What is Photo-to-3D Character Generation?

This feature leverages innovative AI technology to automatically create a unique, animation-ready 3D digital character from a single 2D picture representing a frontal view of the character's head. The input picture can be the photography of a real person, a generated photography, or a reasonably photo-realistic concept art picture.

Importantly, the generated character is in the technical style defined by the user (same mesh topology, material set, animation rig). Unlike classical 2D-to-3D methods, the method implemented in Popul8 works by applying variations to an existing template, rather than generating meshes and textures from scratch based on non-proprietary training data.

Therefore, automated 2D-to-3D character generation enables the user to populate games with unlimited variations at reduced cost, while ensuring consistency with the game art style and technical specifications.

Use in Popul8

You first need to import a photo file representing a frontal view of a person, in neutral expression. See the detailed instructions to help you produce optimal photos for character generation.

In Popul8, Right-click on the Project panel and select Import → Files. Choose a photo file in PNG or JPEG format The imported photo file will appear in the Project window.

Next, create a photo-to-3D character generator from any existing character Template by right-clicking on the template and selecting Create > Photo to 3D Character .

Link the imported file by dragging it into the Input photo box of the Photo-to-3D Character generator, name the character to be created, and finally click Apply Changes.

References

LinkeIn Post: The Future of 3D Generative AI and How Popul8 Fits In

Dias, Mariana, Pedro Coelho, Rui Figueiredo, Rita Carvalho, Veronica Orvalho, and Alexis Roche. "Creating Infinite Characters From a Single Template: How Automation May Give Super Powers to 3D Artists." In ACM SIGGRAPH 2024 Talks. 2024.

Dias, Mariana, Alexis Roche, Margarida Fernandes, and Verónica Orvalho. "High-fidelity facial reconstruction from a single photo using photo-realistic rendering." In ACM SIGGRAPH 2022 Talks. 2022.