This commit extend the technique of dynamic linked list to the mesh
slots so as to eliminate dumb scan or map lookup. It provides massive
performance improvement in the culling and in the rasterizer when
the majority of objects are static.
Other improvements:
- Compute the opengl matrix only for objects that are visible.
- Simplify hash function for GEN_HasedPtr
- Scan light list instead of general object list to render shadows
- Remove redundant opengl calls to set specularity, shinyness and diffuse
between each mesh slots.
- Cache GPU material to avoid frequent call to GPU_material_from_blender
- Only set once the fixed elements of mesh slot
- Use more inline function
The following table shows the performance increase between 2.48, 1st round
and this round of improvement. The test was done with a scene containing
40000 objects, of which 1000 are in the view frustrum approximately. The
object are simple textured cube to make sure the GPU is not the bottleneck.
As some of the rasterizer processing time has moved under culling, I present
the sum of scenegraph(includes culling)+rasterizer time
Scenegraph+rasterizer(ms) 2.48 1st round 3rd round
All objects static, 323.0 86.0 7.2
all visible, 1000 in
the view frustrum
All objects static, 219.0 49.7 N/A(*)
all invisible.
All objects moving, 323.0 105.6 34.7
all visible, 1000 in
the view frustrum
Scene destruction 40min 40min 4s
(*) : this time is not representative because the frame rate was at 60fps.
In that case, the GPU holds down the GE by frame sync. By design, the
overhead of the rasterizer is 0 when the the objects are invisible.
This table shows a global speed up between 9x and 45x compared to 2.48a
for scenegraph, culling and rasterizer overhead. The speed up goes much
higher when objects are invisible.
An additional 2-4x speed up is possible in the scenegraph by upgrading
the Moto library to use Eigen2 BLAS library instead of C++ classes but
the scenegraph is already so fast that it is not a priority right now.
Next speed up in logic: many things to do there...
- Vast performance increase when removing scene containing large number of
objects: the sensor/controller map was updated for each deleted object,
causing massive slow down when the number of objects was large (O(n^2)).
- Use reference when scanning the sensor map => avoid useless copy.
- Remove dynamically the object bounding box from the DBVT when the object
is invisible => faster culling.
- renamed generic attribute "isValid" to "invalid" since BL_Shader already uses isValid.
- Moved deprecation warnings from CValue
- removed unused KX_Scene::SetProjectionMatrix and KX_Scene::GetViewMatrix
- Added KX_Scene attributes "lights", "cameras", "objects_inactive", to allow access to objects in unseen layers (before the AddObject actuator adds them)
- KX_Camera deprecated cam.enableViewport(bool) for cam.isViewport which can be read as well.
Own bug, conversion function to get an orientation from python - PyOrientationTo() ignored user input completely :| (breaking the orientation attribute)
Also made KX_GameObject worldOrientation writable and minor doc fixes.
CListValue fixes
- Disable changing CValueLists that the BGE uses internally (scene.objects.append(1) would crash when drawing)
- val=clist+list would modify clist in place, now return a new value.
- clist.append([....]), was working like extend.
- clist.append(val) didnt work for most CValue types like KX_GameObjects.
Other changes
- "isValid" was always returning True.
- Set all errors for invalid proxy access to PyExc_SystemError (was using a mix of error types)
- Added PyObjectPlus::InvalidateProxy() to manually invalidate, though if python ever gains access again, it will make a new valid proxy. This is so removing an object from a scene can invalidate the object even if its stored elsewhere in a CValueList for eg.
PyObjectPlus::ProcessReplica() is now called when any of its subclasses are replicated.
This is important because PyObjectPlus::ProcessReplica() NULL's the 'm_proxy' python pointer I added recently.
Without this a replicated subclass of PyObjectPlus could have an invalid pointer (crashing the BGE).
This change also means CValue::AddDataToReplica() can be moved into CValue::ProcessReplica() since ProcessReplica is always called.
Some functions used
ProcessReplica(replica);
others
replica->ProcessReplica()
Use the second method everywhere so the PyObjectPlus's ProcessReplica() can be called from its subclasses.
Note that PyObjectPlus's ProcessReplica isnt used yet.
improved how attribute errors are set so each classes py_getattro function dosnt need to set an error if the attribute doesn't exist.
Now py_base_getattro sets an error on a NULL return value when no errors are set to avoid setting errors at multiple levels.
Separate getting a normal attribute and getting __dict__, was having to do too a check for __dict__ on each class (multiple times per getattro call from python) when its not used that often.
- initialize pythons sys.argv in the blenderplayer
- ignore all arguments after a single " - " in the blenderplayer (like in blender), so args can be passed to the game.
- add a utility function PyOrientationTo() - to take a Py euler, quat or 3x3 matrix and convert into a C++ MT_Matrix3x3.
- add utility function ConvertPythonToMesh to get a RAS_MeshObject from a KX_MeshProxy or a name.
- Added error prefix arguments to ConvertPythonToGameObject, ConvertPythonToMesh and PyOrientationTo so the error messages can include what function they came from.
- deprecated brick.getOwner() for the "owner" attribute.
- More verbose error messages.
- BL_Shader wasnt setting error messages on some errors
- FilterNormal depth attribute was checking for float which is bad because scripts often expect ints assigned to float attributes.
- Added a check to PyVecTo for a tuple rather then always using a generic python sequence. On my system this is over 2x faster with an optmized build.
- comments to PyObjectPlus.h
- remove unused/commented junk.
- renamed PyDestructor to py_base_dealloc for consistency
- all the PyTypeObject's were still using the sizeof() their class, can use sizeof(PyObjectPlus_Proxy) now which is smaller too.
