ehofman [Sun, 17 Oct 2004 17:29:34 +0000 (17:29 +0000)]
Frederic Bouvier:
This is a patch to make display list usage optional. They are on by default.
Use --prop:/sim/rendering/use-display-list=false to use immediate mode.
There is also a change in exception handling in main.cxx and bootstrap.cxx
ehofman [Sun, 17 Oct 2004 17:23:19 +0000 (17:23 +0000)]
Frederic Bouvier:
This is a patch to make display list usage optional. They are on by default.
Use --prop:/sim/rendering/use-display-list=false to use immediate mode.
There is also a change in exception handling in main.cxx and bootstrap.cxx
ehofman [Sat, 16 Oct 2004 12:37:39 +0000 (12:37 +0000)]
Roy Vegard Ovesen:
Instrumentation and systems are now configureable from xml files. The two
generic configurations generic-systems.xml and generic-instrumentation.xml
configures the systems and instrumentation as it was hardcoded. You can
override the generic configurations in a similar way as you override the
autopilot configuration.
ehofman [Thu, 14 Oct 2004 09:19:44 +0000 (09:19 +0000)]
Roy Vegard Ovesen:
I've added some digital filters to the autopilot. They are all low-pass
filters that filter away high frequency signals/noise. There are 4 different
filters:
1. Exponential - The algorithm is essentially the same as the one used in the
fgGetLowPass() function.
2. Double exponential - Two exponential filters in series. This filter has a
"steeper" frequency response curve. It filters "better" than the single
exponential.
3. Moving average - Averages a number of inputs.
4. Noise spike - limits the amount that the output value can change from one
sample to the next.
Filters 1 and 2 are characterised by it's filter-time in seconds. For filter 3
you have to set the number of input samples to average over. For filter 4 you
set the maximum allowed rate of change as [1/s]. Since the sampling interval
(dt) isn't constant we have to calculate the maximum allowed change for every
update.
Example of a double exponential filter with filter time 0.1 seconds, that is
1/0.1 = 10 Hz.
ehofman [Thu, 7 Oct 2004 20:34:56 +0000 (20:34 +0000)]
Melchior FRANZ:
As I had reported on 2004/8/4 00:02:56 ("yasim + bo105 + vrp + @#%$#@ == argh!")
there must be a bug somewhere in YASim, which is responsible for the Bo105
turning around the FDM origin (nose tip) rather than the CG. Some people assumed
that I was just another victim of the "view offset" illusion, but this wasn't
and isn't the case.
Maik Justus (the rotor man) has now supposedly found the bug in YASim[1].
Look at this code in FDM/YASim/Integrator.cpp:35--66:
ehofman [Wed, 6 Oct 2004 09:57:37 +0000 (09:57 +0000)]
Reverse the declaration order. jpgRenderFrame (form
erly known as trRenderFrame) is now declared as a NULL function pointer and ass
ignment of the proper function is now done in FlightGear (jpgRenderFrame=FGRend
erer::update).
curt [Wed, 6 Oct 2004 03:29:14 +0000 (03:29 +0000)]
Continued work to clean up the hires screen shot code and move it back towards
a working state. I still see an anomoly when taking a screen shot from inside
a 3d cockpit, but external (chase/tower) views seem to work well. I also
added a property to control how many screen-res tiles are generated in the
output. Theoretically, you can now generate unlimited resolution screen shots,
or limited only by your disk space and patience.
Today I successfully generated a 20*1024 x 20*768 (20480x15360) resolution
screen shot. If you rendered that at 100 dpi it would cover a poster of
about 17 feet by 12.8 feet.
Good luck trying to display something that big or convert it to anything
useful on a typical PC. :-)
david [Sat, 2 Oct 2004 22:35:39 +0000 (22:35 +0000)]
Restore trRenderFrame (which used to be in main.cxx before the
split). If SimGear is configured --with-jpeg-factory, then FlightGear
will fail to build unless this function is present.
FIXME: this is very messy architecturally -- find a better solution,
like passing this explicitly as a callback to the libJPEG class
(SimGear should not have a dependency on FlightGear).
ehofman [Sat, 2 Oct 2004 12:02:04 +0000 (12:02 +0000)]
Vivian Meazza:
I've finished the variation of Cd with Mach number.
The calculations are only applicable to ballistic
objects, and then strictly one shape: non boat-tailed bullets/shells, so
I've put them in AIBallistic rather than AIBAase. For all inputs, Cd
should be the sub-sonic value, so bullets will need changing.
Use the Transmit-Render functions for tower output instead of directly calling ATCdisplay. This is part of preparations for optionally handing the strings off to a TTS engine if required.
