Author Topic: Adaptive real-time denoise filtering  (Read 1818 times)

2013-08-08, 07:27:37

Richy

  • Active Users
  • **
  • Posts: 43
    • View Profile
Adaptive Manifolds for Real-Time High-Dimensional Filtering

http://www.inf.ufrgs.br/~eslgastal/AdaptiveManifolds/

This could cut rendertimes a bit. On low sampled render it will be total fine detail loss, but on higer sampling with low noise, i think it should do good job.
« Last Edit: 2013-08-08, 07:39:20 by Richy »

2013-08-08, 08:51:14
Reply #1

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 8884
  • Turning coffee to features since 2009
    • View Profile
there are multiple papers on similar topic. I am planning to try it one day, since it should be easy to implement such algorithms into Corona. The only problem is that they usually require huge amounts of RAM (tens of gigabytes) even for simple images
Rendering is magic.
Private scene uploader | How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2013-08-08, 09:21:17
Reply #2

Richy

  • Active Users
  • **
  • Posts: 43
    • View Profile
To minimize RAM usage, i think it could be real-time adjustable region on render vindow that shows denoised result, and when render finish sampling, denoise image with something like bucket, or scanline to avoid put all image data into RAM at once.
« Last Edit: 2013-08-08, 10:25:53 by Richy »