This changes how the BGE classes and Python work together, which hasnt changed since blender went opensource.
The main difference is PyObjectPlus - the base class for most game engine classes, no longer inherit from PyObject, and cannot be cast to a PyObject.
This has the advantage that the BGE does not have to keep 2 reference counts valid for C++ and Python.
Previously C++ classes would never be freed while python held a reference, however this reference could be problematic eg: a GameObject that isnt in a scene anymore should not be used by python, doing so could even crash blender in some cases.
Instead PyObjectPlus has a member "PyObject *m_proxy" which is lazily initialized when python needs it. m_proxy reference counts are managed by python, though it should never be freed while the C++ class exists since it holds a reference to avoid making and freeing it all the time.
When the C++ class is free'd it sets the m_proxy reference to NULL, If python accesses this variable it will raise a RuntimeError, (check the isValid attribute to see if its valid without raising an error).
- This replaces the m_zombie bool and IsZombie() tests added recently.
In python return values that used to be..
return value->AddRef();
Are now
return value->GetProxy();
or...
return value->NewProxy(true); // true means python owns this C++ value which will be deleted when the PyObject is freed
Other small changes...
- KX_Camera and KX_Light didnt have get/setitem access in their PyType definition.
- CList.from_id() error checking for a long was checking for -1 against an unsigned value (own fault)
- CValue::SpecialRelease was incrementing an int for no reason.
- renamed m_attrlist to m_attr_dict since its a PyDict type.
- removed custom getattro/setattro functions for KX_Scene and KX_GameObject, use py_base_getattro, py_base_setattro for all subclasses of PyObjectPlus.
- lowercase windows.h in VideoBase.cpp for cross compiling.
Common cases of this are when python references an object from the AddObject actuator that has ended, or a scene has been loaded and the old objects freed.
This means some scripts will raise errors now in certain cases but better give the error early rather then failing silently with strange hard to track down behavior & crashes.
Added "isValid" attribute for checking objects are in a scene.
At the moment it uses the SceneGraph Node to check of the objects valid but it might be better to do this in a more generic way so scenes, meshes etc also have this check.
There is a problem importing 3ds files where I cant find a way to check if the transforms are applied to the vertex locations or not.
Since 2.44 I made the importer assume they were not since you can manually remove transformations, but not reverse.
Nevertheless most 3ds files have the matrix applied, better not give a bad import by default.
Did some research and other 3ds importers (lib3ds for eg), have the same problem and just assume the transformations applied.
3dsMax imports both correctly so there must be a way to tell but I could not link it to the 3ds version or other mesh options.
Added an option to workaround this problem in rare cases where its needed.
- KX_GameObject.cpp & KX_Scene.cpp, clear the dict before removing the reference in case there is a circular reference.
Clamp objects min/max velocity.
Accessed with bullet physics from the advanced button with dynamic and rigid body objects.
- useful for preventing unstable physics in cases where objects move too fast.
- can add linear velocity with the motion actuator to give smooth motion transitions, without moving too fast.
- minimum velocity means objects don't stop moving.
- python scripts can adjust these values speedup or throttle velocity in the existing direction.
Also made copy properties from an object with no properties work (in case you want to clear all props)
Added occlusion culling capability in the BGE.
More info: http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.49/Game_Engine#BGE_Scenegraph_improvement
MSVC, scons, cmake, Makefile updated.
Other minor performance improvements:
- The rasterizer was computing the openGL model matrix of the objects too many times
- DBVT view frustrum culling was not properly culling behind the near plane:
Large objects behind the camera were sent to the GPU
- Remove all references to mesh split/join feature as it is not yet functional
Support for assigning any Type to a KX_GameObject
so you can do...
gameOb.follow = otherGameOb
gameOb[otherGameOb] = distanceTo
gameOb["path"] = [(x,y,x), (x,y,x)]
del gameOb[mesh]
* types that cannot be converted into CValue types are written into the KX_GameObject dict
* the KX_GameObject dict is only initialized when needed
* Python properties in this dict cannot be accessed by logic bricks
* dir(ob) and ob.getPropertyNames() return items from both CValue and Py dictionary properties.
Also found that CType was converting python lists to CType Lists but very buggy, would crash after printing the list most times.
Use python lists instead since logic bricks dont deal with lists.
There were also some problems with int to python conversion
- assigning a PyLong to a KX_GameObject from python would raise an error
- PyLong were coerced into floats when used with internal CValue arithmetic
Changes...
- PyLong is converted into CIntValue for coercing and assigning from python
- CValue's generic GetNumber() function returns a double rather then a float.
- Print an error when a PyType cant be coerced into a CValue
Tested with python, expressions and property sensor.
- Only try and remove light objects from the light list.
- Only loop over mesh verts once when getting the bounding box
- dont return None from python attribute localInertia when theres no physics objects. better return a vector still.
- add names to send message PyArg_ParseTuple functions.
This commit contains a number of performance improvements for the
BGE in the Scenegraph (parent relation between objects in the
scene) and view frustrum culling.
The scenegraph improvement consists in avoiding position update
if the object has not moved since last update and the removal
of redundant updates and synchronization with the physics engine.
The view frustrum culling improvement consists in using the DBVT
broadphase facility of Bullet to build a tree of graphical objects
in the scene. The elements of the tree are Aabb boxes (Aligned
Axis Bounding Boxes) enclosing the objects. This provides good
precision in closed and opened scenes. This new culling system
is enabled by default but just in case, it can be disabled with
a button in the World settings. There is no do_version in this
commit but it will be added before the 2.49 release. For now you
must manually enable the DBVT culling option in World settings
when you open an old file.
