but yeah, testscenes... I'm super busy right now to make new project (and my current work renders(also using luxcore ) are under NDA and they aren't interiors to test)
Side note: try to export a scene and move around the camera in LuxCoreUI ... an hardware vendor could setup a nearly real-time demo station by using some serious hardware.
Actually, I've never used luxcore ui. I only know how to export from blender.... Don't know what to do next
Dade wrote: Wed Feb 13, 2019 4:41 pm
Side note: try to export a scene and move around the camera in LuxCoreUI ... an hardware vendor could setup a nearly real-time demo station by using some serious hardware.
Actually, I've never used luxcore ui. I only know how to export from blender.... Don't know what to do next
Run luxcoreui.exe, load the exported .bcf(or .cfg) file, after the rendering started, you can just use arrow keys or the mouse to move the camera around.
P.S. disable HDR/sky visibility map or it will be updated at every camera movement.
Support LuxCoreRender project with salts and bounties
Just manage to find time to make some PGI test vs Pure Pathtracing. here are the result :
Look Like my cache is not that good ( seem that 500.000.000 photton count/Max size + lookup radius = 5cm could give better cache) But this computer only have 8giga of ram (with 4go already used ).blender slowdown and crash when i test with such number
PGI Debug
PGI Settings
cache settings.jpg (16.2 KiB) Viewed 5134 times
PGI 10mn CPU i7 2600
PGI Denoise OIDN
Pure Path 10mn CPU i7 2600
Pure Path Denoise OIDN
Support LuxCoreRender project with salts and bounties
Sharlybg wrote: Wed Feb 13, 2019 10:19 pm
Just manage to find time to make some PGI test vs Pure Pathtracing. here are the result :
Look Like my cache is not that good ( seem that 500.000.000 photton count/Max size + lookup radius = 5cm could give better cache) But this computer only have 8giga of ram (with 4go already used ).blender slowdown and crash when i test with such number
PGI Debug
Japan PGI + cache 10mn.jpg
PGI Settings
cache settings.jpg
PGI 10mn CPU i7 2600
Japan PGI 10mn.jpg
PGI Denoise OIDN
Japan PGI + OIDN 10mn.jpg
Pure Path 10mn CPU i7 2600
Japan PathCPU 10mn.jpg
Pure Path Denoise OIDN
Japan PathCPU+ OIDN 10mn.jpg
the whole floor has no samples, I wonder where is the issue with this
I assume we are in a pretty good spot if people complain for rendering this in 10mins ... on a i7 2600
Sure !
I also still think there are room to improve the render on PGI to get better performance as my cache seem to be poor in quality. Or no
I assume cache quality must be interpreted like this :
Bad
Good
Very Good
Or i'am Completely Wrong
the whole floor has no samples, I wonder where is the issue with this
I think all shiny surface at some point are not included in the cache. The table / the Lamp / Tv + Tv furniture / white chair in the other room / and the floor.
Support LuxCoreRender project with salts and bounties
If there would be a way to estimate cache quality then we might end up with quality parameter instead photon count and cache size.
Have photon count 5 or 10x the cache size and scale them using some estimation that would be controlled using a multiplier.
Don't know how to set a proper lookup radius as this can be very prone for leaking/bleeding or cache fireflies as Dade called them.
atair wrote: Thu Feb 14, 2019 2:01 am
regarding fireflies and averaging:
wouldn't a median filter be more suitable to avoid fireflies in low radius search?
Its a shot in the dark..
The problem about the median filter is that it is very slow, if I'm not mistaken.
) are under NDA and they aren't interiors to test)
( seem that 500.000.000 photton count/Max size + lookup radius = 5cm could give better cache) But this computer only have 8giga of ram (with 4go already used ).blender slowdown and crash when i test with such number 
