Presentation
Interactive Monte Carlo Denoising Using Affinity of Neural Features
SessionTechnical Papers
Event Type
Technical Paper
Research & Education
Ultimate Supporter
Ultimate Attendee
Exhibitor Ultimate
Time
Location
DescriptionOur method computes weights of a series of dilated kernels using pairwise affinities between learned per-pixel features. It achieves state-of-the-art, temporally stable results at interactive frame rates at 1080p resolution.
