Octane AI denoiser - initial tests, results and next steps Tweet

[Goldorak]

Details below, you can read the post without a facebook account:

https://www.facebook.com/groups/OctaneR ... 102819403/

[Zay]

Holy crap! This look awesome!

[Roboto]

FFFFFFFFFFFFFFFFF&CKKKKKKKKKKKKK!!!

Awesome Goldorak!! AWESOME!!!

[andrey.krivulya]

Woooooooooooooooooooooooooow! Awesome result!

[Lewis]

Interesting, so that is going to be part/feature of Octane 3.07 ?

[coilbook]

so i i would bake it to 1000 samples and still have a little noise to denoise it would be way sharper than denoising at 10 samples?

[Goldorak]

We'll have to train it on 1k vs 10. Basically each training dataset gives you a different denoiser. In out case we started with real time and will then work on offline. That being said, a properly trained denosier can be scary good at very low spp, which is why we want to get that done first.

[coilbook]

We'll have to train it on 1k vs 10. Basically each training dataset gives you a different denoiser. In out case we started with real time and will then work on offline. That being said, a properly trained denosier can be scary good at very low spp, which is why we want to get that done first.

thanks Do you train it or it will train itself while i render? Animation will change frame to frame hopefully denoiser can handle changes like this

[mbetke]

Will it be available for all customers with a license or only for Octane cloud?

[3dworks]

can't wait for its release, hopefully it can be integrated well into plugins? will it work with networked GPU's?