Page 5 of 11

Re: playing with caustics

Posted: Thu Jan 24, 2019 8:51 pm
by epilectrolytics
Thanks guys! :D
FarbigeWelt wrote: Wed Jan 23, 2019 10:06 pm I would add some water below this ice or a fine, small layer of ice here and there as if the wind blew snow over the ground and in some grooves it stuck.
I'd love to do that but mixing snow, ice and ground materials in a convincing way is yet beyond my abilities I fear.
I tried an area with more reflection for a wet look but those bright reflections started to compete with the ice, diminishing the aesthetic effect.
Initially I wanted to put something on a rock and chose a PBR material with displacement map, only it looks like dirt instead, maybe because of an adequate PBR shader missing or the low sun position.
FarbigeWelt wrote: Thu Jan 24, 2019 12:12 pm I‘ve planned to setup example scenes and write a short what-to-consider for CPU BiDir, Metropolis and Filters settings to get quite fast renders for caustics and/or scattering volumes, with a reasonable noise.
That will come handy, looking forward to it!

________
cleandirt.jpg
(final version)

Indirect + caustic caches

Posted: Wed Mar 13, 2019 11:40 pm
by epilectrolytics
Not denoised.
bunker6b.jpg

Re: playing with caustics

Posted: Thu Mar 14, 2019 7:48 am
by Sharlybg
How many time for this one on wich computer ?

Re: playing with caustics

Posted: Thu Mar 14, 2019 8:10 am
by epilectrolytics
It took more than an hour (sorry I don't recall the exact time) on my PC with PGI OCL on RTX 2070 + GTX 1060.
Caustic cache is much bigger than indirect and slows down rendering to 1/3 or less of indirect only speed.
Figuring out the parameters is awful too: In debug mode it took always 8 min to build the cache and I needed more than 10 times to find working settings.
All in all in its current state it is too difficult to use IMHO.
Some progressive version of the photon pass converging to a specified density would be nice but I don't know if this is feasible.

Spectral volumetric Psor cube animation

Posted: Sun May 19, 2019 1:06 pm
by epilectrolytics
epilectrolytics wrote: Sat Dec 29, 2018 8:13 pm So here is a quick and dirty dispersion render:
Here's a hires attempt with OIDN, still a bit noisy.
Rendered at 720p á 3500 samples (42min/frame) and blown up with DaVinci Resolve.

Re: Spectral volumetric Psor cube animation

Posted: Sun May 19, 2019 6:18 pm
by FarbigeWelt
epilectrolytics wrote: Sun May 19, 2019 1:06 pm
epilectrolytics wrote: Sat Dec 29, 2018 8:13 pm So here is a quick and dirty dispersion render:
Here's a hires attempt with OIDN, still a bit noisy.
Rendered at 720p á 3500 samples (42min/frame) and blown up with DaVinci Resolve.
This is a very nice one. How to you chose an optimal glass body for your caustic images?

Re: playing with caustics

Posted: Sun May 19, 2019 7:51 pm
by epilectrolytics
Thanks!
This is not my modelling but the famous Psor-cube from the old Luxrender forum, which I tested in a volumetric situation.
In my opinion caustics are most effective in minimalistic environments without distracting geometry, textures, colors and such.
There even simple bodies produce amazing effects.
So I have started here with simple rounded cubes, toroi or spheres.

I was planning to proceed to more complex shapes, but with BiDir lacking SDS reflections and PGI being very problematic with small merge radii I'm kinda stuck now.

Re: Indirect + caustic caches

Posted: Sat Jun 08, 2019 11:52 pm
by madcat117
epilectrolytics wrote: Wed Mar 13, 2019 11:40 pm Not denoised.
bunker6b.jpg


Mother of....

BiDir vs periodic cache

Posted: Sun Jun 23, 2019 12:17 pm
by epilectrolytics
bidir.jpg
sppm.jpg
.
Rendered ca 45min.
Both struggle with the caustics behind the big ring (SDS-paths):
BiDir still noisy there after 2000 samples,
SPPM shows visibility particle pattern
(1200 samples, step 3, radius 5mm).

BiDir renders darker caustics,
SPPM more highlights in the glass
(bloom already reduced from 25 to 10).

Re: playing with caustics

Posted: Sun Jun 23, 2019 2:40 pm
by provisory
Nice!

Maybe tests don't need bloom at all.

Have you experimented with PGI caustic cache and volumes? I mean the light rays in the fog/smoke type? I can't get good results.