The above improvements speed up scenegraph and culling up to 5x.
However, this performance improvement is only visible when
you have hundreds or thousands of objects.
The main interest of the DBVT tree is to allow easy occlusion
culling and automatic LOD system. This will be the object of further
improvements.
- Make BGE's ListValue types convert to python lists for printing since the CValue GetText() function didnt work well- printing lists as [,,,,,] for scene objects and mesh materials for eg.
- Check attributes are descriptor types before casting.
- py_getattr_dict use the Type dict rather then Method and Attribute array.
Use each types dictionary to store attributes PyAttributeDef's so it uses pythons hash lookup (which it was already doing for methods) rather then doing a string lookup on the array each time.
This also means attributes can be found in the type without having to do a dir() on the instance.
KX_PolygonMaterial and KX_BlenderMaterial - Added a print function (would raise a python error on printing)
* Crashes *
KX_GameObject SetParent - Disallowed setting a parent to its self, caused a recursion crash.
KX_MeshProxy "materials" attribute was segfaulting because of my recent change - I was wrong, you do need to check material types (no idea why since they are both PyObject * at the base)
KX_VisibilityActuator - Wasn't initialized with PyType_Ready() making it crash on access (own fault)
* Crashes because of missing NULL checks *
KX_PolygonMaterial's "gl_texture" attribute wasnt checking for a valid m_tface
KX_GameObject - added checks for GetPhysicsController()
KX_RayCast::RayTest - didnt check for a valid physics_environment
KX_SceneActuator's getCamera python function wasnt checking if there was a camera.
- made camera use PyAttributeDef's
- removed unneeded duplicate matrix type checks
- fixed own bug (added since 2.48a) that broke a converting 4x4 matrix to a PyObject
- Initialize python types with PyType_Ready, which adds methods to the type dictionary.
- use Pythons get/setattro (uses a python string for the attribute rather then char*). Using basic C strings seems nice but internally python converts them to python strings and discards them for most functions that accept char arrays.
- Method lookups use the PyTypes dictionary (should be faster then Py_FindMethod)
- Renamed __getattr -> py_base_getattro, _getattr -> py_getattro, __repr -> py_base_repr, py_delattro, py_getattro_self etc.
From here is possible to put all the parent classes methods into each python types dictionary to avoid nested lookups (api has 4 levels of lookups in some places), tested this but its not ready yet.
Simple tests for getting a method within a loop show this to be between 0.5 and 3.2x faster then using Py_FindMethod()
Added the method into the PyType so python knows about the methods (its supposed to work this way).
This means in the future the api can use PyType_Ready() to store the methods in the types dictionary.
Python3 removes Py_FindMethod and we should not be using it anyway since its not that efficient.
- Bugfix for running dir() on all BGE python objects. was not getting the immediate methods and attributes for each class.
- Use attributes for KX_Scene (so they are included with dir())
- Override __dict__ attributes for KX_Scene and KX_GameObject so custom properties are included with a dir()
ob.someProp = 10
can now be...
ob["someProp"] = 10
For simple get/set test with an objects 10 properties, this is ~30% faster.
Though I like the attribute access, its slower because it needs to lookup BGE attributes and methods (for parent classes as well as KX_GameObject class).
This could also be an advantage if there are collisions between new attributes added for 2.49 and existing properties a game uses.
Made some other small optimizations,
- Getting and setting property can use const char* as well as STR_String (avoids making new STR_Strings just to do the lookup).
- CValue::SetPropertiesModified() and CValue::SetPropertiesModified(), were looping through all items in the std::map, advancing from the beginning each time.
Deprecated..
getPosition, setPosition, getOrientation, setOrientation, getState, setState, getParent, getVisible, getMass
* swapped set/get to get/set in KX_PYATTRIBUTE_RW_FUNCTION macro to match pythons getsetattrs.
* deprecation warnings in the api and notes in epydocs.
* added 'state' attribute
* gameob.mass = 10 # now works because its not checking only for float values.
* dir(gameob) # includes attributes now
Python dir(ob) for game types now includes attributes names,
* Use "__dict__" rather then "__methods__" attribute to be Python 3.0 compatible
* Added _getattr_dict() for getting the method and attribute names from a PyObject, rather then building it in the macro.
* Added place holder *::Attribute array, needed for the _getattr_up macro.
* Where possible use vec.setValue(x,y,z) to assign values to a vector instead of vec= MT_Vector3(x,y,z), for MT_Point and MT_Matrix types too.
* Comparing TexVerts was creating 10 MT_Vector types - instead compare as floats.
* Added SG_Spatial::SetWorldFromLocalTransform() since the local transform is use for world transform in some cases.
* removed some unneeded vars from UpdateChildCoordinates functions
* Py API - Mouse, Ray, Radar sensors - use PyObjectFrom(vec) rather then filling the lists in each function. Use METH_NOARGS for get*() functions.
- variables that shadow vers declared earlier
- Py_Fatal print an error to the stderr
- gcc was complaining about the order of initialized vars (for classes)
- const return values for ints and bools didnt do anything.
- braces for ambiguous if statements
This was committed in revision 2832 but never accounted for existing object name hashes which existed since revision 2.
Its possible to update the names elsewhere but unlikely anyone ever used this successfully so removing.
* fixed segfaults in CListValue.index(val) and CListValue.count(val) when the pyTypes could not be converted into a CValue.
* added scene.objects to replace scene.getObjectList()
* added function names to PyArg_ParseTuple() so errors will include the function names
* removed cases of PyArg_ParseTuple(args,"O",&pyobj) where METH_O ensures a single argument.
