Example-based Creation of Virtual Imagery

The goal of my research is to develop and implement a practical method for generating virtual content. In practical applications, we often need a huge quantity of content to fill up a virtual world. This may serve as a backdrop for some primary content we wish to show off in context, or as a basis for some VR activities. In most cases, we want this filler content to resemble something encountered in real world, but in practice, this content is mostly created manually.

I aim to develop an automatic method capable of generating the greatest possible range of content types from examples by utilizing machine learning techniques. Areas my research is based on include:

• Graphical Markov Models
• Markov Model Sampling
• Shape- and Geometry-generating Grammars
• Texture Synthesis
• Image Inpainting
• Efficient Similarity Search Algorithms
• Efficient Shape Descriptors