View all the Finalists for Rookie Awards 2024
Giacomo Cervi - The Rookie Awards 2024

Giacomo Cervi - The Rookie Awards 2024

Giacomo Cervi
by PaulVarious on 1 Jun 2024 for Rookie Awards 2024

I'm excited to kick off my presentation for the Rookies Awards 2024 with 3 projects! The FIRST ONE is my latest work. I took care of all aspects of the model, from sculpting to the final render.

99 1870 25
Round of applause for our sponsors

Inspired by the incredible work of Kris "Antropus" Costa, this was my very first time tackling such a complex project, diving deep into not only modeling but also look development, grooming and rendering.

Everything you see was created using the following software: ZBrush, Maya, XGen, Marvelous Designer, Substance Painter, Substance Sampler, Arnold Renderer, and Photoshop.

Gathering sufficient references is crucial when it comes to realism. One of the best resources for reference images of the human head is Daniel Boschung's website.

The goal was to create a realistic character resembling a Native American. To achieve this, I didn't rely on a single reference but rather a collection of them. By taking cues from various sources for different facial features, I aimed to create a fictional character that would still look believable and realistic.

We see dozens of faces every day, which is why it's easy for us to differentiate between a CG model and a photograph. The well-known Uncanny Valley.

The lookdev part was, for sure, the most intensive. To achieve highly believable realism, it is essential to properly set the shading parameters for all facial components, not just the skin (particularly specular, displacement and subsurface scattering) but also the eyes and grooming.

For the eye, I tried to get the most realistic result possible while keeping the model as simple as I could. I divided the eyeball into just 2 parts: a mesh for the Iris and a single mesh for the Sclera and Cornea combined.

I used a ramp node to handle the transition between them, allowing me to achieve a realistic and physically accurate rendering of the eye without needing separate meshes.

Additionally, I sculpted the caruncula and modeled the meniscus and the pupil shapes to complete the final eye.

The Character was entirely modeled in Zbrush. To achieve such a high level of detail I relied on sculpting in HD Geometry.

Everything you see is hand sculpted without the use of texture maps of any kind.

Below you can see the main costum brushes I created to achieve the various skin details.

Once the modeling phase was complete, I used ZBrush's polypaint to create the skin texture and all the necessary maps for shading the skin in Maya.

I promptly created an initial set of HDRI renders in Maya to begin testing the maps exported from ZBrush right from the start.

After completing the shading of the skin and eyes, ensuring that displacement, subsurface scattering and specular/coat highlights were working properly, the final major step was to create the groom for both the head and clothes.

To achieve this, I decided to use Maya's XGen Interactive Groom. I created guides for each description to maintain a hybrid workflow between the classic XGen and the interactive version.

This project was an invaluable learning experience for me, significantly expanding my skills and understanding. 

Can't wait to apply the knowledge and techniques I have acquired to future projects, further enhancing my skills and creativity.

The second project is "C'MON," my winning entry in The Meet MAT 3 Contest, representing BigRock Institute of Magic Technologies. Everything you see is created using displacement. No additional geometry was added!

I kicked off my work with an extensive search for references, drawing inspiration from various sources. These references served as the foundation for my project.

The "modeling" was done entirely in Substance Painter using only displacement.

I started with the primary shapes. Once these were completed and verified to work correctly with the rest of the model, I began modifying them by adding various details and textures.

The most complex part has been for sure the Front Camera Lens. Below is a brief breakdown of the process.

The main trick that allowed me to achieve such an extrusion was to use not one but two different overlapping alphas as the starting base.

All the details, such as the red line or the dents on the border that appear to be modeled on the side of the lens are actually on the front of one of the two alphas. Due to the strong displacement, they get stretched on the sides.

Another significant challenge was ensuring coherence between all parts of the model to make it both visually appealing and functional, as if it could be a truly working camera.

For my third and last project, I'd like to briefly showcase the work I had the opportunity to do in collaboration with Wonder Dynamics.

Their motion capture tool, Wonder Studio, is truly phenomenal! In just over two weeks, starting with a stock video, I was able to create a finished product, handling all the production steps myself.

The software provided me with the 3D model of the robot, a Blender scene containing motion capture data and an HDRI, and an actorless clean plate of the video.

I took care of the following tasks:

- animation/clean plate cleanup

- basketball tracking

- dress creation/simulation with Marvelous Designer

- texturing

- lookdev/rendering

- compositing

Thank you for getting this far!

I wish the best to all participants! The level of entries this year is something crazy!

A huge shout-out to the organizers, judges and sponsors for giving us the opportunity to participate to this amazing contest!

Best, Paul Various

Comments (25)