* Made PyObjectFrom use ugly python api rather then Py_BuildValue(), approx %40 speedup for functions that return Python vector and matrix types like ob.orientation.
Use 'const char *' rather then the C++ 'STR_String' type for the attribute identifier of python attributes.
Each attribute and method access from python was allocating and freeing the string.
A simple test with getting an attribute a loop shows this speeds up attribute lookups a bit over 2x.
This patch modifies the way the setParent actuator and KX_GameObject::setParent() function
works when parenting to a compound object: the collision shape of the object being parented
is dynamically added to the coumpound shape.
Similarly, unparenting an object from a compound object will cause the child collision shape
to be dynamically removed from the parent shape provided that is was previously added with
setParent.
Note: * This also works if the object is parented to a child of a compound object: the
collision shape is added to the compound shape of the top parent.
* The collision shape is added with the transformation (position, scale and orientation)
it had at the time of the parenting.
* The child shape is rigidly attached to the compound shape, the transformation is not
affected by any further change in position/scale/orientation of the child object.
* While the child shape is added to the compound shape, the child object is removed from
the dynamic world to avoid superposition of shapes (one for the object itself and
one for the compound child shape). This means that collision sensors on the child
object are disabled while the child object is parent to a compound object.
* There is no difference when setParent is used on a non-compound object: the child
object is automatically changed to a static ghost object to avoid bad interaction
with the parent shape; collision sensors on the child object continue to be active
while the object is parented.
* The child shape dynamically added to a compound shape modifies the inertia of the
compound object but not the mass. It participates to collision detection as any other
"static" child shape.
#18045] [patch] A patch that exposes the rest of the motion functions of KX_GameObject to Python.
*applyForce => setForce
*applyTorque => setTorque
*applyRotation => setDRot
*applyMovement => setDLoc
Rename PHY_GetActiveScene() to KX_GetActiveScene(): more logical name
Add KX_GetActiveEngine()
new KX_KetsjiEngine::GetClockTime(void) to return current
render frame time: if the CPU does not keep up with the
frame rate, up to 5 consecutive logic frames are processed
between each render frame, so that the logic system stays
accurate even if the graphic system is slow. For the video
texture module, it is important to stay in sync with the
render frame: no need to update the texture for logic frame.
BL_Texture::swapTexture(): texture id manipulation
BL_Texture::getTex() : return material texture
Enable video support in ffmpeg for Linux.
added gameObject.replaceMesh(meshname) - needed this for an automatically generated scene where 100's of objects would have needed logic bricks automatically added. Quicker to run replace mesh on all of them from 1 script.
- Reset hit object pointer at end of frame of touch sensor to avoid returning invalid pointer to getHitObject().
- Clear all references in KX_TouchSensor::m_colliders when the sensor is disabled to avoid loose references.
- Test GetSGNode() systematically for all KX_GameObject functions that can be called from python in case a python controller keeps a reference in GameLogic (bad practice anyway).
Implementation of the PHY_IPhysicsController::SetMargin(),
GetMargin(), SetRadius() and GetRadius() for Bullet and Sumo
to allow resetting the Near sensor radius. For bullet use
the new setUnscaledRadius() function to change sphere radius.
In pPreparation of a Fh constraint actuator:
- Add KX_IPhysicsController::GetRadius()
- Fix implementation of KX_BulletPhysicsController::GetVelocity()
(velocity at a point in geometric coordinate)
- Don't try to set velocity on static object (Bullet will assert)
- Add KX_GameObject::GetVelocity() for C access to local velocity
I'm getting this error now:
GPG_Application.cpp: In member function 'void GPG_Application::stopEngine()':
/System/Library/Frameworks/Python.framework/Versions/2.3/include/python2.3/marshal.h:12: error: too many arguments to function 'PyObject* PyMarshal_WriteObjectToString(PyObject*)'
GPG_Application.cpp:720: error: at this point in file
Are we offically not supporint older versions of python now? :)
Kent
the features that are needed to run the game. Compile tested with
scons, make, but not cmake, that seems to have an issue not related
to these changes. The changes include:
* GLSL support in the viewport and game engine, enable in the game
menu in textured draw mode.
* Synced and merged part of the duplicated blender and gameengine/
gameplayer drawing code.
* Further refactoring of game engine drawing code, especially mesh
storage changed a lot.
* Optimizations in game engine armatures to avoid recomputations.
* A python function to get the framerate estimate in game.
* An option take object color into account in materials.
* An option to restrict shadow casters to a lamp's layers.
* Increase from 10 to 18 texture slots for materials, lamps, word.
An extra texture slot shows up once the last slot is used.
* Memory limit for undo, not enabled by default yet because it
needs the .B.blend to be changed.
* Multiple undo for image painting.
* An offset for dupligroups, so not all objects in a group have to
be at the origin.
rayCast(to,from,dist,prop,face,xray,poly):
The face paremeter determines the orientation of the normal:
0 or omitted => hit normal is always oriented towards the ray origin (as if you casted the ray from outside)
1 => hit normal is the real face normal (only for mesh object, otherwise face has no effect)
The ray has X-Ray capability if xray parameter is 1, otherwise the first object hit (other than self object) stops the ray.
The prop and xray parameters interact as follow:
prop off, xray off: return closest hit or no hit if there is no object on the full extend of the ray.
prop off, xray on : idem.
prop on, xray off: return closest hit if it matches prop, no hit otherwise.
prop on, xray on : return closest hit matching prop or no hit if there is no object matching prop on the full extend of the ray.
if poly is 0 or omitted, returns a 3-tuple with object reference, hit point and hit normal or (None,None,None) if no hit.
if poly is 1, returns a 4-tuple with in addition a KX_PolyProxy as 4th element.
