fly/fgdata
1
0
Fork 0
Code Activity
fgdata/Docs/README.scenery
Stuart Buchanan c974f5242d Clean up README files. 
- Remove trailing whitespace
- Replace tabs with spaces
- Reformat to < 85 characters to avoid autogenerate readme.pdf bleeding over
  margins.
2013-06-20 23:02:17 +01:00

580 lines
19 KiB
Text
Raw Blame History

This document describes how FlightGear searches and loads scenery, how to
add static objects to the scenery as well as the syntax of *.stg files.
Contents ----------------------------------------------------------------------
1 scenery path
2 terrasync
3 stg files
3.1 OBJECT_BASE
3.2 OBJECT
3.4 OBJECT_SHARED
3.3 OBJECT_STATIC
3.5 OBJECT_SIGN
4 model manager ("/models/model")
4.1 static objects
4.2 dynamic objects
4.3 loading/unloading at runtime
5 tools for object placing
5.1 calc-tile.pl
5.2 ufo scenery object editor
6 embedded Nasal
6.1 static models
6.2 AI models
1 scenery path ----------------------------------------------------------------
FlightGear loads scenery by default from the Scenery/ subdirectory of its
data directory. The path to this data directory can be set via environment
variable FG_ROOT or the --fg-root option. The scenery path can be set
independently via environment variable FG_SCENERY or option --fg-scenery.
The order of precedence is as follows:
--fg-scenery=/some/dir ... highest priority
$FG_SCENERY
$FG_ROOT/Scenery/ ... lowest priority
A scenery specification may be a list of paths, separated by the OS-specific
path separator (colon on Unix/OSX, semicolon on MS Windows). The paths are
searched in the order from left to right:
FG_SCENERY=/first/dir:/second/dir:/third/dir
(likewise with --fg-scenery option)
Each of the scenery paths can follow one of two possible layouts: with or
without Terrain/ and Objects/ subdirectories. As soon as either or both
of these subdirectories are found, scenery is only searched *in* these two,
but not in any other directory on the same hierarchy level!
This example shows which directories are used to search for scenery:
$ ls /first/dir
w130n30/ searched
$ ls /second/dir
Objects/ searched
Terrain/ searched
w130n30/ *not* searched
$ ls /third/dir
Terrain/ searched
w130n30/ *not* searched
If FlightGear searches for a particular "tile" file, let's say for
"w130n30/w123n37/942050.stg", then (using the above examples) it looks
into
/first/dir/w130n30/w123n37/942050.stg (A)
/second/dir/Terrain/w130n30/w123n37/942050.stg (B)\__ same path element
/second/dir/Objects/w130n30/w123n37/942050.stg (C)/ /second/dir
/third/dir/Terrain/w130n30/w123n37/942050.stg (D)
but as soon as it finds an OBJECT_BASE entry it only finishes this
path element and then stops scanning. So, if (B) contains an entry
"OBJECT_BASE 942050.btg, then the twin Objects/ directory (C) will
be read, too. But (D) will *not*! Objects/ and Terrain/ directories
are laid out equally. Airport and elevation data, as well as airport
inventory objects are usually put into Terrain/, while other objects
are put into Objects/.
This searching behavior is usually used to collect user-added
custom objects first, then to read in standard scenery and objects
that came with the distribution (San Francisco Bay area), and to
use locally added scenery everywhere else. So a typical scenery
path specification could look like this:
FG_SCENERY=$HOME/.fgfs/Scenery:$FG_ROOT/Scenery:$FG_ROOT/WorldScenery
The third path would then be populated by the user with unpacked scenery
archives downloaded from http://www.flightgear.org/Downloads/scenery.html,
or by using terrasync (see next section).
Additional objects can be downloaded from the FlightGear Objects
Database (http://scenemodels.flightgear.org/download/). (Note that
those objects are occasionally merged into the flightgear.org/terrasync
packages, so you may end up with doubled entries!)
Using a private directory for downloaded add-on scenery and adding
that path to FG_SCENERY is the preferred way. This separates default
data from locally added data, and makes administration and later updates
easier.
HINT: if you want to see where FlightGear is searching and finding
terrain/objects, start it with the --log-level=info option.
