Rayleigh and/or Wie Atmospheric Scattering

Started by aptyler, January 09, 2018, 06:57:17 AM

Previous topic - Next topic

0 Members and 1 Guest are viewing this topic.

aptyler

I often use KeyShot to render Sci-fi space scenes, and one thing that often enhances the image aesthetic is to have an alien-looking planet in the background. I have plenty of meshes and textures with which to create these planets, but they never end up looking particularly realistic due to their lack of atmospheric scattering. I've tried using the motion blur trick to try to simulate the volumetric behavior of the atmosphere, but it doesn't work well with spherical objects, and doesn't account for the the change in density as a function of altitude.

Any thoughts?

KeyShot

In KeyShot 7 you can use the cloudy plastic material to simulate a homogeneous scattering medium. You may be able to approximate the height variation using a textured cloudiness parameter - at least this can be used to increase the scattering density. You can also add some color to the scattered light to approximate the effect of Rayleigh scattering (where blue light scatterings 16 times as frequently as red light). Another option if you want to explore the result from an artistic point of view could be to use the xray material.

We are looking to support genereal heterogeneous scattering media in KeyShot 8.

Both cloudy plastic and the upcoming scattering media support directional scattering via the Henyey-Greenstein phase function (under advanced settings in cloudy plastic). Henyey-Greenstein is a commonly used phase function, and it is perfectly suited for simulating atmospheric aerosols. Mie (or more precisely Lorenz-Mie) scattering is a general solution to scattering from a perfectly smooth homogeneous sphere. It is a very complex phase function and I do not think any rendering programs use the actual Lorenz-Mie phase function. I co-authored a paper ten years ago that investigated the use of Lorenz-Mie theory for predicting the appearance of scattering materials:
https://dl.acm.org/citation.cfm?id=1276452
While the results were great the images took forever to render. The paper has one image of a few glasses of milk and it took over 12,000 cpu hours to render (in 2007).

aptyler

Thanks for the tip. The cloudy plastic sort of works, but with a few unexpected behaviors. Was very surprised at the results in the second pic. The scattering must have been so good that it completely lit the back side of the planet. After adding the caustic lighting setting, the results were maybe a bit more realistic, but for some reason the tessellation of the planet was more pronounced along the day/night terminator. I'm going to have to keep tweaking settings. I'd really like to see if I can't get something to work without the caustics, due to the processing time, and its negative impact on other areas of the render.


KeyShot

Would you be able to share the scene? You should be seeing a shadow without having to include caustics?
If you can send it or a link to info@luxion.com then we will take a look?

Will Gibbons

Quote from: aptyler on January 10, 2018, 12:41:48 PM
Thanks for the tip. The cloudy plastic sort of works, but with a few unexpected behaviors. Was very surprised at the results in the second pic. The scattering must have been so good that it completely lit the back side of the planet. After adding the caustic lighting setting, the results were maybe a bit more realistic, but for some reason the tessellation of the planet was more pronounced along the day/night terminator. I'm going to have to keep tweaking settings. I'd really like to see if I can't get something to work without the caustics, due to the processing time, and its negative impact on other areas of the render.

Just wanted to say thanks for including images based on your findings and hopefully you're able to find a solution that works well. This is interesting and I haven't seen many planets rendered with KeyShot. Super-cool!