The KX_PolyProxy object holds information on the polygon hit by the ray: the index of the vertex forming the poylgon, material, etc.
Attributes (read-only):
matname: The name of polygon material, empty if no material.
material: The material of the polygon
texture: The texture name of the polygon.
matid: The material index of the polygon, use this to retrieve vertex proxy from mesh proxy
v1: vertex index of the first vertex of the polygon, use this to retrieve vertex proxy from mesh proxy
v2: vertex index of the second vertex of the polygon, use this to retrieve vertex proxy from mesh proxy
v3: vertex index of the third vertex of the polygon, use this to retrieve vertex proxy from mesh proxy
v4: vertex index of the fourth vertex of the polygon, 0 if polygon has only 3 vertex
use this to retrieve vertex proxy from mesh proxy
visible: visible state of the polygon: 1=visible, 0=invisible
collide: collide state of the polygon: 1=receives collision, 0=collision free.
Methods:
getMaterialName(): Returns the polygon material name with MA prefix
getMaterial(): Returns the polygon material
getTextureName(): Returns the polygon texture name
getMaterialIndex(): Returns the material bucket index of the polygon.
getNumVertex(): Returns the number of vertex of the polygon.
isVisible(): Returns whether the polygon is visible or not
isCollider(): Returns whether the polygon is receives collision or not
getVertexIndex(vertex): Returns the mesh vertex index of a polygon vertex
getMesh(): Returns a mesh proxy
New methods of KX_MeshProxy have been implemented to retrieve KX_PolyProxy objects:
getNumPolygons(): Returns the number of polygon in the mesh.
getPolygon(index): Gets the specified polygon from the mesh.
More details in PyDoc.
* bugfix for BGE python api - SetParent actuator getObject would segfault if the object was not set.
* Added utility function ConvertPythonToGameObject() that can take a GameObject, string or None and set the game object from this since it was being done in a number of places.
* allow setObject(None), since no object is valid for actuators, Python should be able to set this.
* added optional argument for getObject() so it returns the KX_GameObject rather then its name, would prefer this be default but it could break existing games.
* GBE Python API's alignToVect wasnt clamping the align ammount from 0.0-1.0
* Generated images arnt animated - use for a test to see if the textures animated.
* Material.c - functions for get/setRayTransGlossSamples were not being used.
* BPY_interface.c - removed function GetName(), since everything else just uses id->name+2.
* header_info.c - added ifdef win32 around copy_game_dll since its not needed for other os's yet
With this patch, only sensors that are connected to
active states are actually registered in the logic
manager. Inactive sensors won't take any CPU,
especially the Radar and Near sensors that use a
physical object for the detection: these objects
are removed from the physics engine.
To take advantage of this optimization patch, you
need to define very light idle state when the
objects are inactive: make them transparent, suspend
the physics, keep few sensors active (e,g a message
sensor to wake up), etc.
=======================================
Alpha blending + sorting was revised, to fix bugs and get it
to work more predictable.
* A new per texture face "Sort" setting defines if the face
is alpha sorted or not, instead of abusing the "ZTransp"
setting as it did before.
* Existing files are converted to hopefully match the old
behavior as much as possible with a version patch.
* On new meshes the Sort flag is disabled by the default, to
avoid unexpected and hard to find slowdowns.
* Alpha sorting for faces was incredibly slow. Sorting faces
in a mesh with 600 faces lowered the framerate from 200 to
70 fps in my test.. the sorting there case goes about 15x
faster now, but it is still advised to use Clip Alpha if
possible instead of regular Alpha.
* There still various limitations in the alpha sorting code,
I've added some comments to the code about this.
Some docs at the bottom of the page:
http://www.blender.org/development/current-projects/changes-since-246/realtime-glsl-materials/
Merged some fixes from the apricot branch, most important
change is that tangents are now exactly the same as the rest
of Blender, instead of being computed in the game engine with a
different algorithm.
Also, the subversion was bumped to 1.
First batch of optimizaton of the bullet adaptation layer in the BGE.
- remove circular motion state update.
- optimization of physic adaptation layer for bullet: bypass
unecessary conversion of rotation matrix to quaternion and back.
- remove double updates during object replication.
Certain actuators hold a pointer to an objects: Property,
SceneCamera, AddObject, Camera, Parent, TractTo. When a
group is duplicated, the actuators that point to objects
within the group will be relinked to point to the
replicated objects and not to the original objects.
This helps to setup self-contained group with a camera
following a character for example.
This feature also works when adding a single object
(and all its children) with the AddObject actuator.
The second part of the patch extends the protection
against object deletion to all the actuators of the above
list (previously, only the TrackTo, AddObject and
Property actuators were protected). In case the target
object of these actuators is deleted, the BGE won't
crash.
=============================
* Clean up and optimizations in skinned/deformed mesh code.
* Compatibility fixes and clean up in the rasterizer.
* Changes related to GLSL shadow buffers which should have no
effect, to keep the code in sync with apricot.
(can be improved to rotate correctly but for our use ist ok for now, would also be useful to have an argument to clamp the maximum rotation angle to get a constant rotation speed),
This will used to make franky upright when falling from an angle, to track to a surface when hanging onto a ledge and setting the glide pitch.
