Claudia Marcucci walked us through the Frog Fighter project, detailing how a 2D illustration was brought into 3D with a focus on achieving a stylized look using Substance 3D and V-Ray.
Introduction
Hello! I'm Claudia Marcucci, a 3D Artist from Italy specialising in Character Art at Think Tank Training Centre.
I got into 3D art when I was in high school. I was always interested in technologies and art, focusing on how they could mix together in different media. I started experimenting with video editing using green screens, and then, while watching breakdowns of my favourite movies and video games, I got into 3D art.
After getting a degree in Creative Technologies at NABA (Fine Art Academy in Milan, Italy) as a 3D generalist, I wanted to specialize in 3D Characters, especially for video games. At this point, I followed Scott Eaton's Digital Sculpture course, and then I enrolled at Think Tank Training Centre to step up my artistic and technical knowledge.
The Frog Fighter Project
For our first term at Think Tank, we had to choose a 2D concept to recreate and translate into 3D and try to match it as closely as possible. I found this incredible artwork on ArtStation made by Vladimir Gerasimov, and I fell in love with it. Take a look at his project here:
As we had a limited time to complete it (four weeks and an extra week for tuning and refining), I thought that the best solution was to choose a concept with fewer subjects, to push myself to make it as similar as possible. At first I was not too sure about which style to use, if going for something more realistic or stylized. In the end, I wanted to challenge myself and try to make a 3D render look like a 2D illustration.
I started with planning and collecting the references. When doing the planning, I wanted to push all my strength into the first couple of weeks to give myself enough time to fine-tune the project.
References
My main reference was, of course, the concept art, but I also wanted to collect some real-life images of what it would be in the scene, to better understand the shapes, colors, and material properties. Collecting stylized movie references also helped me to understand how to layer the painted strokes in textures. For example, I used some screenshots from Puss in Boots: The Last Wish, Arcane, and The Wild Robot.
Modeling
I started by importing the concept as an image plane visible only through the camera, and then began a 3D blockout with primitives to match the reference. Setting the right focal length was essential to match the 3D blocking with the 2D concept. After finding a good compromise, I started to refine the blockout in Maya, adding the first details such as the frog's mustache and the scarf.
I preferred modelling everything in Maya except for the frog, which needed some sculpting in ZBrush to create some extra volumes and details to match the concept and make the light interact correctly with the model (like bulges on the frog's belly). I also decided to sculpt smoke in ZBrush because I wanted to make everything in 3D without paintovers or texture planes.
To set up the ZBrush project, I exported the frog blockout and the camera and then imported them into ZBrush. The camera matching was essential for me because I could easily switch between my camera view and the perspective view directly in ZBrush under the Draw window > Channels > Select Camera.
To make the process of texture matching easier, I chose to mark lightly with the Dam Standard brush all of the texture details that I had to match later in Substance 3D Painter (like the black inner lines on the frog's gloves), just to have a reference on where to texture the line later in Painter.
After I reached a satisfying result in ZBrush, I created different Polygroups based on where I wanted to put UV seams after, to make my work easier in Maya. Then I ZRemeshed the model in ZBrush and reprojected all the details from the high poly version on it, so as not to lose any sculpting detail I had made before. Normally, after importing the model in Maya, I would have made the topology manually using Quad Draw, but since I had a limited time and I wanted to be sure to have enough time to focus on texturing and lighting, I opted to keep the topology from the ZRemesher in ZBrush.
I organized the UVs into UDIMs, where I separated the Frog body, hat, scarf, and other elements, such as the smoke and the pipe. The sword also used two UDIMs, divided by materials (wrap and main structure). Then the background reeds occupied a whole row of UDIM tiles, making sure all the mesh in the scene had the same texel density.
After making UVs, I transferred them to the high-poly remeshed version in ZBrush using copy and paste UVs (ZPlugin > UV Master > Copy UVs on the subtool with UVs and Paste UVs on the one that didn’t have it). At this point, I exported a high-poly version with all the subtle marks I made before so I could use them for my baking process in Substance 3D Painter. Using this trick really helped me save a lot of time in the texture process when I was trying to match the original concept. Having the subtle marks on the baked model was just a temporary solution visible only in Substance 3D Painter to speed up the process.
Texturing
After doing the baking process, I began experimenting with texturing. The first mesh to be textured was the scarf, so I could understand how to achieve the effect I wanted with a single color.
