Tabula Rasa: Monte Carlo estimation of unit-variance noise with controlled spatio-temporal correlation
arXiv cs.GR details Tabula Rasa, a Monte Carlo method for generating time-varying Gaussian noise with controlled variance and temporal correlation. The paper frames the problem as joint Monte Carlo estimation of pixel reconstruction and variance, borrowing the idea of sketching from database research.
The practical hook for game and graphics developers is temporal control: the method is meant to support downstream tasks that need stable noise over time, which is often where denoisers, stochastic effects, and temporal filters get tricky. The authors say the approach produces simpler and faster code than earlier methods while still giving explicit control over correlation.
That makes it relevant anywhere noise quality affects visible stability, from rendering research to real-time pipelines that lean on stochastic sampling. The exact downstream integrations aren't spelled out here, but the core idea is useful for anyone trying to balance variance, coherence, and implementation cost.
“simpler and faster code than previous methods”
- what
- Tabula Rasa proposes Monte Carlo estimation of unit-variance noise with controlled spatio-temporal correlation.
- who
- Authors include Tobias Ritschel, Yang Zhou, Nick Milef, Mikhail Dereviannykh, Chen Liu, Christophe Hery, and Carl Marshall.
- when
- Submitted to arXiv on 8 Oct 2026.
- impact
- Could help rendering and simulation pipelines manage temporally coherent noise with simpler code.
Promising graphics method with practical pipeline benefits
Follow graphics updates
See relevant stories in your personalized news feed.
Discussion