Thought I'd hare this here:
https://elementacular.com (https://elementacular.com)
Currently it's only available for Maya (renders with Mental Ray and Arnold). The neat thing is how they approach the modeling of clouds and that they fully support OpenVDB.
Looks like OVDB is the main new non-app-specific feature in 1.5:
https://elementacular.com/content/elementacular-1-5-supports-rendering-in-arnold-and-renderman-through-openvdb
The render on that page, is that minutes and seconds, or seconds and milliseconds?? It has to be the latter, right?
- Oshyan
I'd say it's hard to believe these are minutes...41 seconds probably, Arnold should be quite fast with that. But they really need to specify this along with that number.
I found that the way the cloud modeling is really interesting, you could automate a lot of this to produce all kinds of cloud forms while keeping control of the details.
I agree, Arnold should be faster than that, especially for the relatively unrealistic shading/scattering. That's why I mentioned it.
The only video I've seen of their cloud modeling just looks like a bunch of spheres that act sort of like metablobs, and I saw that back in 2014. Is that what you're referring to?
- Oshyan
Yes. I wonder how they create the volume data but you probably know more ;)
Meta-blobs can be done in TG quite easily. You define a point then measure the length away from that point, convert that to a gradient (linear step), then use a PF or noise combination to alter the black level in another linear step.
What would be useful is a visual cue to realise the volume around that point when constructing your shape out lots of meta-balls. I've been thinking that a particle system could also use this approach.
?????
Quote from: Hetzen on December 17, 2016, 08:08:05 AM
Meta-blobs can be done in TG quite easily. You define a point then measure the length away from that point, convert that to a gradient (linear step), then use a PF or noise combination to alter the black level in another linear step.
What would be useful is a visual cue to realise the volume around that point when constructing your shape out lots of meta-balls. I've been thinking that a particle system could also use this approach.
This is incredibly cool, Jon!!!
Here's another render with three blobs stuck together. Took just over a minute to render on default settings, just AA=4.
[attachimg=1]
And this one took 6 mins with 300 million voxels, Cloud Quality of 0.7, AA=6. With slightly different positions and sizes for the blobs.
[attachimg=1]
And another. Same settings as above, different POV to fill the frame, with slightly different blob positions. This took just over 9 mins to render. This and the above two renders use the same PF, just different blob scales and positions.
[attachimg=1]
Well dang Jon, that's cool stuff! Need to look into and think about this more...
- Oshyan
Those are some interesting cloud experiments.
Here are some older threads concerning Metacloud experiments.
http://forums.planetside.co.uk/index.php?topic=3691.msg50489#msg50489
One example in the File Sharing thread.
http://www.planetside.co.uk/forums/index.php/topic,3691.150.html
Cool, Jon. Off to experiment right away, but I can't get it to work, didn't find a way to change the blobs location. Is this the way you set this up somehow?
Linear step 01 delivers an output at all places between 0 and 1
So it means no sense to have a limit greater than one on linear step 02.
Ulco, I've sent this set up over to Oshyan to have a look at, so perhaps I shouldn't give the details away at the moment. It's quite different to those examples AP. When I re-wrote this yesterday I realised I only needed one Linear step in the graph. And Mogn I stuck a conditional in there :P
I figured as much after I studied more based on what this was about. I was tempted to run some experiments of my own, however not directly related to blobs but if anyone wants to look at these I can post the files later after a few more are finished.
@ Jon; no problem.
@ AP; those are pretty cool. There are clouds like those. I guess you can use a sinus function or some kind of PF multiplied by altitude ranges.
Just did a quick test with localised clouds as blobs. Like that you can roughly see where the blobs are.
I guess this won't work with easy clouds, but with V2 or V3 it does.
There's some nice work there AP. Looks like you're using the altitude to drive a perlin in some of those. They remind me of a Battlestar Galactica - Cylon Basestar.
@Hannes: That's how DandelO did his I think.
Quote from: Hetzen on December 18, 2016, 11:54:08 AM
@Hannes: That's how DandelO did his I think.
Ah, I have to confess, that I haven't read all the postings... :-[
Thanks, Jon!
Not exactly an altitude to drive the noise but function nodes, however I intend to use altitude driven ideas to attempt to create a possible anvil top for cumulonimbus clouds but the starting idea there is to attempt the more tall primary cloud patterns and tight secondary cauliflower-like patterns which make up the cloud convection effect before flattening in the troposphere. Right now, what you all see is a general idea for mesocyclone super cell storm systems, hence the Cylon Basestar look. I just need to have the tight billowy patterns to grab all over the primary cloud. No success as of current.
A few more.
Quick question. Is there a way to limit any effect using a surface layers altitude for example with the clouds? I had a Surface layer attached to the cloud density on top of everything else using another noise as it's child but no change what so ever.
I sometimes used a distribution shader with altitude settings to influence clouds, used as a mask. If you use a few and have them mask fractals, you can get different fractals at different altitudes within one cloud. Or use a color adjust to alter the same fractal but at different altitudes.
I tried that and it did not work.
Does...
It appears I overlooked some things. Here is the result.
Wow, you are getting some results that are very close to huge "anvil" thunderstorm clouds. Keep it up! And share when you nail it. ;)
- Oshyan
Indeed, I will. There is some more testing to be had. I should start a new post based on these experiments as this diverged from the original post subject.