I have done a patch to eliminate the jitter of 3D-objects near the viewpoint
(for example 3D cockpit objects).
The problem is the roundoff accuracy of the float values used in the
scenegraph together with the transforms of the eyepoint relative to the
scenery center.
The solution will be to move the scenery center near the view point.
This way floats relative accuracy is enough to show a stable picture.
To get that right I have introduced a transform node for the scenegraph which
is responsible for that shift and uses double values as long as possible.
The scenery subsystem now has a list of all those transforms required to place
objects in the world and will tell all those transforms that the scenery
center has changed when the set_scenery_center() of the scenery subsystem is
called.
The problem was not solvable by SGModelPlacement and SGLocation, since not all
objects, especially the scenery, are placed using these classes.
The first approach was to have the scenery center exactly at the eyepoint.
This works well for the cockpit.
But then the ground jitters a bit below the aircraft. With our default views
you can't see that, but that F-18 has a camera view below the left engine
intake with the nose gear and the ground in its field of view, here I could
see that.
Having the scenery center constant will still have this roundoff problems, but
like it is now too, the roundoff error here is exactly the same in each
frame, so you will not notice any jitter.
The real solution is now to keep the scenery center constant as long as it is
in a ball of 30m radius around the view point. If the scenery center is
outside this ball, just put it at the view point.
As a sideeffect of now beeing able to switch the scenery center in the whole
scenegraph with one function call, I was able to remove a one half of a
problem when switching views, where the scenery center was far off for one or
two frames past switching from one view to the next. Also included is a fix
to the other half of this problem, where the view position was not yet copied
into a view when it is switched (at least under glut). This was responsible
for the 'Error: ...' messages of the cloud subsystem when views were
switched.
have done a patch to eliminate the jitter of 3D-objects near the viewpoint
(for example 3D cockpit objects).
The problem is the roundoff accuracy of the float values used in the
scenegraph together with the transforms of the eyepoint relative to the
scenery center.
The solution will be to move the scenery center near the view point.
This way floats relative accuracy is enough to show a stable picture.
To get that right I have introduced a transform node for the scenegraph which
is responsible for that shift and uses double values as long as possible.
The scenery subsystem now has a list of all those transforms required to place
objects in the world and will tell all those transforms that the scenery
center has changed when the set_scenery_center() of the scenery subsystem is
called.
The problem was not solvable by SGModelPlacement and SGLocation, since not all
objects, especially the scenery, are placed using these classes.
The first approach was to have the scenery center exactly at the eyepoint.
This works well for the cockpit.
But then the ground jitters a bit below the aircraft. With our default views
you can't see that, but that F-18 has a camera view below the left engine
intake with the nose gear and the ground in its field of view, here I could
see that.
Having the scenery center constant will still have this roundoff problems, but
like it is now too, the roundoff error here is exactly the same in each
frame, so you will not notice any jitter.
The real solution is now to keep the scenery center constant as long as it is
in a ball of 30m radius around the view point. If the scenery center is
outside this ball, just put it at the view point.
As a sideeffect of now beeing able to switch the scenery center in the whole
scenegraph with one function call, I was able to remove a one half of a
problem when switching views, where the scenery center was far off for one or
two frames past switching from one view to the next. Also included is a fix
to the other half of this problem, where the view position was not yet copied
into a view when it is switched (at least under glut). This was responsible
for the 'Error: ...' messages of the cloud subsystem when views were
switched.
Changes
=======
- corrected some strange behavior when playing with the render dialog options
- the density slider is now working : if you are fps limited and still want to see clouds in
the distance you should play with that
- added the choice for texture resolution, its more comprehensible now (before it was
wrongly allways choosing 64x64 textures)
- changed the initial texture size : you now have 64 texture of 64x64, this uses 1Mo of
texture memory (before it was 20 texture of 256x256, that took more memory and there was
not enought impostors)
- sun vector is now right so the lighting is a bit better
- removed useless sort and light computations for impostors, this should save a lot of cpu
- blending of distant cloud is more accurate now
- clouds are now positioned correctly, they don't try to escape you anymore
- no more red/white boxes around cloud
- textures are now filtered (no more big pixels)
known bugs
==========
- distant objects are seen in front of clouds
- new and updated sources for the new volumetric clouds
- 2 new textures for the clouds
- an update to the render dialog to enable/disable and change a few parameters
for the new clouds
foreach, except that the variable gets the index instead of the list
element. Should be useful, and took almost no code to implement.
Support for operator/assignment syntax: +=, -=, *=, /= and ~= now do
what you think they should.
Library support for a bind() function (see the docs Andy is still
writing), allowing runtime modifications to function lexical
environments.
that the column math goes wacky when lgalloced is allowed to be
zero.
Augment the find() function to take a starting index.
Fix strc() to use a default index of zero.
Fix parser precedence of TOK_MINUS, so that "a-b-1" means (a-b)-1 and
not a-(b-1).
Re-organisation: <diffuse>, <ambient>, <emission>, <specular> are
now groups with members <red>, <green>, <blue>, <factor>, <offset>,
and their <*-prop> forms. Additionally, there's an option <property-base>
that can be used to set a path that is prepended to all <*-prop> paths.
It defaults to an empty string. Rationale: see model-howto.html.
here is the promised material animation. It looks a bit longish, but that
wasn't avoidable -- there are simply too many parameters to consider. I tried
hard, though, to make the animation fast by only doing the necessary stuff.
