Page 1 of 4
Posted: Mon Apr 21, 2014 4:12 pm
is hard, like PhD research topic hard.
It's not even easy to do one kind of cloud :/
Links for reading:
- this ones not practical for us but it is pretty
- this is a collection of links and further reading
- Niniane Wang - Fast and efficient cloud rendering (PDF)
NB: if you do checkout my "clouds" branch (coders only) then just load up the modelviewer with "-mv wave" or something and you'll get some clouds rendered, zoom out and you'll see the ones that are billboarded, it's a straight copy of the code from the Mark Harris 2003 source code and I won't be using it, this was just for research.
Posted: Sat Jul 26, 2014 6:40 pm
I have no idea if this is of interest to you. I just noticed that there were some files "GLCloud1", "GLCloud2" etc.
Aweful screenshot, of not a single cloud, but this is part of what the code can do apparently:
Posted: Mon Jul 28, 2014 9:17 am
The Intel samples
look interesting too and have GitHub sources
Posted: Mon Jul 28, 2014 10:46 am
I think the Intel thing is specific to their hardware... I don't know if there's much chance of translating it to work on anything else. The web page says "demonstrates high quality cloud rendering technique which uses Pixel Shader Ordering, a DirectX extension available on Intel HD graphics hardware". Still, I guess there must be useful things that can be learned from it, even if the method isn't directly usable for us.
Posted: Mon Jul 28, 2014 11:11 am
Yeah but it runs without that at reduced quality and that's mostly the rendering side of things.
The other thing is that it appears to mostly be a method for order independent transparency and there's plenty of other options for that including just doing depth sorting.
Posted: Mon Jul 28, 2014 12:39 pm
FluffyFreak wrote:Yeah but it runs without that at reduced quality and that's mostly the rendering side of things.
Oh, ok, I didn't realise it had a fallback option.
Posted: Mon Dec 08, 2014 11:20 am
I didn't forget, I just ran out of time/patience with this for bit :)
On my cloud-sphere
branch however you can see my work-in-progress hacking based on AndyJ's description of what he's doing to FFED3DAJ's cloud rendering
They're using a selection of static cloud textures which, in the long run, won't really fit with Pioneers procedural universe, however for testing they're interesting and useful.
This will only produce a cloud "layer" at a fixed height but it can look quite nice from space and the surface. Passing through that layer it becomes painfully obvious that it's a single polygon thick :( also watching it for any amount of time shows that it's static.
So, lots to do but it could be good.
Posted: Mon Dec 08, 2014 11:29 am
Posted: Mon Dec 08, 2014 11:49 am
Posted: Mon Dec 08, 2014 11:54 am
Looks really impressive!
nozmajner posted this vimeo
video, (with narration by Carl Sagan) on IRC the same day it came out, I think. Awesome inspiration. Checkout the 3D clouds 00:35-00:45, and see all your ambitions crushed. But maybe some day, in the distant future we'll have something like it as well. Nothing wrong with a little day dreaming.
But static 2D clouds are far better than none. I'm not sure I even remember how clouds worked in FFE.
I really hope your awesome looking clouds make it into master branch (when ready)