The setup I used was pretty simple: I used a fill layer as my base with only base color information. Then I used some color variation fill layers that shifted a bit in values and saturation. For example, in the scarf’s textures, yellow and purple tones worked amazingly to create the desired effect, especially while using a directional noise as a mask on the fill layer to emulate the fabric texture. The main rule I followed was to use curvature information for lighter colors and more purplish/dark colors on occluded areas.
The main advice I would like to share with you is to export your texture maps and already apply them to the material in Maya, even in an early stage of texturing. Plus, having a blockout of the lights in this stage helps you to understand the right colors to use and helps you understand where you need to paint shadows/highlights on the texture.
For the bandages on the frog, I saved the custom material I created for the scarf and applied it to them, with some fine-tuning. I did pretty much the same texturing process for everything in the scene. It's important not to be afraid of layering more colors, even different from the main one: color variation is essential to achieve a stylized and painted effect. Using complementary colors helps a lot.
For the black outline, I used both texturing and toon shading mixed together. I mainly used V-Ray Toon on most of the renderings, but I added a painted outline in texture only when the Toon effect was insufficient. In Substance 3D Painter, I used a Fill Layer with a black base color and curvature masked with some tuning and hand-painted control. It's important to create some noise texture or a filter in the masks that breaks the continuity and perfection of the line to make it look like a hand-painted pencil stroke.
Lastly, I exported a base color map for the smoke (texture painted with tones of blue and white) and two opacity maps: one that helped me control the fading of the external faces, and the second one to make the mesh look transparent in some mesh parts.
Lighting & Rendering
For lighting and rendering, I used V-Ray by Chaos in Autodesk Maya. As I said before, for the lighting, it really helped me to have a rough blockout of how the light would interact with the model in a very early stage of the project. This way, I had enough time to understand how to position every light. It was quite a challenge to match every shadow and highlight in the render with just a few lights, so I opted to use multiple Vray Rect Light strategically where interaction was needed. The two main issues I had while lighting this scene were to create very sharp shadows, essential to make it look like an illustration, and to keep the frog's face in shadow.
For the first issue, I changed the lights' settings and increased Directional and Directional Strength, under the Directionality Palette in the Attribute Editor. This way, I had more sharp lights in the scene. Then, to make the lights not interact with other elements in the scene, I used the Light-Linking Editor, where I could choose for each light which object it would interact with. In addition to that, I also used some planes as light blockers to make the light bounce on a specific surface.
I also created a V-Ray Toon in the Rendering Palette in Maya and tuned the settings to create an outline similar to the concept. To achieve a better result, I also turned on Inner Line Control to have more lines even on the inside of the models. After setting up everything, I selected the Render Elements I wanted to export from the Render Settings.
At this point in Nuke I rebuilt the Beauty and separated every light source in the scene thanks to the Light Mix render element. This way I could have more control on single lights, creating roto masks or changing the intensity.
I also made different renders where I had the foreground separated from the middle ground and background, so that I could composite them together in Nuke. Thanks to the Cryptomatte render element I isolated every geometry in the scene so that I could make a more accurate color correction to make it match perfectly to the concept. Checking the RGB channels individually was vital to achieve the same color.
When compositing the reeds, I added fog using a constant node masked by the Z Depth render element. I used this method instead of creating an Environment Fog in Maya to create a more stylized effect without making it too realistic.
Near the end, I added the Toon Render element. To make it look less CG, I tried to create some line weight with two Erode nodes merged and masked together, as one was used to make the line look thinner and the other one to thicken. Then applied some noise texture to make it look less sharp and overlaid the Beauty color information on it, after blurring it and making it less saturated, to create some value variation in color. At this point, I finally finished the Frog Fighter!
Conclusion
So, some final considerations. I actually reached almost the final result after three weeks, so I took my time to refine it and fix some imperfections. What helped me the most was having a strict schedule where I had to rush in the first weeks to make sure I could do everything in time. It's important to break down your concept before starting a project, understanding what can be challenging and how many hours you can work on it.
I found myself redoing some elements over and over again, experimenting without being afraid to delete everything and restart, sometimes with a clearer mind. The texturing of the frog face took me a ton of time to make it work as the concept. You should not be afraid to delete and redo when something is not fully satisfying you. This makes you faster and helps you understand what you have done wrong before.
I would like to thank my supervisor, Matthew Clapperton, who gave me precious advice on how to improve my project and followed every step of my Frog Fighter. I would also give a special thanks to my classmates and the whole Think Tank community for supporting me! Thank you!