It doesn't affect the frame rate here with my test model. The animation is
heavily based on Jim's "material-emission" animation.
* implementation of the "material" animation (this required to make the
texture path available) + documentation update ($FG_ROOT/Docs/)
* fix some more return values (texture animations, and select) for the
shadow problem (and some in anticipation of other problems :-)
* fix compiler warning
the cause for the disappearing shadows is, that SimGear doesn't tell plib
to call the pre-traversal-callback function on culled objects. These calls,
however, are necessary to execute the transform animation that does, for
example, translate a shadow back into the frustum! Curretnly, the callback
is only executed, and the shadow only magically pops up again, when the object
enters the frustum because the view has changed significantly.
The plib documentation does only talk about TRUE and FALSE for possible
return values from the pre-traversal-callback. But src/ssgEntity.cxx reads
like this:
int ssgEntity::preTravTests ( int *test_needed, int which )
...
int result = (*preTravCB)(this,which) ;
if ( result == 0 ) return FALSE ;
if ( result == 2 ) *test_needed = 0 ;
...
So the return value needs to be 2 to bypass the cull test for pretraversal,
and get the pretraversal in any case. I only changed the return values in
four animations: scale, rotate, translate, and range, because these are
the most likely to move an object out of the frustum. It's not necessary
for blend/alpha/texture manipulation etc. Of course, this is a bit more work
for plib, but the performance will probably not be affected, because:
* these four animations are mainly used for the aircraft model (the spin
and billboard (trees!) animations are not affected)
* the number of extra nodes to process is quite low
* a part of the time spent for the extra nodes to be processed, was before
used for workarounds that are now not necessary any more
I didn't observe a frame rate drop, at least.
code generation of constant objects to use real identity and not Nasal
equality, so (e.g.) the constants 1 (number) and "1.0" (string) do not
get turned into the same object in the generated code.
This patch adds support to the model animation system for modifying emissive
states on the fly so that it is possible to make "lights" appear to dimm.
This is an example of a configuration entry which should explain how it is used:
<animation>
<type>material-emission</type>
<object-name>Face</object-name>
<property>/controls/lighting/instruments-norm</property>
<emiss-red>1.0</emiss-red>
<emiss-green>0.8</emiss-green>
<emiss-blue>0.5</emiss-blue>
</animation>
Note the color entries are the emissive colors when the "property" value is
1.0. They are useful for tinting the light. The "property" itself must be
float or double and is clamped to values between 0 ~ 1.0 inclusively. The
"property" value is multiplied against the colors to get the actual material
properties. Thus property value 0.0 = darkest, and 1.0 = brightest.
If alcOpenDevice( NULL ) is NULL, then context is never assigned a
value, and it's pointless to ask for it in the next "if". But as the
ALCcontext that context points to doesn't seem to be fully defined
(OpenAL bug), valgrind still complains ...
Erik Hofman:
Extend this some further and define context=0 otherwise and check for
context != 0 before using it.
Trying to find the bug in tower.cxx (that crashes fgfs quite frequently
for me!), I'm playing with valgrind again. Until I'm in the ATC subsystem
there will be some other bugs and nitpicking along the way.
valgrind doesn't like that imgage->tmp is once allocated with new and
once with new[], sometimes with malloc() (via map), and sometimes freed
with delete (not delete[]!) and sometimes with free(). With simple types
such as GLubyte this shouldn't really make a difference, but anyway.
Also, I promised that I'd send patches for "if (foo) delete foo;" as
I'm making other changes to concerned files. texture.cxx is one with a
few occurrences thereof. (Remember: C++ explicitly allows to delete
null-pointers, so this check is redundant, and hence not tolerated in
other projects, such as KDE. Doesn't have to impress us, of course. :-)
Also, fixes 4 signed/unsigned warnings (gcc 3.3.4)
The error properly belongs to the enclosing scope, not the called
(non-)function. This bug was fixed a few months back in my private
tree, but Melchior just discovered that the new Concorde scripts
tickle it. I really need to re-synchronize SimGear with my own Nasal
tree...
this is the animation code that do randomisation of the spin animation. The XML tags are modified to support the syntax below :
<use-personality type="bool">true</use-personality>
<factor>
<random>
<min>1.8</min>
<max>2.2</max>
</random>
</factor>
<starting-pos-deg>
<random>
<min>0</min>
<max>360</max>
</random>
</starting-pos-deg>
instead of usual :
<factor>1.42</factor>
<starting-deg-pos>42.0</starting-deg-pos>
1. Remove the dependency on alut which (on certein platforms) might pose
some restrictuons on commercial use.
2. Create a sound source just prior to playing the sound and destroy it
again when the sound has stopped. This should greatly reduce the
error reports from Windows users.
The following patches to SimGear & FlightGear ...
- create an FGMetar abstraction layer, whose purpose is:
* provide defaults for unset values
* interpolate/randomize data (GREATER_THAN)
* derive additional values (time, age, snow cover)
* consider minimum identifier (CAVOK, mil. color codes)
- add rain/hail/snow/snowcover support on the METAR side
- add max age of METAR data handling (currently set to
- add support for an external METAR cache proxy server
- add CAVOK handling
- set missing year/month in regular METAR messages
- fix a small bug in metar.cxx (wrong return value)