2 terrasync -------------------------------------------------------------------
FlightGear comes with a utility "terrasync" that allows downloading
scenery (literally) "on-the-fly. Given the scenery path setup from
section 1 you could use terrasync with a script like this:
#!/bin/bash
PORT=5503
nice terrasync -p $PORT -d $FG_ROOT/WorldScenery&
fgfs --atlas=socket,out,1,localhost,$PORT,udp $*
killall terrasync
If you name it "fgfsterra", then you can use it just like you would use
"fgfs", but behind the scenes it would update your scenery everywhere in
sight and save the files to $FG_ROOT/WorldScenery. Example:
$ ./fgfsterra --aircraft=ufo --airport=LOXZ
Note, however, that if it downloads scenery for the area around your
starting location, then you'll only see that after the next start, or
after you flew or teleported to a distant location and then back.
terrasync depends on the rsync application and an open port 873,
so it may not be available/usable on MS Windows.
3 stg files -------------------------------------------------------------------
stg files ("static terragear") define the static elements of a scenery
"tile", including the terrain elevation data, airport geometry, and all
static objects placed on this tile. (See section 5 for how to find out which
geo coordinates belong to which tile.) Four of the available key words
are followed by a string and four numbers. The meaning of these numbers
is always the same and described in section 3.3.
3.1 OBJECT_BASE
----------------
specifies the terrain elevation data file. These files are generated with
the TerraGear tools (http://www.terragear.org/) and have file extension
".btg" ("binary terragear"; there used to be an "*.atg" file, too, where
the 'a' stood for ASCII).
Example:
OBJECT_BASE 942050.btg
The entry may be anywhere in the 942050.stg file, on a separate line.
3.2 OBJECT
-----------
specifies an airport geometry 'drop-in' file. The scenery elevation file
has cut out holes for airports, that are filled with such objects. They
are usually called after the airport ICAO id:
Example:
OBJECT KSFO.btg
These files are, again, created by TerraGear tools and are usually gzipped,
so you'll find that file stored as KSFO.btg.gz.
3.3 OBJECT_SHARED
------------------
add static object to the tile.
Example:
OBJECT_SHARED Models/Airport/tower.xml -122.501090 37.514830 15.5 0.00 0.00 0.00
Syntax:
OBJECT_SHARED <object-path> <lon> <lat> <elev-m> <hdg-deg> <pitch-deg> <roll-deg>
The <object-path> is relative to the data directory (FG_ROOT).
<elev-m> is in meter and relative to mean sea-level (in the fgfs world).
<hdg-deg> is in degree, counter-clockwise with North being 0.0. Note
that this differs from about every other place in FlightGear, most notably
the /orientation/heading-deg entry in the property system, which is clockwise.
<pitch-deg> and <roll-deg> are in degree and optional.
OBJECT_SHARED models are cached and reused. They are only once in memory
and never freed. (See also the next section.)
3.4 OBJECT_STATIC
------------------
add static objects to the tile, just like OBJECT_SHARED. There are three
differences to OBJECT_SHARED (apart from the name):
(A) the path is relative to the tile directory where the *.stg file with
this entry is located. For example, relative to 130n30/w123n37/. This
usually means that all 3D object files, textures, and XML animation
files are in this tile directory, too.
(B) these objects are *not* cached and kept loaded, but rather freed with
the tile (that is, when you leave that area).
(C) the animation XML files may contain Nasal blocks <nasal><load> and
<nasal><unload> which are executed on loading/unloading.
Example:
OBJECT_STATIC ggb-fb.xml -122.4760494 37.81876042 0 105 0.00 0.00
3.5 OBJECT_SIGN
---------------
defines taxiway or runway sign. The syntax is much like that of OBJECT_SHARED
entries, except that the path is replaced with a sign contents specification
and that there is an additional size value at the end of the line.
Example:
OBJECT_SIGN {@R}10L-28R{@L}C -122.35797457 37.61276290 -0.5398 74.0 2
The sign specification defines the sign contents. We try to resemble the
apt.dat 850 specifications in our implementation.
In the simplest form it contains just 'normal' text, for example:
EXIT
This will create a black panel of 1m height with "EXIT" written on it
in white versal letters. Actually, each of those characters are
single-letter glyph names that are looked up in the <glyph> map of a
texture font <material> entry in $FG_ROOT/materials.xml. It just
happens that the <glyph> entry for <name> 'E' maps to a drawn 'E' in
the font texture. This isn't true for all ASCII characters. Many aren't
mapped at all (and thus not available), others are mapped to non-standard
drawings. The '_', for example, is mapped to an empty black area and can
therefore be used as a space. (The sign specification must not contain
real spaces.) The '*' is mapped to a raised period.
Some glyph names consist of more than one character, and can't, thus, be
used directly. They have to be put in a pair of curly braces:
{^rd}
This creates an arrow that points to the right and down. Braces can really
contain a list of glyph names, separated by commas (no space!).
