Page 6 of 9
Re: LuxCore v2.3 proposed features
Posted: Wed Oct 02, 2019 5:07 pm
by provisory
Sharlybg wrote: ↑Wed Oct 02, 2019 9:39 am
If Bidir have to exist it need more love...
BiDir must exist (and needs solution for SDS) to be the full featured "holistic" reference renderer.
Path Tracing with Light Tracing and PGI caches is good for the most scenes, but it's overly complicated and still doesn't cover complex cases, like when the full scene is illuminated through real glass.
Re: LuxCore v2.3 proposed features
Posted: Wed Oct 02, 2019 5:39 pm
by epilectrolytics
provisory wrote: ↑Wed Oct 02, 2019 5:07 pm
BiDir must exist (and needs solution for SDS) to be the full featured "holistic" reference renderer.
Yep, and that would be:
Dade wrote: ↑Mon Sep 30, 2019 10:21 am
- New complete Bidirectional path tracing with vertex merging (BIDIRCPUVM)
BiDirVM currently is an option in LuxCoreUI but it is broken (though @Fox has managed to get some great renders out of it).
I would also like to see an LightCPUVM option (like that already realised with PSR) but I'm afraid I'm the only one
)
Re: LuxCore v2.3 proposed features
Posted: Wed Oct 02, 2019 11:17 pm
by FarbigeWelt
Due to my intense experience trying to render orange juice in a glass I suggest an additional volume type.
Custom Volume
Basic parameters like heterogeneous with following differences
Scattering size (particle size, length on axis)
Scattering density (distance between particles, 1 means distance equals particle size)
Scattering color / intensity (RGB, 0,0,0 passes all, 1,1,1 scatters homogeneous in particle volume).
Scattering density, size and color / intensity are related to x, y and z; they all accept 3D texture input or a function (linear, quadratic, cubic) respectively free hand drawn curve with only one value per axis point (value 0..1, axis 0..255). Input default is 1. Input is multiplied with its related scattering value.
Particles can be cubes or Rhombic dodecahedrons (RhDdh). Like cubes RhDdh can tesselate space gapless.
Possible uses for custom volumes are: clouds, fog, sand clouds, fog, sedimenting inhomogeneous mixtures, fruit juice, milk, turbid lakes
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 12:21 am
by Fox
epilectrolytics wrote: ↑Wed Oct 02, 2019 5:39 pm
BiDirVM currently is an option in LuxCoreUI but it is broken
The MLT support for VM would be insanely huge upgrade. Now even with classic BiDir, the MLT beats random in underwater caustics so fast, i can't even blink my eyes before caustics are "all done".
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 1:30 pm
by Dade
FarbigeWelt wrote: ↑Wed Oct 02, 2019 11:17 pm
Due to my intense experience trying to render orange juice in a glass I suggest an additional volume type.
Custom Volume
Basic parameters like heterogeneous with following differences
Scattering size (particle size, length on axis)
Scattering density (distance between particles, 1 means distance equals particle size)
Scattering color / intensity (RGB, 0,0,0 passes all, 1,1,1 scatters homogeneous in particle volume).
Scattering density, size and color / intensity are related to x, y and z; they all accept 3D texture input or a function (linear, quadratic, cubic) respectively free hand drawn curve with only one value per axis point (value 0..1, axis 0..255). Input default is 1. Input is multiplied with its related scattering value.
Particles can be cubes or Rhombic dodecahedrons (RhDdh). Like cubes RhDdh can tesselate space gapless.
Possible uses for custom volumes are: clouds, fog, sand clouds, fog, sedimenting inhomogeneous mixtures, fruit juice, milk, turbid lakes
I added this as "Add procedural textures specifically tailored for heterogeneous volumes (for instance clouds, fog, sand clouds, fog, sedimenting heterogeneous mixtures, fruit juice, milk, turbid lakes)".
All is already possible using LuxCore OpenVDB support but without procedural textures, you are limited by Blender/BlendLuxCore capabilities to edit/export heterogeneous volumes. Procedural textures have also a very small (GPU) memory foot print while OpenVDB volumes require a lot of ram.
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 1:30 pm
by Dade
Updated the original list and going to create the poll.
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 2:34 pm
by kintuX
What Glass material really needs is multi-scattered surface model, one like Cycles has. Here, even Corona ain't as good
Next, it would be great to really fully fix the shadow terminator. Should be pretty simple, like what appleseed's dev. Franz did.
And of course, most important is resulting render consistency between BlendLuxCore and LuxCoreUI!
Being able to use network rendering should work without hassle.
Oh, finally better sampling. IDK what it is, but the way it is now, image quality is far less then that of Corona. Just can't reach such
sharpness & clarity. In this regard, even faster GPU rendering doesn't help. Corona wins.
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 2:36 pm
by marcatore
kintuX wrote: ↑Thu Oct 03, 2019 2:34 pm
And of course, most important is resultind render consistency between BlendLuxCore and LuxCoreUI, being able to use network rendering without hassle.
I'm quite happy that someone else feel this as important.
I hope that soon (2.3) or later this will be fixed.
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 2:44 pm
by kintuX
marcatore wrote: ↑Thu Oct 03, 2019 2:36 pm
kintuX wrote: ↑Thu Oct 03, 2019 2:34 pm
And of course, most important is resultind render consistency between BlendLuxCore and LuxCoreUI, being able to use network rendering without hassle.
I'm quite happy that someone else feel this as important.
I hope that soon (2.3) or later this will be fixed.
This shouldn't be hard. We need to find a way to bypass Blender's disfiguration of colour.
PS
I believe added option to also support direct export of rendered image from LuxCore (as EXR or optionally other formats / like Corona exporter supports) could be the way.
Re: LuxCore v2.3 proposed features
Posted: Thu Oct 03, 2019 4:18 pm
by Dade
kintuX wrote: ↑Thu Oct 03, 2019 2:34 pm
What Glass material really needs is multi-scattered surface model, one like Cycles has. Here, even Corona ain't as good
It looks like LuxCore roughglass material
