System and method for the automated creation of shaders in 3D graphics design

A system for automated three-dimensional (3D) graphics shader design, creation, and optimization, comprising:

• a computing system comprising a processor, a memory, a display, and a network interface;
• a shader optimization engine comprising a first plurality of programming instructions that, when operating on the processor, cause the computing system to:
• receive a plurality of design parameters describing one or more characteristics of a three-dimensional digital model of an object;
• process the plurality of design parameters through a statistical, probabilistic latent variable model (LVM), by performing probabilistic inference conditioned on the plurality of observed design parameters using learned probability distributions, to generate one or more latent variables associated with the plurality of design parameters;
• process the one or more latent variables through a convolutional neural network (CNN) to generate a high-dimensional representation of an initial shader for the three-dimensional digital model of the object;
• process the high-dimensional representation of the shader through the statistical, probabilistic LVM as a second, post-synthesis probabilistic inference operation that evaluates the high-dimensional representation against learned distributions of valid shader configurations and infers a final shader that satisfies probabilistic design and performance constraints for the three-dimensional digital model of the object;
• and
• display the object on the display using the final shader.