Description
A recurring problem in computer graphics is the representation of radiance information. For instance, it plays a central role for path guiding, a highly relevant photorealistic rendering method that relies on such a representation to facilitate the rendering process. Here, various methods have been explored, including trees [3, 6, 7], basis-function-based structures [2], mixture models [4, 9, 10], and neural networks [1, 5, 8, 11, 12]. Surprisingly, there has been little theoretical work on evaluating different radiance representations w.r.t. to their accuracy, performance, memory efficiency, etc.
In this project, you will work with us on closing this gap by analyzing different state-of-the-art representations. Moreover, we will work on novel representations that have the potential to provide significant improvements to the state of the art. You will join a highly motivated research team collaborating with Wenzel Jakob, a world-class leading researcher in realistic graphics. If you are highly motivated and would like to contribute to research in this context, please contact Hiroyuki Sakai.
Requirements
- High intrinsic motivation to contribute to research
- Experience with C++ (preferably C++17 or higher)
- Willingness to deal with theory and mathematics
Monetary Bonus
We offer a monetary bonus for female students who complete their thesis in the context of this project: for bachelor students EUR 500 and for master students EUR 1000.
References
[1] S. Bako et al. “Offline Deep Importance Sampling for Monte Carlo Path Tracing”. In: Comput. Graph. Forum 38.7 (2019). doi: 10.1111/cgf.13858.
[2] T. Bashford-Rogers, K. Debattista, and A. Chalmers. “A Significance Cache for Accelerating Global Illumination”. In: Comput. Graph. Forum 31.6 (2012). doi: 10.1111/j.1467-8659.2012.02099.x.
[3] J. J. Guo et al. “Primary Sample Space Path Guiding”. In: [26]. doi: 10.2312/sre.20181174.
[4] S. Herholz et al. “Volume Path Guiding Based on Zero-Variance Random Walk Theory”. In: ACM Trans. Graph. 38.3 (2019). doi: 10.1145/3230635.
[5] Y. Huo et al. “Adaptive Incident Radiance Field Sampling and Reconstruction Using Deep Reinforcement Learning”. In: ACM Trans. Graph. 39.1 (2020). doi: 10.1145/3368313.
[6] E. P. Lafortune and Y. D. Willems. “A 5D Tree to Reduce the Variance of Monte Carlo Ray Tracing”. In: [19]. doi: 10.1007/978-3-7091-9430- 0_2.
[7] T. Müller, M. H. Gross, and J. Novák. “Practical Path Guiding for Efficient Light-Transport Simulation”. In: Comput. Graph. Forum 36.4 (2017). doi: 10.1111/cgf.13227.
[8] T. Müller et al. “Neural Importance Sampling”. In: ACM Trans. Graph. 38.5 (2019). doi: 10.1145/3341156.
[9] L. Ruppert, S. Herholz, and H. P. A. Lensch. “Robust fitting of parallax-aware mixtures for path guiding”. In: ACM Trans. Graph. 39.4 (2020). doi: 10.1145/3386569.3392421.
[10] J. Vorba et al. “On-line learning of parametric mixture models for light transport simulation”. In: ACM Trans. Graph. 33.4 (2014). doi: 10.1145/2601097.2601203.
[11] Q. Zheng and M. Zwicker. “Learning to Importance Sample in Primary Sample Space”. In: Comput. Graph. Forum 38.2 (2019). doi: 10.1111/ cgf.13628.
[12] S. Zhu et al. “Photon-Driven Neural Reconstruction for Path Guiding”. In: ACM Trans. Graph. 41.1 (2021). doi: 10.1145/3476828.