Metal Reflection & Brighter Refraction - Bug with OCL

Use this forum for general user support and related questions.
Post Reply
kintuX
Posts: 491
Joined: Wed Jan 10, 2018 2:37 am

Metal Reflection & Brighter Refraction - Bug with OCL

Post by kintuX » Sun Jan 12, 2020 4:24 pm

Maybe it's only on my side | W10, B281a, latest LuxCore 2.3beta (9addec117) | :? So, as mentioned in ENV cache persistance thread,
... There are differences between CPU vs OCL :?
Using OCL:
A. Glass produces brighter caustics (refractive) /// Notice darker shadow with CPU
B. Metal is missing reflections
Image

& here's .blend ready to be checked. Issue also occurs in the viewport, just switch engines.
HDRI "Barcelona Rooftops" from sIBL archives included
Attachments
CPUvsOCL_bug.7z
(7.61 MiB) Downloaded 5 times

User avatar
Dade
Developer
Posts: 3571
Joined: Mon Dec 04, 2017 8:36 pm
Location: Italy

Re: Metal Reflection & Brighter Refraction - Bug with OCL

Post by Dade » Mon Jan 13, 2020 10:18 am

I fixed a problem, recently introduced, with normals of instanced/motion blurred object. Now the metal/mirror/glass object look the same between CPU and OpenCL.
Support LuxCoreRender project with salts and bounties

kintuX
Posts: 491
Joined: Wed Jan 10, 2018 2:37 am

Re: Metal Reflection & Brighter Refraction - Bug with OCL

Post by kintuX » Mon Jan 13, 2020 5:13 pm

Dade wrote:
Mon Jan 13, 2020 10:18 am
I fixed a problem, recently introduced, with normals of instanced/motion blurred object. Now the metal/mirror/glass object look the same between CPU and OpenCL.
Confirming the fix. Working correctly now.
Quick comparison gif:
Image


Note:
The difference in rendered image between OCL vs CPU is really minimal now, basically just a difference caused by sampling (tested with Sobol @ 500 samples).
001_OCL-CPUsobol_difference.png
Great work!

Post Reply