures quickly. Also, when I work in Substance I add post-processing effects ; those I use a lot include the Iray renderer with my own HDRI. Like this, I can see how the project will look in the final render.
Substance Workflow, ACES configuration
I modify the post-processing and texture configuration to adapt it to my workflow with the ACES color profile. ACES allows me to present a visual aspect that’s closer to the range of perception of color of the human eye. I highly recommend the ACES color profile, because it can greatly improve your projects.
First of all I open the Display Settings menu and modify the color profile, by adding this specific file, ACES LUT for Substance Painter.
Then, in the Post-Effects tab, I activate Tone Mapping and change the Function to Log. As a last step I modify the Channels of the textures in Texture Set Settings with this configuration:
After this, I can visualize the Substance Painter file in the ACES color profile, export the textures, and test each character’s shader.
4: Xgen, Lighting and Rendering
In this section I will explain the process of creating the hair of the characters, and the lighting and rendering of the scene.
The characters’ hair was challenging for me because it was a new area for me; I learned how to create hair on this project, which I really enjoyed. I found that it’s easier to recreate uniform and relatively compact hair, or this kind of hairstyle, than it is to recreate messy or curly hair, or a complex hairstyle such as a braid.
Creating the characters’ hair involves a process of adding guides to create hair strands of different lengths, and to provide the direction of each character’s hairline or hairstyle. Doing this again and again, I’m able to gradually fill in each character’s entire head, readjusting parts if necessary. Of the two characters, the tailor’s hairstyle is more complicated, due to the very characteristic shapes of his locks of hair. I have to reproduce these perfectly in order to stay true to the original concept.
In the case of the lighting, I have to distort it and play with a lot of parameters, as the lighting of the scene is not realistic. For example, the tailor’s left arm casts a very hard, sharp shadow on his apron, while the tape measure does not seem to be affected by that shadow. This also occurs with the boy’s hat, which should be casting a shadow on his forehead – but his face is perfectly illuminated, while a relatively hard shadow is created on the brim of the hat. I have to analyze all these aspects in order to resolve the scene with lighting composed of white lights, or sometimes slightly bluish lights, with some bounces of orange light and a slightly complex structure.
With the lighting setup ready, it’s time to render the whole project. I carry out the rendering process with Arnold; this is the engine I’ve used from the moment I started learning and working in the 3D field. Arnold provides versatility and an excellent finish; that’s why it’s always been my main choice. I usually work a lot with Arnold Render View, which allows you to rapidly preview the final result of the render. When configuring the sampling, I always adapt it depending on the project I am working with. I use this configuration:
— I set the Camera Anti-Aliasing high, and I increase it even more if the project includes XGEN.
— If the project is composed of multiple metal objects and crystals, I increase the specular and transmission parameters a lot.
— Finally, since all the projects I work with tend to have subsurface scattering (SSS) to a greater or lesser extent, I always leave it medium-high.
These renders are in EXR or PNG format, depending on the type of project. I also tend to include the alpha pass so I can later do post-production in Adobe Photoshop. I usually take different camera shots to show unique or different aspects of the illustration. Even with a project like this one, based on a 2D concept, this of course differentiates it from a flat 2D image, allowing for more inventiveness, and many more possibilities. These planes I use for these camera shots are, main (the one that most closely resembles the reference illustration), American shot, long shot (high-angle or low-angle), medium shot and close up. I always render a turntable or semi-turntable as well to show the model at different angles. The resolution of the static planes is 4K, while the turntable images are 2K, with fewer samples.
And finally, once we have all the shots, it is time to add the final post-processing touches in Photoshop. Here, I play with the following different elements of Photoshop:
— I tune the details: specifically, I duplicate the image, desaturate it, add a high pass effect and put it in High Light blending mode. I also adjust the opacity.
— I add a Color Lookup. The choice of the LUT varies depending on the mood we want to convey; it also simply depends which one best suits the final illustration.
— I modify the image with different filters: Brightness/Contrast, Hue/Saturation, and Curves.
— Finally, I add some aesthetic touches, such as enhancing the brightness of the eyes or adding a little depth to them, or retouching some colors.
At last, the render has gone through the final process and is ready to be exported. I export it in PNG at highest quality.
5: Problems and solutions in modeling to translate the concept
I encountered various problems in interpreting the illustration. The, first and most complicated, has been the perspective of the tailor’s right arm. This arm goes from the lapel of the vest to his own hand, and it is completely distorted. It’s been really difficult to adapt this arm to make it believable in the 3D model.
A side-by-side comparison, Leyendecker’s original image (left) and Manuel’s 3D adaptation (right).
Another problem was the expressions of the characters. In the original image, the child has a neutral pose, and his frame of mind is difficult to identify; trying, therefore, to imbue him with a particular feeling or emotion was a complicated task. At the same time, the hands of the tailor that are posing and holding the vest, have their own rather complex form. Making that form credible and faithful to the concept has been quite difficult.
As I mentioned above, one of the great challenges of this project was the wrinkles, specifically the types of wrinkles that Leyendecker has illustrated on the tailor’s left arm. And so the complication here was not only creating the wrinkles, but when generating lighting also taking into account that the projection of the wrinkles has to be the same as or similar to the original work.
Finally, getting the same camera shot as the original concept was really tricky. The shot with the boy is more low-angle, while the perspective of the tailor is much straighter – and so this is only really possible if the tailor’s torso were much more angled. I had to play around with the field of view quite a bit to get that look of an illustration that’s present in the original.
These points have made this illustration a real challenge, as well as a beautiful exercise in self-improvement and adaptation.
6: Final thoughts and final renders
I’ve worked on many animation cinematics for projects like Apex Legends, Sky and Call of Duty, and companies like HBO or Facebook. Right now, I’ve taken a few months off from work as my old job this past year proved to be a pretty heavy burden, mentally and physically, particularly during this global pandemic. I’m still creating personal projects, still growing and learning, and charging my batteries to come back better and stronger. This kind of project, Fitted for a Suit, has been helping me a lot. Creating the artwork for this project has helped me to channel all the energy accumulated throughout this stressful time since the pandemic began. It’s been really comforting and satisfying to see how, little by little, the illustration was coming to life in 3D.
Fitted for a Suit marks a ‘before and after’ turning point in my personal portfolio, and in my personal career as an artist. and representing an illustration of this magnitude, and being able to adapt such a wonderful piece from Joseph Christian Leyendecker, has been an adventure. This is the biggest challenge I’ve ever encountered when it comes to representing something in 3D. This piece exudes magic, feelings and tradition, and being able to represent the features of this fantastic piece in my own way is a real honor.
Trying to capture that style, and that quality that Leyendecker had as an artist, has been complex; I think that my overall representation of this wonderful artist’s work has proven a worthy homage. I couldn’t be happier or more satisfied with the final result.
Thanks to the Substance team for providing me with the freedom to reinterpret this piece however I wanted to. Leyendecker was a phenomenal artist, but perhaps isn’t so well known in the 3D community. Being able to give him more visibility in this way is something I’m immensely proud of.
Going forward from this project, in the short term, I simply plan to keep growing and improving as an artist, feeling increasingly fulfilled with the work that I do, and being able to have an emotional impact on the people who see my artwork. In the longer term, perhaps I’ll one day be able to fulfil my professional dream of working with Disney or Pixar – this would, in a sense, bring my career full circle, back to my love of Toy Story that started everything off.
Meet Manuel Arroyo Arrebola
Manuel Arroyo Arrebola is a freelance character artist and 3D generalist currently based in Alicante, Spain. His professional credits to date include work on projects such as Apex Legends, Sky, and Call of Duty.