Experiments..

[dmn]

For the clouds:
I'll probably do something like:
you can have an arbitrary number of layers, for each layer you can specify
* the start altitude
* end altitude
* a list of Perlin noise (frequency, amplitude) pairs
* overall noise bias; controls sparseness of clouds

Wow. very decent looking clouds. I'm sure they will get better as people experiment with the settings.

Am I understanding correctly that your atmosphere simulation is actually wrapped around a sphere primitive, rather than being arbitrary like the ground plane? Could it be another primitive or polygon? Could you have multiple 'atmospheres' around different 'planets'?

How do you maintain the origin of the model in relation to the atmosphere (I'm guessing it's the same as ground plane).

[Kram1032]

I wonder, what's the progress, on that, currently...

[filippo]

very nice!!!
filippo

[Kram1032]

For the clouds:
I'll probably do something like:
you can have an arbitrary number of layers, for each layer you can specify
* the start altitude
* end altitude
* a list of Perlin noise (frequency, amplitude) pairs
* overall noise bias; controls sparseness of clouds

Wow. very decent looking clouds. I'm sure they will get better as people experiment with the settings.

Am I understanding correctly that your atmosphere simulation is actually wrapped around a sphere primitive, rather than being arbitrary like the ground plane? Could it be another primitive or polygon? Could you have multiple 'atmospheres' around different 'planets'?

How do you maintain the origin of the model in relation to the atmosphere (I'm guessing it's the same as ground plane).

First of all, the clouds aren't currently available. Only the atmospheric scattering.

2nd: The atmosphere is supposed to represent a real atmosphere as close as possible and as such one, you also could make universe shots, which (besides some other physical effects) males it necessary to use a sphere primitive

The multiple-atmosphere part *should* work, as the atmosphere is a material, but you can't really tweak much with it - most stuff is hardcoded. You couldn't fake a Mars- or Venus-atmosphere with it

the origin of the atmosphere is the same as the origin of the sphere, as it's a sphere, itself
In total, you have three layers (actually still quite a low detailed model):
ground sphere
lower atmo - sphere (without clouds)
higher atmo - sphere (with clouds, which you can't use, currently, as I already said)

Ono said, he's gonna go on working on that, when he's got his new quad-core but since he said that, the topic kinda died

He didn't say a word about that, since a long time, now - not even "I didn't get into it, yet" or "I wont develop it any further" or "I'm just working on it, right now" or anything like that....

[zsouthboy]

Ono, I've been meaning to ask you about this:

Isn't it faster to simply render an .exr of your desired sky, then use it in the desired scene?

vs.

Clouds with scattering material, atmospheric simulation, IN ADDITION to your scene?

It's totally cool, but my guess (especially since rays just get terminated at the .exr map) is that the former is much faster.

And, when you think about it - it's pretty much correct for most scenes. How many scenes are actually going to influence the lighting of the clouds (vs. simply returning the light from them)?

[Kram1032]

lol
zsouthbody, you can't know it, but months ago, before this *really* started, I chatted with Ono in the Indigochat and talked with him about this idea.
He told me, exr is far faster, which is true, obviously, but I (not sure, if *only* I...) asked for this, because of several interesting cases, which you can't really get with an exr map.

For example, you can make a shot from the universe down the planet (this sky simulation is (or was?) also supposed to once support stars and different sun - earth - moon constellations, like full and semi eclipses.

Or what if one once models a full featured air plane, that flies?
Sure, you could use an exr map for it, but there wont be too many exr maps from very high in the air... - and what's about parts of clouds, which might cross the air plane.

In even farther future, where Indigo will also have particle- and volume emitter support (now, that I mention it: what's about the particles, Ono?), there might also be a limited physical weather simulation (Indigo'd have a very complex but good GUI, 'till then, else, it wouldn't be tweakable, I guess), like snow, rain and thunderstorm (and tornado with randomly created stuff and cows, flying around )

Especially the last part might never happen. - Except Ono is once thinking of making a full featured physics prog out of the current physics-of-light-renderer-only

[StompinTom]

buuuuump. whatever happened to this thread and all those awesome experiments Ono was doin?

it seems this forum has gotten much quieter in the past bit.... hopefully that will change come the fall.

[PureSpider]

Yes what happened to it?!
Those experiments look awesome!

[CTZn]

Wow looks very interesting!

I'm greatly anxious to see more about it!

And I hope it'll find its way in one of the next releases soon.

[Deus]

Yeah give us fricking clouds already!

[CTZn]

Can we blend medium and null yet ? If yes then clouds could be an option already isn't it ?

[Jambert]

These tests are very impressive,

up

[neo0.]

Sorry, if this sounds noobish, but are they volumetric?

[CTZn]

These clouds ? Sure they are

[dmn]

buuuuump.

*cough*