B.Y.O.B. wrote: Tue Feb 05, 2019 9:06 pm
Not yet.
I think the plan was to make this mostly automatic, determined by roughness - is that already implemented?
No idea - how will automatic roughness work with roughness maps (think PBR material)?
Other question: In Luxcore there is a choice of random or metropolis for photon cast - is metropolis the default in BlendLuxCore?
epilectrolytics wrote: Tue Feb 05, 2019 8:59 pm
Dade added some option to exclude glossy materials from indirect cache, is that exposed somwhere?
Not yet.
I think the plan was to make this mostly automatic, determined by roughness - is that already implemented?
I reverted the plan at the end and only implemented an On/Off flag because glossy roughness can be textured, so it is variable and pretty much unknown (so I can not implement an automatic setting). So, yup, the material flag should be exposed.
Support LuxCoreRender project with salts and bounties
epilectrolytics wrote: Tue Feb 05, 2019 8:53 pm
First results are quite sobering, caustic cache doesn't like environmental light:
It is more the usual problem of power balance between sky (very low) and sun (very high): it looks like photons are fired mostly from the sun. Have you tried alternative light strategies ?
Support LuxCoreRender project with salts and bounties
Could the lookup radius be calculated from normal pass in screen space? So, large areas would have larger lookup radius and when approaching edges it would be smaller, maybe at the very edge no photons just pathtracing? Is this a stupid idea?
Or maybe just somehow force more acurate calculation on the edges and corners...
lacilaci wrote: Wed Feb 06, 2019 5:53 am
Could the lookup radius be calculated from normal pass in screen space? So, large areas would have larger lookup radius and when approaching edges it would be smaller, maybe at the very edge no photons just pathtracing? Is this a stupid idea?
Or maybe just somehow force more acurate calculation on the edges and corners...
It requires ray differentials and it would be the optimal solution but, unfortunately, we don't have ray differentials.
Support LuxCoreRender project with salts and bounties
Dade wrote: Tue Feb 05, 2019 11:40 pm
It is more the usual problem of power balance between sky (very low) and sun (very high): it looks like photons are fired mostly from the sun. Have you tried alternative light strategies ?
Changing light strategy or importance doesn't change anything.
It looks like the sky is excluded from direct sampling.
It looked normal before you fixed the fireflies (direct hits) for infinite lights.
Here's a file:
I'd like to suggest that the debug option showing the photon map is removed and instead it is always contained as a photon AOV when PhotonGI is activated so that it can be toggled as a channel during the render for inspection.
Also it would be nice to have a percentage count in Blender view during cache building.
caustictest7_LuxCore.zip