Without this rotation is instant and jerky.
currently this is done with Mathutils which isnt available in Blender Player.
def do_rotate_up(own):
own.alignAxisToVect([0,0,1], 2, 0.1)
replaces...
def do_rotate_up(own):
up_nor = Vector(0,0,1)
own_mat = Matrix(*own.getOrientation()).transpose()
own_up = up_nor * own_mat
ang = AngleBetweenVecs(own_up, up_nor)
if ang > 0.005:
# Set orientation
cross = CrossVecs(own_up, up_nor)
new_mat = own_mat * RotationMatrix(ang*0.1, 3, 'r', cross)
own.setOrientation(new_mat.transpose())
M source/gameengine/Ketsji/KX_GameObject.cpp
M source/gameengine/Ketsji/KX_GameObject.h
New Add mode for Ipo actuator
=============================
A new Add button, mutually exclusive with Force button, is available in
the Ipo actuator. When selected, it activates the Add mode that consists
in adding the Ipo curve to the current object situation in world
coordinates, or parent coordinates if the object has a parent. Scale Ipo
curves are multiplied instead of added to the object current scale.
If the local flag is selected, the Ipo curve is added (multiplied) in
the object's local coordinates.
Delta Ipo curves are handled identically to normal Ipo curve and there
is no need to work with Delta Ipo curves provided that you make sure
that the Ipo curve starts from origin. Origin means location 0 for
Location Ipo curve, rotation 0 for Rotation Ipo curve and scale 1 for
Scale Ipo curve.
The "current object situation" means the object's location, rotation
and scale at the start of the Ipo curve. For Loop Stop and Loop End Ipo
actuators, this means at the start of each loop. This initial state is
used as a base during the execution of the Ipo Curve but when the Ipo
curve is restarted (later or immediately in case of Loop mode), the
object current situation at that time is used as the new base.
For reference, here is the exact operation of the Add mode for each
type of Ipo curve (oLoc, oRot, oScale, oMat: object's loc/rot/scale
and orientation matrix at the start of the curve; iLoc, iRot, iScale,
iMat: Ipo curve loc/rot/scale and orientation matrix resulting from
the rotation).
Location
Local=false: newLoc = oLoc+iLoc
Local=true : newLoc = oLoc+oScale*(oMat*iLoc)
Rotation
Local=false: newMat = iMat*oMat
Local=true : newMat = oMat*iMat
Scale
Local=false: newScale = oScale*iScale
Local=true : newScale = oScale*iScale
Add+Local mode is very useful to have dynamic object executing complex
movement relative to their current location/orientation. Of cource,
dynamics should be disabled during the execution of the curve.
Several corrections in state system
===================================
- Object initial state is taken into account when adding object
dynamically
- Fix bug with link count when adding object dynamically
- Fix false on-off detection for Actuator sensor when actuator is
trigged on negative event.
- Fix Parent actuator false activation on negative event
- Loop Ipo curve not restarting at correct frame when start frame is
different from one.
General
=======
- Removal of Damp option in motion actuator (replaced by
Servo control motion).
- No PyDoc at present, will be added soon.
Generalization of the Lvl option
================================
A sensor with the Lvl option selected will always produce an
event at the start of the game or when entering a state or at
object creation. The event will be positive or negative
depending of the sensor condition. A negative pulse makes
sense when used with a NAND controller: it will be converted
into an actuator activation.
Servo control motion
====================
A new variant of the motion actuator allows to control speed
with force. The control if of type "PID" (Propotional, Integral,
Derivate): the force is automatically adapted to achieve the
target speed. All the parameters of the servo controller are
configurable. The result is a great variety of motion style:
anysotropic friction, flying, sliding, pseudo Dloc...
This actuator should be used in preference to Dloc and LinV
as it produces more fluid movements and avoids the collision
problem with Dloc.
LinV : target speed as (X,Y,Z) vector in local or world
coordinates (mostly useful in local coordinates).
Limit: the force can be limited along each axis (in the same
coordinates of LinV). No limitation means that the force
will grow as large as necessary to achieve the target
speed along that axis. Set a max value to limit the
accelaration along an axis (slow start) and set a min
value (negative) to limit the brake force.
P: Proportional coefficient of servo controller, don't set
directly unless you know what you're doing.
I: Integral coefficient of servo controller. Use low value
(<0.1) for slow reaction (sliding), high values (>0.5)
for hard control. The P coefficient will be automatically
set to 60 times the I coefficient (a reasonable value).
D: Derivate coefficient. Leave to 0 unless you know what
you're doing. High values create instability.
Notes: - This actuator works perfectly in zero friction
environment: the PID controller will simulate friction
by applying force as needed.
- This actuator is compatible with simple Drot motion
actuator but not with LinV and Dloc motion.
- (0,0,0) is a valid target speed.
- All parameters are accessible through Python.
Distance constraint actuator
============================
A new variant of the constraint actuator allows to set the
distance and orientation relative to a surface. The controller
uses a ray to detect the surface (or any object) and adapt the
distance and orientation parallel to the surface.
Damp: Time constant (in nb of frames) of distance and
orientation control.
Dist: Select to enable distance control and set target
distance. The object will be position at the given
distance of surface along the ray direction.
Direction: chose a local axis as the ray direction.
Range: length of ray. Objecgt within this distance will be
detected.
N : Select to enable orientation control. The actuator will
change the orientation and the location of the object
so that it is parallel to the surface at the vertical
of the point of contact of the ray.
M/P : Select to enable material detection. Default is property
detection.
Property/Material: name of property/material that the target of
ray must have to be detected. If not set, property/
material filter is disabled and any collisioning object
within range will be detected.
PER : Select to enable persistent operation. Normally the
actuator disables itself automatically if the ray does
not reach a valid target.
time : Maximum activation time of actuator.
