PDA

View Full Version : Rayleigh VS Mie scattering for volumetrics such as hypervoxels/clouds



prometheus
02-15-2015, 09:27 AM
Is hypervoxels in need of a new illumination model?

As I understand it, rayleigh scattering model fits well for air scattering and some gas perhaps, but for clouds? Isnīt mie scattering algorithms better suited.

Ivé been thinking about that since I heard that modo is supposed to use mie scattering in itīs illumination model, and as far as I have seen, the results looks better for volumetric clouds.
Modo has a better blending between particles where the density between them seem to fuse together much better than hypervoxels, I do not know if that has anything to do with the mie illumination algorithms, or something else?
I wonder which of rayleigh and mie is the best for clouds, rayleigh might be best for gas,nebula,explosions, where mie might be more suitable for clouds?

Descriptions of the different types of scattering...
http://www.severewx.com/Radiation/scattering.html

Sky Radiation..
http://en.wikipedia.org/wiki/Diffuse_sky_radiation

ABout mie..
http://en.wikipedia.org/wiki/Mie_scattering

About rayleigh
http://en.wikipedia.org/wiki/Rayleigh_scattering

http://apollo.lsc.vsc.edu/classes/met130/notes/chapter19/graphics/table_rayleigh_mie.jpg



Michael

prometheus
02-21-2015, 08:04 AM
Not much of action going on with this, probably lack of interest...or no one knows a squat about it.

Michael

jasonwestmas
02-21-2015, 09:00 AM
Yes, I would say that hypervoxels needs a lot of new features and probably new shaders. Yes, I think you answered your own questions. :)

erikals
02-21-2015, 10:25 AM
...or no one knows a squat about it
+1 here on that one... :° :]

prometheus
02-21-2015, 02:33 PM
Reckon I have to post modo cloud samples to show the blending difference between particles in modo voxels VS lightwav e hypervoxels, would be interesting to know though if it is because of that Mie illumination method or something else in the volumetrics that makes modo voxels blend much better?

the old dynamite plugin for lightwave had better blending than lightwaveīs native voxels, but it also came with a price to be slower to calculate, I donīt experience any additional slowness in modo with itīs volumetrics.

I need to speed up my own hypervoxel dummie and post here soon, with suggested improvements..particle handling with multiple particle emitters is one addition I would like to see, then we could use several particle emitter cloud planes, or use several different scaled and positioned emitters in the scene, using the same hypervoxel shading instance, that is not possible today, and adding them one by one in the hypervoxel panel would be killing lightwave in render times with so many hypervoxels instances.

Volumetric item (geometry based) is also something lightwave need to catch up on.

Michael

m.d.
02-21-2015, 05:48 PM
here's a paper on using some modified mie algorithm for near realtime cloud rendering

http://www-evasion.imag.fr/Publications/2008/BNMBC08/clouds.pdf