The calculation of submodel mass from weight has been moved from AIBallistic
to Submodel so that it is calculated only once, rather than on every
iteration as a present. The parameter <contents> has been added, primarily
so that droptanks will have the proper mass. It is the path to an
appropriate property containing a weight in lbs.
Care has to be taken with the use of <contents> because after a reset there
appears to be a delay in submodel instantiation (dt not properly reset???)
and the weight property is not always picked up before it is set to zero in
the key bindings. Slightly hard to explain. It works fine if FGFS has not
been reset though. There is a partial solution which involves the rejigging
of the fuel and gui nasal scripts, but there is still the visible delay in
instantiation to be resolved. I've nearly done the nasal fixes, which will
form part of an update to the Hunter only. I'll probably complete those
later today.
The value of rho (air density) varies with height. (Including the upper
stratosphere, ust in case someone wants to model ICBMs.) The standard
atmosphere is used (based on a sea-level temperature of 15 deg C.).
Erik Hofman:
I moved this code over the AIBase::update() so all AIModels can make
use of rho, temperature, pressure, etc.
I have added <Cd> and <weight> to the input parameters in the submodels.xml
script. Raw data may be used, thus avoiding the need to guestimate <eda>.
Eda remains, but should now be used to enter the proper cross-sectional
area.
Finish what was committed in a broken state yesterday.
Split up main.cxx into a program manegement part (which remains in
main.cxx) and a render part (the new renderer.?xx files). Also turn
the renderer into a small class of it's own. At this time not really
exctining because most of the stuff is still global, but it allows us
to slowly migrate some of the global definitions into the new class.
The FGRenderer class is now managed by globals, so to get the renderer
just call gloabals->get_renderer()
At some pijt it might be a good idea to also turn the remaining code in
main into a class of it's own. With a bit of luck we end up with a more
robust, and better maintainable code.
I had to reverse a number of signs to get it right. I took the opportunity
to add roll to the submodel so that droptanks will come off with the right
orientation. I have neither added the rotational speed to the submodel, nor
yaw, so if you release droptanks with significant roll rate or yaw angle on
the aircraft the submodel will not be quite right. Straight and level, or
nearly so, is fine.
Make a subtle change to tile loading/unloading policy in order to make the tile
paging system much more robust when position change is very rapid and sporadic.
Recall that we must load 3d models in the main render thread because model
loading can trigger opengl calls (i.e. with texture loading) and all opengl
calls *must* happen in the main render thread.
To accomplish this we load the base tile in the pager thread and build a work
queue of external models that need to be loaded. We never allow a tile to be
paged out of the tile cache until all it's pending model loads are complete.
However, when changing position very rapidly, we can quickly create a huge
backlog of pending model loads because we are changing positions faster than we
can load the associated models for the existing tiles. The end result is
that tiles that are long out of range can't be removed because there is still
a huge backlog of pending model load requests and memory blows up.
This change being committed allows the tile paging system to remove tiles
if they are out of range, even when there are pending models to load. The
model loading code in the render thread can now check to see if the tile
exists and discard any model load request for tiles that no longer exist.
This situation should never occur in normal operation, but could occur in
"contrived" situations where an external script was rapidly changing
the simulator position to then be able to query FG terrain height, and doing
this for a large number of points that are distributed across a large area.
The maths, so far, is now correct. Roll and pitch are now both in the
correct sense. The aircraft velocity is added correctly to the
submodel velocity, and the submodel is now visible when instantiated.
However, the velocity is measured at the aircraft centre. To be totally
correct we ought to take into account the aircraft's rotational
velocity. We have pitch rate and roll rate available, but not yaw rate
(small anyway).
First stab at a simple script that inputs a list of lon/lat coordinates via
stdin, and outputs the coordinates with FG ground elevation. This requires
a running copy of FlightGear with --fdm=null and the telnet server enabled.
Of course you need to have scenery installed for all areas you are querying.
This is not fast and the scenery load wait time may need to be tuned for
individual systems.
Here are some things I've added to the submodel code.
First, I added a first_time value that is true when the trigger is pressed and
false when the trigger is released. The true value is also made false after
the first pass through release(). Release() then uses this to force the
first dt (per salvo) to be zero. I was hoping this would make the submodel
appear closer to the airplane, but I don't notice a difference with the
tracers. In a prior test I found that the first dt is about 2.5 times larger
than subsequent ones. Maybe this will be effective with slower submodels,
like smoke, contrails, etc.
Secondly, I updated the IC.elevation and IC.azimuth calcs to correctly add in
the yaw and pitch offsets, corrected for bank angle. Actually this is still
an estimation. A proper calculation will sum the submodels vector with the
airplane's vector. Until that's done only models which are fired forward
will have proper IC.
Testing revealed that the code was not reading y-offset - a
typo in the original code, and roll was in the wrong sense. All readily
fixable, and it now works.
