3D Character Artist Andres Zambrano shared a breakdown of his remarkable 3D model of Lady Maria from Bloodborne: The Old Hunters, guiding us through the modeling, texturing, and rendering pipelines and giving tips for new Character Artists.
In case you missed it
You may find these articles interesting
Introduction
I'm Andres Zambrano, a 3D Character Artist located in Florida. I started doing 3D after graduating high school because I liked drawing, video games, and computer graphics growing up. This is why I went to The Art Institute of Tampa and graduated with a bachelor's degree in Game Art and Design.
I've always enjoyed doing different aspects of 3D, including environment and animation, but ultimately decided to specialize in character art during the last year of college after trying other 3D disciplines. Since that, I had opportunities to work on Godfall and its expansion for PS5 and Asgard's Wrath 2 for the Meta Quest 3.
Inspiration
I started this Lady Maria model after seeing a cosplay photoshoot that I liked from Halt, as well as having played the game recently and discussing it with a friend.
For this project, I gathered references from the game along with the official collectible figurines and real-life props that were similar to what the character has in her costume. The main thing that interested me about recreating Lady Maria was the contrast between the rugged and bloodied-up parts of the costume with the more delicate parts of it such as the face, as well as the amounts of different materials that I would have to make for it.
Head & Face
For the head, I went for an idealized realistic face, which means that it has fairly realistic proportions but is designed to look like a character that is aesthetically pleasing and more idealized than a normal person to avoid the head scan look. My workflow was to build a head from different sculpts and scans for its main structure and proportions, and then edit the render features in Photoshop until I liked it, and then I would apply those changes inside of ZBrush.
The face details were made in ZBrush in HD subdivision mode, where I used custom-made pore Array brushes, as well as manually making the directional creases, lip details, and small blemishes. The eye setup consisted of creating the sclera (white part), which includes the lens, a caruncle (fleshy corner) that extends into the eye, and the iris (colored part), which includes the pupil.
The sclera has a circular gradient in Maya for its transmission so that you can see the iris inside, and the caruncle has a horizontal gradient in the opacity so it blends into the eyeball. There is also a tearline (meniscus) around the eye for that highlight you see on the bottom of the eye. All the parts were painted in ZBrush manually, and then the colors were edited with a Circular Color Ramp.
The hat was a lot of trial and error to see what shape looked good and cast shadows on the face in a way that I found appealing, and the feathers were made using XGen Interactive Groom in Maya.
For the hair, I used XGen Interactive Groom instead of regular XGen, since it allowed me to create sculpt layers that I could key-frame, which let me create different poses for the hair.
Outfit & Weapons
For the outfit, I decided to use an old base mesh I had and use Marvelous Designer to make the coat, cape, and jabot. The rest of the parts were either created from scratch in ZBrush or detailed and modified there as well.
The embroidered patterns were made in ZBrush using the Curve Wrap Mode modifier in the brush settings, which enables you to tile your sculpt on the default plane that comes with ZBrush.
All of the weapons were sculpted in ZBrush, and the floral relief details were created by masking out a shape on the base of the weapon, then extracting a mesh from it, and sculpting it. Each weapon took about a week and a half to make.
The material for the coat was challenging in particular, because I needed to use a lot of tiling details on different parts of the coat, and each had to have its own metalness, roughness, color, displacement, and sheen.
I created multiple black and white masks in Substance 3D Painter and used Maya to place the details around the coat. The most important thing about the costume for me was adding elements like the sword sheath and elaborate details on the belts and gems to make them look better or more intricate when seen from up-close in a high-res render, as opposed to far away, which is how you would see it in the game.
Retopology & Texturing Workflow
The topology here is fairly standard grid topology, which I made in Marvelous Designer with its retopology tool so that the seams would line up with edge loops, making them easier to unwrap. Most of the other parts were done with ZRemesher using PolyGroups to help me decide the flow and to have some edge loops where I needed to have seams for the UVs.
The face and hands were also retopologized by hand since their deformation is more important than some other parts. I unwrapped the model in Maya using UDIMs, keep in mind that the UVs are very similar to the way the patterns for the clothes look in Marvelous Designer so that way the tiling details match with each cloth part, have a consistent size and resolution, and are split were the seams of the cloth are.
Most of the character texturing was done in Substance 3D Painter, I mainly focused on getting color and roughness variation on the surfaces. I also used occlusion maps to get more contrast on the cavities and the contact between different meshes. The face was painted by hand as well, and it was a lot of trial and error, using Photoshop to test it before applying the changes in Substance 3D Painter.
Rendering Tricks
For the rendering, I used Arnold in Maya. I used some color correction settings and lens effects like Bloom and Vignette inside of Arnold to tweak the feel of the image. The depth of field was created by capturing the Z-Depth layer in Maya for every render and using that with a Lens Blur inside of Photoshop. The reason I like doing the depth of field in Photoshop is that it's less destructive and of higher quality than the one in Maya, and it also takes less time to render.
Each pose has a different lighting setup for it. My main approach to the lighting was to find a key light that I liked for each pose and then paint the rest of the model with light to accentuate key parts of it. I also used an invisible box environment to get some bounce reflections onto the character.
Conclusion
My main challenge with this character was fixing a lot of the issues I had with the base mesh that I started with, so I had to modify the silhouette and volume on top of the body of the character multiple times after everything was sculpted. The face was also difficult due to all of the trial and error to get it to where I was satisfied with it.
Another challenge was to figure out the workflow to get the cloth and hair to move and look interesting with each pose, as well as find render compositions that looked good and showed the essence of the character.
I would advise beginner artists to use a base mesh that looks right or focus on learning fundamentals of anatomy and body/face proportions before making complicated designs or costumes on characters. Making the character look good and natural before making any clothes will save you many headaches. Lastly, I would say to test out the different features and techniques that you think you will need to know to finish a character by yourself before using them on your character project.