Learning to Learn and Sample BRDFs
This accelerates BRDF modeling for computer graphics and vision applications, though it is incremental as it builds on existing meta-learning methods.
The paper tackles the slow physical acquisition of neural Bi-directional Reflectance Distribution Function (BRDF) models by extending meta-learning to optimize sampling patterns, achieving up to five orders of magnitude fewer acquisition samples for new BRDFs at similar quality.
We propose a method to accelerate the joint process of physically acquiring and learning neural Bi-directional Reflectance Distribution Function (BRDF) models. While BRDF learning alone can be accelerated by meta-learning, acquisition remains slow as it relies on a mechanical process. We show that meta-learning can be extended to optimize the physical sampling pattern, too. After our method has been meta-trained for a set of fully-sampled BRDFs, it is able to quickly train on new BRDFs with up to five orders of magnitude fewer physical acquisition samples at similar quality. Our approach also extends to other linear and non-linear BRDF models, which we show in an extensive evaluation.