0 : unlimited.
>0: number of frames before automatic deactivation.
rotDamp: Time constant (in nb of frame) of orientation control.
0 : use Damp parameter.
>0: use a different time constant for orientation.
Notes: - If neither N nor Dist options are set, the actuator
does not change the position and orientation of the
object; it works as a ray sensor.
- The ray has no "X-ray" capability: if the first object
hit does not have the required property/material, it
returns no hit and the actuator disables itself unless
PER option is enabled.
- This actuator changes the position and orientation but
not the speed of the object. This has an important
implication in a gravity environment: the gravity will
cause the speed to increase although the object seems
to stay still (it is repositioned at each frame).
The gravity must be compensated in one way or another.
the new servo control motion actuator is the simplest
way: set the target speed along the ray axis to 0
and the servo control will automatically compensate
the gravity.
- This actuator changes the orientation of the object
and will conflict with Drot motion unless it is
placed BEFORE the Drot motion actuator (the order of
actuator is important)
- All parameters are accessible through Python.
Orientation constraint
======================
A new variant of the constraint actuator allows to align an
object axis along a global direction.
Damp : Time constant (in nb of frames) of orientation control.
X,Y,Z: Global coordinates of reference direction.
time : Maximum activation time of actuator.
0 : unlimited.
>0: number of frames before automatic deactivation.
Notes: - (X,Y,Z) = (0,0,0) is not a valid direction
- This actuator changes the orientation of the object
and will conflict with Drot motion unless it is placed
BEFORE the Drot motion actuator (the order of
actuator is important).
- This actuator doesn't change the location and speed.
It is compatible with gravity.
- All parameters are accessible through Python.
Actuator sensor
===============
This sensor detects the activation and deactivation of actuators
of the same object. The sensor generates a positive pulse when
the corresponding sensor is activated and a negative pulse when
it is deactivated (the contrary if the Inv option is selected).
This is mostly useful to chain actions and to detect the loss of
contact of the distance motion actuator.
Notes: - Actuators are disabled at the start of the game; if you
want to detect the On-Off transition of an actuator
after it has been activated at least once, unselect the
Lvl and Inv options and use a NAND controller.
- Some actuators deactivates themselves immediately after
being activated. The sensor detects this situation as
an On-Off transition.
- The actuator name can be set through Python.
* 2 returning errors without exception set another return None instead of NULL.
* a missing check for parent relation
* BPY matrix length was incorrect in matrix.c, this change could break some scripts, however when a script expects a list of lists for a matrix, the len() function is incorrect and will give an error. This was the only thing stopping apricot game logic running in trunk.
Also added a function for GameObjects - getAxisVect(vec), multiplies the vector be the objects worldspace rotation matrix. Very useful if you want to know what the forward direction is for an object and dont want to use Blender.Mathutils which is tedious and not available in BlenderPlayer yet.
This is also needed for removing any force that existed before suspending dynamics - In the case of franky hanging, resuming dynamics when he fell would apply the velocity he had when grabbing making dropping to the ground work unpredictably.
Also note in pydocs that enable/disable rigidbody physics doesn't work with bullet yet.
* Action FrameProp was checking if the string was true, not that it contained any text.
* Added GameObject.getVisible() since there is already a getVisible
* Added GameObject.getPropertyNames() Needed in apricot so Franky can collect and throw items in the level without having the names defined elsewhere or modifying his game logic which is stored in a separate blend file.
This patch introduces two options for the motion actuator:
damping: number of frames to reach the target velocity. It takes
into account the startup velocityin the target velocity direction
and add 1/damping fraction of target velocity until the full
velocity is reached. Works only with linear and angular velocity.
It will be extended to delta and force motion method in a future
release.
clamping: apply the force and torque as long as the target velocity
is not reached. If this option is set, the velocity specified
in linV or angV are not applied to the object but used as target
velocity. You should also specify a force in force or torque field:
the force will be applied as long as the velocity along the axis of
the vector set in linV or angV is not reached. Works best in low
friction environment.
Shape Action are now supported in the BGE. A new type of actuator "Shape Action" is available on mesh objects. It can be combined with Action actuator on parent armature. Only relative keys are supported. All the usual action options are available: type, blending, priority, Python API. Only actions with shape channels should be specified of course, otherwise the actuator has no effect. Shape action will still work after a mesh replacement provided that the new mesh has compatible shape keys.
GLEW
====
Added the GLEW opengl extension library into extern/, always compiled
into Blender now. This is much nicer than doing this kind of extension
management manually, and will be used in the game engine, for GLSL, and
other opengl extensions.
* According to the GLEW website it works on Windows, Linux, Mac OS X,
FreeBSD, Irix, and Solaris. There might still be platform specific
issues due to this commit, so let me know and I'll look into it.
* This means also that all extensions will now always be compiled in,
regardless of the glext.h on the platform where compilation happens.
Game Engine
===========
Refactoring of the use of opengl extensions and other drawing code
in the game engine, and cleaning up some hacks related to GLSL
integration. These changes will be merged into trunk too after this.
The game engine graphics demos & apricot level survived my tests,
but this could use some good testing of course.
For users: please test with the options "Generate Display Lists" and
"Vertex Arrays" enabled, these should be the fastest and are supposed
to be "unreliable", but if that's the case that's probably due to bugs
that can be fixed.
* The game engine now also uses GLEW for extensions, replacing the
custom opengl extensions code that was there. Removes a lot of
#ifdef's, but the runtime checks stay of course.
* Removed the WITHOUT_GLEXT environment variable. This was added to
work around a specific bug and only disabled multitexturing anyway.
