Chris Metzler ha scritto:
At one point, I used fgsd to do what I thought was some nice work in the
Washington, D.C. area as preparatory to laying out ground structures I
had done or was going to do. The course of the Potomac River is about
500m from its correct location in the FG scenery as is, so I fixed that,
changed material settings for various ground triangles, made some "inlet"
areas that didn't exist in the FG scenery, etc. It looked nice, I thought.
And KDCA no longer sat on a table that hovered out in the middle of
nowhere like it previously did (and does now).
>
Then a new version of the FG scenery came out, and one of its big
advantages was a fix to a bug which occasionally produced sharp steps
in ground elevation where the ground should slope more smoothly. This
was significant because this bug had made the main runway at KDCA
unusable because of a sharp step in the middle. I went with the new
scenery, getting full access to the runway; but I lost the terrain
changes I'd done with fgsd in the process. I haven't re-done them
since, out of fear that I'd either have to throw them out again with
the next FG scenery set, OR would keep them and at the very least have
odd artifacts around the tile edges where one transitions from old
scenery tile to the new stuff (and of course miss out on any improvements
to the scenery generation algorithms that would have impacted the
tile(s) in question).
I think fgsd is cool, and I really enjoy playing with it; but if I
had the infinite amount of free time all of us wish we had, I'd work
on TerraGear drawing its info from some kind of GIS, and implementing
some way (in fgsd and/or other tools) to update that info, so that
"fixes" to the terrain could propagate upstream and be included in
future scenery builds, removing the need to fix the terrain over and
over and over. I know, I know, we've all talked about this before,
and pretty much everyone thinks its a good idea, and no one has the
time. I really really wish I did.
You know what? You are right. The problems coming out from working with
FGSD on scenery tiles instead of with some kind of primitives data set
(which should be preprocessed) are obvious but I wish people don't stop
just because of that.
I still prefer having a working and beautiful scenery _now_ then waiting
for an immaginary scenery in the future. I don't think there will be
that great improvement in global scenery development in the near future
without some kind of a users' personal involvement in the process. This
is true with commercial Flight Simulators too (look at what people do
around the world with MSFS sceneries personalizations, they really rock!).
As I look at that I notice there are many things to be done in FGFS when
thinking about scenery structures. First of all, I don't think that
roads/ground/lakes/railroads/etc... should share the same surface mesh.
This is nonesense. They are very different in the real world, they don't
share much of their properties (geometrical and visual ones) and should
be simulated with appropriate methods instead of a homogeneous one like now.
I know, the way FGFS terrain file format works now is simple and works
good, but imposes to scenery developers a great amount of limitations
and compromises. I'd like to have a tool which could modify ground
surfaces without touching the roads and the railroads (which could be
adjusted, if needed, in a future time and by other hands). And what
about terrain usage borders (urban, ocean, crop etc...), which, by now,
have to coincide with mesh's triangle borders? It's a limitation and
should not be. The terrain mesh should descript the geometry of the
terrain (altitudes, borders, coast lines); textures are then used to
'paint' something over the mesh, I don't see any need to limit the
borders of those textures to the same borders of the meshes. I see in my
imagination a terrain simulation where these and other details are not
stuck together with a rigid terrain description system like the one FGFS
is using now.
But now, FGFS works in this way. The terrain has a file format which can
be easily modified with FGSD and the only chance a simple user like me
has to fully enjoy the simulation is to make full use of it's potentials
_now_. You know what? I like the Palermo scenery now that I hacked it
:-) I really didn't before that :-(
I hope many other people will walk this way, maybe in the future there
will be so many personalized scenery around the world (it's just a
supposition) that an evolution of the terrain file format should take
into account a good tool for converting old ones without loosing those
users' improvements or should consider some kind of backwards compatible
terrain parser.
Anyway, I think the best way is to develop a new set of world elements
description. The roads, the terrain boundaries, the terrain textures
need a very deep restyling in order to attract users around the world. I
am shure that just like Airplane's 3d models, the terrain simulation
should involve users around the world, and they need a good and easy
approach in doing that in order to be effective. If we remain stuck to
this system, which does not consider those kind of collaborative effort,
we will always have a mediocre terrain simulation.
Roberto
_______________________________________________
Flightgear-devel mailing list
Flightgear-devel@flightgear.org
http://mail.flightgear.org/mailman/listinfo/flightgear-devel
2f585eeea02e2c79d7b1d8c4963bae2d