Yadgar wrote:
> 
[...]
> 
> But even with this simpler version, a new problem arises: which
> pattern/noise modification to choose to get halfway realistic cloud
> coverage, i. e. cyclonic/anticyclonic whirls? I tried a whole range of
> patterns, from agate to spiral1, usually frequency 1, with more or less
> turbulence, scaled between 0.1 and 1 and with different scopes within
> the color map for non-transparent white.
> 

Realistic clouds from above are a difficult thing, you should try using warps
for local changes of the pattern.  A turbulent bozo pattern is often a good
start for cloud structures, but you could also try something else.  

> Also, I thought about some relief for my clouds to emulate cumuli
> structures... up to now, I've tried bumps and dents, but none of them
> yielded satisfying results.
> 

You have to remember, that normal statements only change the normal vectors but
not the geometry of the clouds.  For volume either use media of several cloud
layers. 

Concerning the ground image map, i wonder whether the white parts are already
clouds.  Furthermore i would suggest interpolation for getting rid of the grid
structures and maybe you should consider real mountains with an isosurface.  (it
looks somewhat flat right now)

The atmosphere looks good, but i wonder about the height ratio: if the clouds
are 5 km high, the atmosphere seems to be visible up to more than 100 km which
is too high IMO.  

Christoph

--
Christoph Hormann <chr### [at] gmxde>
Homepage: http://www.schunter.etc.tu-bs.de/~chris/

Post a reply to this message