Single-letter glyph names can be used that way, too, or in any mixture
of both methods:
EXIT
{E,X,I,T}
{E}{X}{I}{T}
EX{I,T}
E{X,I}T{^lu,^rd}
{^u}EXIT{^u}
Multi-letter glyph names are usually used for symbols. Arrow symbol names
always start with a caret ("arrow head") and the left or right direction
always comes first (like the x in a Cartesian coordinate system). Here's
a list of some of the available names (see $FG_ROOT/materials.xml for
more):
^l left arrow
^r right arrow
^u up arrow
^d down arrow
^lu left-up arrow
^ld left-down arrow
^ru right-up arrow
^rd right-down arrow
no-entry "no entry" symbol
critical runway critical area
safety ils safety area
hazard end of taxiway
There are commands for pre-defined sign types according to the FAA
specification (5345-44; see http://www.google.com/search?q=5345-44g).
@Y "Direction, Destination, Boundary" sign (black on yellow)
@R "Mandatory Instruction" sign (white on red with black outline)
@L "Location" sign (yellow text and frame on black)
@B "Runway Distance Remaining" sign (white on black)
Examples:
{@R}10L-28R{@L}C
{@Y,^l}P|{^lu}N{@L}F{@Y}F{^ru}
{@Y,^ld}C ... same as any of {@Y}{@ld}C {@Y,@ld,C}
{@B}17
Syntax errors are reported in --log-level=debug, in the SG_TERRAIN
group. You can use this command line to filter out such messages:
$ fgfs --log-level=debug 2>&1|grep OBJECT_SIGN
4 model manager ("/models/model") --------------------------------------------
4.1 static objects
-------------------
Another way to add objects to the scenery is via the "model manager".
It reads all /models/model entries at startup and places these objects
in the scenery. Just load a definition like the following into the
property tree, for example by putting it into $FG_ROOT/preferences.xml, or
better: an XML file that you load with e.g. --config=$HOME/.fgfs/stuff.xml:
<models>
<model n="0">
<name>pony</name>
<path>Local/pony.ac</path>
<longitude-deg>-115.8352869</longitude-deg>
<latitude-deg>37.24302849</latitude-deg>
<elevation-ft>4534.691321</elevation-ft>
<heading-deg>0</heading-deg>
<pitch-deg>0</pitch-deg>
<roll-deg>0</roll-deg>
</model>
</models>
The <path> is relative to $FG_ROOT, the <name> is optional. One can leave the
heading/pitch/roll entries away, in which case they are set to zero. The values
are fixed and unchangeable at runtime.
4.2 dynamic objects
--------------------
Any of the model properties can be made changeable at runtime by appending
"-prop" and using a property path name instead of the fixed value:
<local>
<pony>
<longitude-deg>-115.8352869/<longitude-deg>
<latitude-deg>37.24302849</latitude-deg>
<elevation-ft>4534.691321</elevation-ft>
<heading-deg>0</heading-deg>
</pony>
</local>
<models>
<model n="1">
<name>pony</name>
<path>Local/pony.ac</path>
<longitude-deg-prop>/local/pony/longitude-deg</longitude-deg-prop>
<latitude-deg-prop>/local/pony/latitude-deg</latitude-deg-prop>
<elevation-ft-prop>/local/pony/elevation-ft</elevation-ft-prop>
<heading-deg-prop>/local/pony/heading-deg</heading-deg-prop>
<pitch-deg>1.234</pitch-deg> <!-- static, just for fun -->
</model>
</models>
Then one can move the pony around by changing the values in /local/pony/ in
the property system. One can, of course, use other animals, too.
4.3 loading/unloading at runtime
--------------------------------
Both dynamic and static model-manager-models can be loaded and unloaded
at runtime. For loading you first create a new <model> entry under <models>,
initialize all properties there (<longitude-deg> or <longitude-deg-prop>,
etc.), and finally you create a child <load> of any type in this group.
This is the signal for the model manager to load the object. You can
remove the <load> property after that. It has no further meaning.
To remove a model-manager model at runtime, you simply delete the whole
<model> group.
5 tools for object placing ----------------------------------------------------
5.1 calc-tile.pl
----------------
For finding out the tile number for a given geo coordinate pair there's
a script "scripts/perl/scenery/calc-tile.pl" in the FlightGear sources.
You feed longitude and latitude to it and it returns the path to the
*.stg file where you have to add the object entry.
$ perl calc-tile.pl 16.1234 48.5678
Longitude: 16.1234
Latitude: 48.5678
Tile: 3220128
Path: "e010n40/e016n48/3220128.stg"
5.2 ufo scenery object editor
-----------------------------
The ufo has a scenery object editor built-in. It uses the model manager
described in section 4. To place objects with it, start fgfs, optionally
with specifying an initial model type ("cursor") and a list of subdirectories
of $FG_ROOT where the ufo should search for available 3D models ("source"):
$ fgfs --aircraft=ufo --prop:cursor=Models/Airport/radar.xml \
--prop:source=Models,Scenery/Objects
Then click anywhere on the terrain to add a model (left mouse button).
