Cloud sculpting

Started by t0mr1ddle, January 18, 2021, 01:55:05 AM

Previous topic - Next topic


hi everyone,
I was searching for a way to describe in a deterministic way the cloud mass in order to place some "objects" in the sky side by side the "natural" clouds' formations.
the final density modulation seems to be the right way to send some masks to the cloud node.
The basic idea is to drive 3 density modulation masks along the 3 axis via transform input shader node.
the first step is to build the modulation masks. these are just the projections of a 3d object shaded with the height map (not just shadows!).
    it seems to be relevant having as much color dynamic as possible to maintain depth's details. (please find enclosed my assets as example)
the second step is to compose the projections in a volumetric space.
here the idea is to amplify the points where the three beams coming from the 3 different directions are intersecting.
this can be achieved via multiplication nodes (the multiplication function/network/logic is heavily influencing the quality)
I connected the netwok to cloud via cloud's fractal shader mask in order to have some more density parameter to use. 

the second line connected to the density shader is useful to add some features to the basic mass.

As mentioned by some of you, the set-up is tricky. (Maybe Matt can give us some easy to use node for that?)
The positioning of the masks and the set-up of right density and lighting is actually matter of try.
this is the reason why I share my starting scene with you, hoping your experience can help to improve.

Up to you to explore more (and please share results!).
enjoy ;)

Immagine 2021-01-17 231342.jpg


Thanks for sharing this. Haven't looked at the file yet, but I thought it would be something like this. The hardest is probably to get the camera's on the right positions around the cloud base, and I thought that there might be a way from inside the proposed cloud base, so 3 camera's would be in the same position at least with ortho projection, and some sort of fake vector displacement can be achieved. But something tells me that may not work, as it needs to be a negative somehow. Anyway; room for experimentation...


Cool, thanks a lot for sharing!!!