Aktuelle Forschungsfragen im Bereich des Visual Computing, der Informationsvisualisierung und -analyse und 'Computational Photography'. Themenvorschlaege finden sich auf der aktuellen Webseite des Seminars.

Topics for this year's seminar include non-photorealistic rendering (NPR), image-based rendering, as well as other topics involving machine learnning, e.g., sketch learning, artistis style transfer, or image-to-image translation.

Non-photorealistic rendering (NPR) is an area of computer graphics that focuses on enabling a wide variety of expressive styles for digital art. It focuses on photorealism in contrast to traditional computer graphics. In image-based rendering, light-field based approaches may be studied for generating new views from arbitrary camera positions. For the rendering purpose, creating 3d structures from images may also be an interesting topic.

In addition, with machine learning methods, especially neural networks, more tasks concerning computer graphics can be better handled and thus may yield some quite interesting applications. Features are better extracted from images, with which people can better understand the images to deal with tasks like classification or sketching. A image may even be merged with features from other images to creat new ones, e.g., turning the horses in images into zebras, or turning a photo into an artistic painting.