Predicting and controlling the appearance of an object is of immense interest in different fields from manufacturing to product visualization. The appearance of a material is a major factor in accurately demonstrating a product in 3D print. Although there have been many attempts to properly demonstrate material appearance in digital format, printing the same doesn’t give the expected results. Moreover, a material with same parameters can look non-identical when viewed in different platforms/software. Ray tracing is a method of graphics rendering which simulates the physical behavior of light.
This PhD project attempts to simplify material properties and represent them without taxing the computational power of general purpose computers as ray tracing is highly expensive. This also includes exploration of neural methods if they can understand the effect of light on a translucent/transparent object. The goal of this project is to utilize the synergy between machine learning algorithms in their understanding of physical phenomenon and computer graphics to accurately represent material properties digitally.
This project is carried out as part of the PRIME Innovative Training Network with DTU as a partner and Jeppe Revall Frisvad as the Supervisor.