Author Topic: AI based interactive denoising from Nvidia  (Read 4161 times)

2017-11-16, 14:41:48

agentdark45

  • Active Users
  • **
  • Posts: 577
    • View Profile
https://blogs.nvidia.com/blog/2017/05/10/ai-for-ray-tracing/


Around the 2 minute mark in the video above.

Very interesting happenings! I wonder what else neural nets could speed up in the rendering process?
Vray who?

2017-11-21, 16:54:50
Reply #1

cecofuli

  • Active Users
  • **
  • Posts: 1577
    • View Profile
    • www.francescolegrenzi.com
It looks very interesting!
Maybe it will be possible to integrate in Corona as a "fast denoise GPU solver" =)

2017-11-21, 19:40:21
Reply #2

Ondra

  • Administrator
  • Active Users
  • *****
  • Posts: 9048
  • Turning coffee to features since 2009
    • View Profile
maybe ;)
Rendering is magic.How to get minidumps for crashed/frozen 3ds Max | Sorry for short replies, brief responses = more time to develop Corona ;)

2017-11-21, 23:40:35
Reply #3

cecofuli

  • Active Users
  • **
  • Posts: 1577
    • View Profile
    • www.francescolegrenzi.com

2017-11-22, 09:31:29
Reply #4

Christa Noel

  • Active Users
  • **
  • Posts: 911
  • God bless us everyone
    • View Profile
    • dionch.studio

2017-11-22, 10:07:04
Reply #5

vlado

  • Chaos Group Team
  • Active Users
  • ****
  • Posts: 37
    • View Profile
NVidia will actually ships the denoiser as part of the OptiX 5.0 SDK so anyone can integrate it in their application. I've looked at the OptiX samples briefly and it doesn't seem too complicated to use it. The denoiser can be used separately - the renderer itself doesn't need to be done with OptiX or use GPUs at all.

However keep in mind that the OptiX denoiser *requires* an NVidia GPU in order to run. If you wanted to use it on a render farm for final frames, you'll need GPUs on the farm (which is what NVidia wants, for sure :))

Best regards,
Vlado

2017-11-22, 10:27:39
Reply #6

romullus

  • Global Moderator
  • Active Users
  • ****
  • Posts: 8843
  • Let's move this topic, shall we?
    • View Profile
    • My Models
And what about memory requirements? Does implementing this, would mean a. lesser system RAM consumption during denoising phase and b. would it be limited by the available amount of GPU RAM?
I'm not Corona Team member. Everything i say, is my personal opinion only.
My Models | My Videos | My Pictures

2017-11-22, 12:44:48
Reply #7

Juraj

  • Active Users
  • **
  • Posts: 4761
    • View Profile
    • studio website
Wouldn't mind to try either :- )
Please follow my new Instagram for latest projects, tips&tricks, short video tutorials and free models
Behance  Probably best updated portfolio of my work
lysfaere.com Please check the new stuff!

2017-11-22, 16:55:26
Reply #8

alexyork

  • Active Users
  • **
  • Posts: 701
  • Partner at Recent Spaces
    • View Profile
    • RECENT SPACES
Faster denoising would be a huge bonus. We occasionally do some stupidly big renders (20k or more) and denoising those takes crazy long, and even on standard 5-8k renders it can be a really slow process. So this would be pretty exciting to see implemented!
Alex York
Partner
RECENT SPACES
recentspaces.com