dark juice

Use this forum for general user support and related questions.
Forum rules
Please upload a testscene that allows developers to reproduce the problem, and attach some images.
Post Reply
Fox
Posts: 437
Joined: Sat Mar 31, 2018 11:17 am

Re: dark juice

Post by Fox »

@ lacilaci
The asymmetry 0 does not make caustics at all (it's like candle wax). Juice is around 0.9
The values in page 7 "Orange Powder" work pretty well http://graphics.ucsd.edu/~henrik/papers ... erties.pdf
User avatar
lacilaci
Donor
Donor
Posts: 1969
Joined: Fri May 04, 2018 5:16 am

Re: dark juice

Post by lacilaci »

Fox wrote: Fri Aug 16, 2019 1:18 pm @ lacilaci
The asymmetry 0 does not make caustics at all (it's like candle wax). Juice is around 0.9
The values in page 7 "Orange Powder" work pretty well http://graphics.ucsd.edu/~henrik/papers ... erties.pdf
welp... I wouldn't drink that
asymmetry09.png
Anyway it doesn't matter, cause luxcore won't show caustics through refractions cause that's sds path:
PTLT.png
and without lighttracing it won't let any light through glass:
PT.png
There's no fixing this, until Dade brings new features... Transparent shadows, vertex merging, ability to hide objects from certain types rays(like lightpath node in cycles)
provisory
Posts: 235
Joined: Wed Aug 01, 2018 4:26 pm

Re: dark juice

Post by provisory »

I just realized that path depth is the most important part with this juice case.

Below are renders with Path / Sobol (without light tracing and any other "tricks").
The settings are the same, except path depths (total, diffuse, glossy and specular depths are set the same per render).
Render times were also the same.
(Illuminated by built in Sun+sky)

Path depths: 8
path-depth-8.jpg

Path depths: 16
path-depth-16.jpg

Path depths: 64
path-depth-64.jpg

The volume settings here were:
volume-settings-for-path-test.png
volume-settings-for-path-test.png (5.2 KiB) Viewed 3546 times
provisory
Posts: 235
Joined: Wed Aug 01, 2018 4:26 pm

Re: dark juice

Post by provisory »

It looks very useful, but to be honest I didn't even try to understand fully.
Do you have a formula, that can convert these values to LuxCore volume settings (and not just asymmetry but absorption and scattering too)?
User avatar
lacilaci
Donor
Donor
Posts: 1969
Joined: Fri May 04, 2018 5:16 am

Re: dark juice

Post by lacilaci »

Mate, the fact that you start seeing juice volume color after so many path depth increases should be a clear hint to you that this is GI light not direct light, which is the core problem of this...problem.

Also lighttracing isn't actually a trick, it's what bidir does too.
epilectrolytics
Donor
Donor
Posts: 790
Joined: Thu Oct 04, 2018 6:06 am

Re: dark juice

Post by epilectrolytics »

provisory wrote: Sat Aug 17, 2019 12:19 pm I just realized that path depth is the most important part with this juice case.
When I tried your scene with path sobol (path depth 30 all types) it was very confusing to see it were the diffuse bounces doing the trick.
Apparently multiscattering in a volume is constrained by diffuse path depth!

I'm convinced LuxCore is doing the volume thing physically correct but that means tracing billions of bouncing paths in a dense scattering medium, together with the very low efficiency of SDS resolving with BiDir/Path+Metro it is taking its toll on render time.
(There are lots of papers done about doing efficient ray marching, probably addressing this very problem.)

Using a not volume based surface-scattering approach as with Cycles' principled shader is really advantageous in this case.
provisory
Posts: 235
Joined: Wed Aug 01, 2018 4:26 pm

Re: dark juice

Post by provisory »

epilectrolytics wrote: Sat Aug 17, 2019 12:58 pm ...
I'm convinced LuxCore is doing the volume thing physically correct but that means tracing billions of bouncing paths in a dense scattering medium, together with the very low efficiency of SDS resolving with BiDir/Path+Metro it is taking its toll on render time.
...
Using a not volume based surface-scattering approach as with Cycles' principled shader is really advantageous in this case.
I absolutely agree, however I hope, this "brute force" approach stays in LuxCore - as a reference solution - after it will have a faster method.
Fox
Posts: 437
Joined: Sat Mar 31, 2018 11:17 am

Re: dark juice

Post by Fox »

provisory wrote: Sat Aug 17, 2019 12:27 pm
It looks very useful, but to be honest I didn't even try to understand fully.
Do you have a formula, that can convert these values to LuxCore volume settings (and not just asymmetry but absorption and scattering too)?
I can convert extinction coefficient to transmittance, but the scale of things is unclear to me.
https://chem.libretexts.org/Bookshelves ... ambert_Law

In volume the dilution ratio is about 390 (4 spoons powder or 59 ml diluted in 23 l water), but 390 seems too little for scattering, and too much for absorption.

Cube root from 0.059 l is 0.389, and from 23 l is 2.844
2.844 / 0.389 = in linear direction the scale is 7,31 . That also is too little for absorption.

I'm so bad in math :oops: maybe it's related to (×10−2 mm−1), does it mean multiply by 0.01 for 0.1 mm or with 100 for 1m :roll:

I went this way by scaling scale 390 up x times for scattering in tandem i made absorption stronger same x times
SSS scale 2.5 x 390 = 975
ABS scale 2.5 x 7,31 = 18,275 maybe then 1/18,275 so absorption at depth about 0.05 meters

I played with G values (originals 0.919 0.918 0.922), scaled down by 0.9 and lost almost all light transmission, then scaled up by 1.01 and got more transmission.
provisory
Posts: 235
Joined: Wed Aug 01, 2018 4:26 pm

Re: dark juice

Post by provisory »

Thank you Fox!
I'll try to play with it when I find some time.
Fox
Posts: 437
Joined: Sat Mar 31, 2018 11:17 am

Re: dark juice

Post by Fox »

provisory wrote: Sat Aug 17, 2019 2:26 pm Thank you Fox!
I'll try to play with it when I find some time.
The converted orange juice values in old Indigo manual are very low, i guess it's lemonade like fanta.
Post Reply