I've added another parameter to the submodel - wind.
It's activated by the entry <wind>true</wind> in the ../submodel.xml file.
If true, the submodel is affected by the local wind, otherwise not. The
parameter defaults to false. This is useful for exhausts and smoke, and
possibly all objects.
ehofman [Mon, 30 Aug 2004 11:13:29 +0000 (11:13 +0000)]
Tie the count into the property tree instead of using snprintf every frame. Add the life-timer to the property tree under sim/time/elapsed-sec for AI object animation (blending and/or scaling).
ehofman [Mon, 30 Aug 2004 09:15:04 +0000 (09:15 +0000)]
David Culp: This is n idea from Vivian, it allows setting the count to a negative number which will give unlimited submodels (as long as the trigger is true). This is useful for contrails.
ehofman [Thu, 26 Aug 2004 16:25:54 +0000 (16:25 +0000)]
David Culp:
Silly me. I was starting the timer at zero, so the first tracer didn't fly
until 0.25 seconds after pulling the trigger. Now the timer starts at the
same value as "delay", so the first round comes out immediately.
Also, I've added an optional configuration attribute that allows you to change
the ballistics of the submodel. This allows parachutes, or anything else
that has ballistics different from a bullet. The attribute is called "eda",
which is the equivalent drag area. Default value is 0.007, which gives the
same ballistics as the current tracers. Increasing this value gives more
drag. A value of 2.0 looks good for a parachute.
math stuff
########################################################################
The deceleration of the ballictic object is now given by:
[ (rho) (Cd) ] / [ (1/2) (m) ] * A * (V * V)
where rho is sea-level air density, and Cd and m are fixed, bullet-like
values. So the calculation is:
The value "A" is what I'm calling the "eda" (equivalent drag area).
########################################################################
A parachute model will have to be built so that the parachutist's feet
are in the forward x-direction.
Here is the submodel.xml config I use for "parachutes":
ehofman [Thu, 26 Aug 2004 12:27:39 +0000 (12:27 +0000)]
David Culp, Vivian Meazza:
Last night I sent these new files to Vivian to fix a problem he found. Since
we can have more than one submodel we need more than one "count" property.
The new code creates a property systems/submodels/submodel[n]/count for each
submodel. Vivian is using this count property for his sound.
ehofman [Thu, 26 Aug 2004 08:38:43 +0000 (08:38 +0000)]
David Culp:
Here is an update for the submodel system. This will allow submodels to be
defined for any aircraft, and there are no default submodels. To use this
submodel system you need to set up a binding (slight change in property name
from last one, but you can use any property name you like, as long as it
matches the name in the submodels.xml file, see below):
Then you put the submodel configuration file in your aircraft's directory.
I've attached a file, submodels.xml, that can be used to define a gun that
works just like the former one did.
There are two things remaining to be done. One is to change the function
SubmodelSystem::transform() to properly position the submodel. This will
require some complicated matrix code that I might borrow from Yasim.
ehofman [Tue, 24 Aug 2004 08:40:41 +0000 (08:40 +0000)]
Move the atexit() handler over to bootstrap.cxx (which makes libMain.a behave like a real library) and fix a segmentation fault when running 'fgfs -h'. The exit handler tried to change the (still uninitialized) mouse cursor causing a core dump.
In the *-set.xml file NOT within <sim></sim>. Trouble is it still defaults
to serviceable=true, but it also defaults to amount=0. Unless this bit of
code is included it wont run, so other models can't fire inappropriate
tracer. Revised submodel.cxx file attached.
I've changed to a non-billboard solution. I tried to make the viewing angle
of the tracer less than 180 degs, but failed. Close enough I think. It's
still a bit big, but a reasonable compromise. .AC file attached.
I remain concerned about the tracer colour. In practice, red is better (and
it's what I am used to) but David seems set on white, and I don't want to
upset him. I'm going to adjust the texture a bit more tomorrow.
ehofman [Sun, 22 Aug 2004 16:22:18 +0000 (16:22 +0000)]
David Culp:
Right now the code is not very configurable, and there is only one submodel per airplane possible. It is implemented as an SGSubSystem, just like the electrics, vacuum, etc. systems. To make it work you need to make a release binding like this (for my joystick trigger):
Then, each airplane that uses the system should have something like this added to its *-set.xml file (note that this does *not* go within the <sim></sim> tags):
Future improvements will include:
1) more configurability, so the user can create multiple submodels, and can assign them different locations, and pitch and yaw adjustments, and nitial velocity.
2) sound?
3) a more accurate calculation of the submodels location at any pitch/roll/yaw.
4) a way to pre-load the model, so the AI code doesn't have to parse the model every time it creates an instance.