Planetside Software Forums

General => Terragen Discussion => Topic started by: bigben on January 18, 2007, 05:54:07 PM

Title: Cloud distribution tests (Hurricane Katrina)
Post by: bigben on January 18, 2007, 05:54:07 PM
Hi All

A tip and a test...

First the tip... Rendering clouds in the preview window takes a long time so I came up with an alternative way to see where the clouds are.

Rather then preview the clouds, I set up a coloured surface with the fractal breakup set to the density fractal for the cloud layer in question. This makes positioning the camera between clouds a lot easier and also makes for a quick, accurate preview of the cloud distribution.

[attachthumb=#2]
[attachurl=#1]

I've also figured out a way to use image masks to help control the distribution of clouds.
[attachurl=#4] (No image included)

This works pretty well with masks generated from IR satellite weather images, although finding raw images seems to be getting harder these days. Everyone seems to be adding map overlays to them.  Here's a quick test render. The cloud settings need a lot of work but you get the idea.
[attachthumb=#3]
[attachthumb=#5]
Yes it's Katrina
4 cloud layers, 5km thick cumulus, 3km thick cumulus, 300m thick altocumulus and 10m thick cirrus layers. I'll add some closer renders as they finish. Masks deriverd from the high res link on http://en.wikipedia.org/wiki/Image:Hurricane_Katrina_August_28_2005_NASA.jpg (http://en.wikipedia.org/wiki/Image:Hurricane_Katrina_August_28_2005_NASA.jpg)

It also works quite well with low res masks as the density fractal help to disguise the borders of the mask.
Title: Re: Cloud distribution tests
Post by: Will on January 18, 2007, 05:56:52 PM
Good tip and neat pictures,  is intresting that you can but IR weather maps in I should try that ;)
Thank man!

Regards,

Will
Title: Re: Cloud distribution tests
Post by: MeltingIce on January 18, 2007, 06:29:53 PM
Very cool, how about a way to smooth out the clouds now?
Title: Re: Cloud distribution tests
Post by: bigben on January 18, 2007, 06:40:04 PM
Quote from: MeltingIce on January 18, 2007, 06:29:53 PM
Very cool, how about a way to smooth out the clouds now?

That'll be in my cloud settings, probably in the density fractal scaling. This may not be as much of a problem in the registered versions as they will have additional rendering options which may reduce the noise.  I have another render on the way (another hour or so to go) much closer to the eye of the hurricane which isn't as bad. For animations it would appear to be necessary to also animate the density fractal settings of clouds to allow for large changes in camera altitude... but then I'm still learning  ;)
Title: Re: Cloud distribution tests
Post by: Will on January 18, 2007, 06:45:53 PM
Keep us posted on anything new you find out.

Regards,

Will
Title: Re: Cloud distribution tests (Hurricane Katrina)
Post by: Oshyan on January 18, 2007, 11:44:33 PM
Very cool. Definitely post the higher quality versions when done rendering!

- Oshyan
Title: Re: Cloud distribution tests (Hurricane Katrina)
Post by: bigben on January 22, 2007, 06:55:10 PM
My initial masking wasn't really correct, it just looked like it was doing the right thing. I went back to the drawing board to nut out a real subtractive image mask. The first problem was to get around the norml cloud distribution outside of the image mask:

In theory this should have been enough, however in practise I found that there was enough noise in the area outside of the local image mask to put scattered clouds over the rest of the terrain. For the global white shader I tried both a white image mask and a white shader, both with the same result. To get around this I repeated the image mask subtraction.

Here is a clip file of the cloud mask. Edit the image mask with your own image.
[attachurl=#1]

I then built up multiple cloud layers with this clip file, tidying it up by using just the one global white shader.
[attachthumb=#2]

The cloud settings still need work but it's an improvement.
[attachthumb=#3]

Looking more closely at the image I was basing my masks on, it was obvious that some manual editing was necessary to overcome some dense shadows on some of the taller, billowing cumulus clouds. It's possibly easier to use lower resolution mask images and simulate the rest with the cloud settings (which is a good thing as it saves your RAM for surface image masks)
Title: Re: Cloud distribution tests (Hurricane Katrina)
Post by: littlecannon on January 23, 2007, 05:05:07 PM
Nice ideas and renders bigben... what would be a cool feature would be to have a Twirl function for the cloud fractals, then we could create our own worldly weather chaos, eh guys... ;)
Title: Re: Cloud distribution tests (Hurricane Katrina)
Post by: Superza on January 23, 2007, 06:32:27 PM
Great Work and tips!
Thank you!