You can open the adjustment dialog (Tab-key) to make adjustments to
position and orientation. Click as often as you like, choose further
models from the space-key dialog. You can select an already placed object
by Ctrl-clicking at its base (not at the object itself, but the surface
point where it's located!). By also holding the Shift key down, you
can select several objects or add them to a selection. You can remove
the selected object(s) with the Backspace-key. (See the ?-key dialog
for futher available keys.) After clicking on the input field right
over the status line (invisible if there's no text in it) you can enter
a comment/legend for the selected object.
And finally, you dump the object data to the terminal (d-key) or export
them to a file $HOME/.fgfs/ufo-model-export.xml (Unix) or
%APPDATA%\flightgear.org\ufo-model-export.xml (MS Windows).
You can now put the generated object entries into the specified *.stg
file to make them permanent. Or load the whole exported *.xml file
via --config option:
$ fgfs --config=$HOME/.fgfs/ufo-model-export.xml
If you choose the sign placeholder object from the m-key dialog (first
entry; "Aircraft/ufo/Models/sign.ac"), then an OBJEC_SIGN *.stg line
will be generated with the legend used as sign contents. If you didn't
insert any legend, then the sign text will be: NO CONTENTS and a 4 digits
random number for later identification in the *.stg file.
Unfortunately, objects added with this method are kept in memory, no
matter where you are actually flying, so the *.stg method is preferable.
6 embedded Nasal in XML files (static objects and AI) -------------------------
6.1 static models
-----------------
Objects loaded via OBJECT_STATIC in *.stg files as well as AI models loaded
via scenarios may contain embedded Nasal code. This can be used to drive
more advanced animations. An example is a lighthouse with specific light
signals, or hangar doors that open when the "player"'s aircraft is nearby.
The Nasal code is added to the object's XML wrapper/animation file, anywhere
on the top level, for example:
<PropertyList>
<path>lighthouse.ac</path>
<nasal>
<load>
var loop_id = 0;
var light = aircraft.light.new("
"/models/static/w120n30/w118n35/lighthouse/light",
[2, 1, 2, 1, 2, 1, 2, 5]);
var loop = func(id) {
id == loop_id or return;
light.switch(getprop("/sim/time/sun-angle-rad") > 1.37);
settimer(func { loop(id) }, 30);
}
loop(loop_id += 1);
</load>
<unload>loop_id += 1</unload>
</nasal>
<animation>
<type>select</type>
<object-name>light-halo</object-name>
<property>/models/static/w120n30/w118n35/lighthouse/light/state</property>
</animation>
...
</PropertyList>
The <load> part is executed when the scenery tile on which the model is placed
is loaded into memory. It can start timers or listeners that modify properties,
which are then queried by the <animation>. As a convention developers are requested
to use "/models/static/" + <tile-path> + <file-basename>. So, in the above example
file "$FG_ROOT/Scenery/Objects/w120n30/w118n35/lighthouse.xml" all properties
are stored under "/models/static/w120n30/w118n35/lighthouse/". That way collisions
with other models are quite unlikely.
An optional <unload> part is executed when the tile and model is removed from
memory. Note that this is only when the "player" is already far away! To
cause minimal impact on the framerate it is recommended to do as few
calculations as possible, to use as large timer intervals as possible, and to
stop all timers and listeners in the <unload> part, as shown in the example.
All Nasal variables/functions are in a separate namespace, which is named
after the file name. It's recommended not to access this namespace from
outside for other than development purposes.
What the above code does: as soon as the model is loaded, an aircraft.light
is created with a specific light sequence. Then, in half-minute intervals,
the light is turned on or off depending on the sun angle. On <unload> the
loop identifier is increased, which makes the loop terminate itself. For
more info about this technique, see the Nasal wiki.
6.2 AI models
-------------
Here the syntax is the same like for static models. The only two differences
are:
- these models are currently only removed at program end, so it's more
important to consider effects on performance.
- AI models don't need to store their properties in /models/static/...,
but get a separate node under /ai/models/, for example /ai/models/carrier[1].
The embedded Nasal code can access this dynamically assigned property
via cmdarg() function, which returns a props.Node hash. Example:
<nasal>
<load>print("my data are under ", cmdarg().getPath())</load>
<unload>print("Currently I'm only called at fgfs exit!")</unload>
</nasal>
View git blame Copy permalink