Frederic Bouvier wrote:

So the 3d code shouldn't exclude 2d clouds but fade into them.
But if we want to model front one day, it is a thing to ponder now.
Overcast layers are flat at the base but often show towers from the top, at
altitude a 737 or an Airbus are supposed to fly.
I wish we could one day see really big clouds in FG and not only small cumulus

Is that towering cumulus or perhaps some cumulonimbus ?

I agree cirrus can stay 2d.


BTW: how are computed cloud shapes ? In the M. Harris code, they are modeled
with a tool he never released and stuck to the same set of shape ( stored in a
binary format file ). Is it a procedural function or something fixed ?
I am dreaming loudly here but we could envision in a distant future that clouds
could be reshaped at runtime by wind and current. So implementing a very simple
procedural function rather than something fixed could be seen as the first step
of something more ambitious later. In fact I was thinking about implicit
surfaces ( see ) aka
metaballs in blender to picture the sky at a rough level and refine individual
clouds with impostors like M. Harris' or now Harald's code. </dreaming>

ideas provider

Flightgear-devel mailing list

I made 3 screenshot of my lab test program. First I give position and size of containers, those will give the general shape
of the cloud :
With that, and given some hardcoded rules (cu bottom are flat, stratus only use 1 metaball, etc), metaballs are generated :
We have now a sort of volumetric shape that we render with simple quads for speed :

The idea is to build the cloud with a very simple definition (position of boxes, size, content). This definition will be in an xml so
everybody will be able to test new shapes. To reply at your question, yes it is fixed, but there is still the possibility to alter the shape by moving the metaballs or adding new.


Flightgear-devel mailing list

Reply via email to