CGL-Sim

BRINGING A SMILE TO YOUR FACE

Together with Align Technology and the University of Geneva, we push the frontiers of the medical metaverse by developing data-driven methods for 3D patient modeling and physics-based simulation of soft tissue changes arising from orthodontic treatments.

FOR THE YOUNGEST PATIENTS

Our vision is to enable a stress-reduced treatment for every child born with a cleft - worldwide. In collaboration with USB and supported by BRCCH, we use graphics and AI to automatically compute a plate for the presurgical treatment of infant patients.

Learning 3D Models of Infants

In collaboration with USB, we learn 3D anatomical shape models of infants, including the facial surface and bones, from uncontrolled and incomplete data, addressing a critical gap in computer graphics and medical applications by focusing on the youngest among us.

Surgical planning with 3D head models

In collaboration with USZ, we develop AI-driven methods to reconstruct precise 3D models of patients' pre-trauma head morphology, addressing a critical need in facial-skull injury treatment.

AI DRIVEN PHYSICAL FACES

With Disney Research | Studios we develop a methods to inject physics into face animation technologies. By treating the face as an actuated soft body, we can leverage differentiable physics and neural networks to animate expressions.

INTERACTIVE CHARACTERS

Digital characters are increasingly used in entertainment, education or customer support. We develop methods for data-driven animation synthesis to enable expressive and engaging conversational stylized characters.

AFFECTIVE STATES

Identification of affective states enables the design of emotionally sentient systems. We have developed methods for affective state prediction based on camera recordings, low cost mobile biosensors, handwriting data, and smartphone touch and sensor data.

ART-DIRECTED FLUIDS

Artistically controlling the shape, motion and appearance of fluid simulations pose major challenges in visual effects production. We use differentiable simulation and physics-informed neural networks for style transfer, keyframe matching and artistic deformation.

3D FLUID RECONSTRUCTION

In collaboration with TUM, we enable the reconstruction of volumetric flow motions from sparse and single-view input videos. Our algorithm uses transport constraints, differentiable rendering and neural networks for robust end-to-end reconstructions.