It might also have caused a slowdown since it was retrieving the
environment variable for every vertex in immediate mode (bug #13680).
* Refactored the code to allow drawing skinned meshes with vertex
arrays too, removing some specific immediate mode drawing functions
for this that only did extra normal calculation. Now it always splits
vertices of flat faces instead.
* Refactored normal recalculation with some minor optimizations,
required for the above change.
* Removed some outdated code behind the __NLA_OLDDEFORM #ifdef.
* Fixed various bugs in setting of multitexture coordinates and vertex
attributes for vertex arrays. These were not being enabled/disabled
correct according to the opengl spec, leading to crashes. Also tangent
attributes used an immediate mode call for vertex arrays, which can't
work.
* Fixed use of uninitialized variable in RAS_TexVert.
* Exporting skinned meshes was doing O(n^2) lookups for vertices and
deform weights, now uses same trick as regular meshes.
The current layer information is now stored in KX_GameObject and inherited from the parent object when dynamically added. This information is used during the rendering the select the lamps. As the selected lamps are always coming from active layers, their position and orientation are correct.
This patch consists in new KX_GameObject::SetParent() and KX_GameObject::RemoveParent() functions to create and destroy parent relation during game. These functions are accessible through python and through a new actuator KX_ParentActuator. Function documentation in PyDoc.
The object keeps its orientation, position and scale when it is parented but will further rotate, move and scale with its parent from that point on. When the parent relation is broken, the object keeps the orientation, position and scale it had at that time.
The function has no effect if any of the X/Y/Z scale of the object or its new parent are below Epsilon.
rayCastTo(other,dist,prop)
Look towards another point/KX_GameObject and return first object hit within dist with a property that match prop, None if no object found or if it does not match prop.
Parameters:
other = 3-tuple (xyz coordinates) or object reference (target=center of object)
(type = list [x,y,z] or object reference)
dist = max distance of detection (can be negative => look behind)
If 0 or omitted => detect up to other
(type=float)
prop = property name that object must have
If empty or omitted => detect any object
(type=string)
AddObject actuator forces last created object to hang in memory even after object is removed from scene => bad link between object and physic controller that causes Blender to crash in case a python script tries to use it (bad programming anyway).
This patch avoids the crash by maintaining consistent links at all time.
Armatures are back
Split screen
Double sided lightning
Ambient lighting
Alpha test
Material IPO support (one per object atm)
Blender materials
GLSL shaders - Python access
Up to three texture samplers from the material panel ( 2D & Cube map )
Python access to a second set of uv coordinates
See http://www.elysiun.com/forum/viewtopic.php?t=58057
added raycast support for bullet (no triangle-mesh support, soon)
added python methods for 'getHitObject', getRayDirection, getHitPosition and getHitNormal for mouse over sensor,
which makes it easy for a shootout.blend demo :)
from brian hayward (bthayward)
Detailed description:
Currently, when an armature deformed object's mesh is replaced by the ReplaceMesh actuator, the new mesh fails to deform to the armature's movement.
My patch fixes this by properly replacing the deform controller along with the mesh (when appropriete).
For instance, if one had an animated character using any of the standard deformation techniques (armature, ipo, RVK, or AVK), that character's mesh would currently be prevented from changing mid-game. It could be replaced, but the new mesh would lack the controller which tells it how to deform. If one wanted to dynamiclly add a hat on top of the character's head, it would require storing a secondary prebuilt character (mesh, armature, logic, ect...) on another layer FOR EACH HAT the character could possibly wear, then swapping out the whole character when the hat change was desired. So if you had 4 possible hat/character combos, you would have 4 character meshes, 4 armatures, 4 sets of logic, and so on. I find this lack of flexibility to be unresonable.
With my patch, one could accomplish the same thing mearly by making one version of the character in the main layer, and adding an invisible object atop the character's head (which is parented to the head bone). Then whenever it becomes desirable, one can replace the invisible object's mesh with the desirable hat's mesh, then make it visible. With my patch, the hat object would then continue to deform to the character's head regardless of which hat was currently being worn.
*note 1*
for armature/mesh deformations, the new mesh must have properly assigned vertex groups which match one or more of the bones of the target armature before the replaceMesh call is made. Otherwise the vertices won't react to the armature because they won't know how. (not sure if vertices can be scripted to change groups after the game has started)
*note 2*
The added processing time involved with replacing the object's deform controller is negligible.
Profiling revealed that the SceneGraph updated every physics object, whether it moved or not, even though the physics object was at the right place. This would cause SOLID to go and update its bounding boxes, overlap tests etc.
This callback handles the special case (parented objects) where the physics scene needs to be informed of changes to the scenegraph.
Added Python attributes (mass, parent, visible, position, orientation, scaling) to the KX_GameObject module.
Make KX_GameObject use the KX_PyMath Python <-> Moto conversion.
Consider:
gameobj->getClientInfo()->m_auxilary_info = (matname ? (void*)(matname+2) : NULL);
It works if matname is "MAblah", but not if matname is "".
2. Added constructor for struct RAS_CameraData.
3. Added initializers to the struct KX_ClientObjectInfo constructor
4. Collision sensors won't detect near sensors.
5. A stack of minor tweaks, adjusting whitespace, using ++it for stl stuff.
[SCons] Build with Solid as default when enabling the gameengine in the build process
[SCons] Build solid and qhull from the extern directory and link statically against them
That was about it.
There are a few things that needs double checking:
* Makefiles
* Projectfiles
* All the other systems than Linux and Windows on which the build (with scons) has been successfully tested.