worked properly with modifiers. Needs more testing I am sure.
No, honestly, I wasn't just cleaning for the hell of it, it
was *necessary* (I would never do such a thing). Selection should
work completely with cage options of modifiers now.
- added DerivedMesh foreach functions to iterate over mapped
verts/edges/face centers. These replaced some of the drawing
functions and are more general anyway. Special edge drawing
functions remain for performance reasons.
- removed EditFace xs, ys fields
- added general functions to iterate over screen coordinates of
mesh/curve/lattice objects
- removed all calc_*verts* functions that were used for storing
screen coordinates in objects. they were recalc'd on the fly
for most situations anyway, so now we just always do that.
calc_*verts_ext was one of those calls that did dirty things
deep down in the callstack (changing curarea and poking at
matrices)
- rewrote all vertex level selection routines (circle, lasso, bbox)
and closest vertex routines (rightmouse select) to use the new
system. This cleaned up the selection code a lot and the structure
of selection is much easier to see now. This is good for future
work on allowing modifiers to completely override the selection
system. It also points out some discrepancies in the way selection
is handled that might be nice to resolve (mesh vertex selection has
fancy stuff to try to help with selecting overlapping, but it only
works w/o bbuf select, and curves/lattices don't have at all).
- had to remove ton's code to move Manipulator to cage location, this
is not reliable (can come up with a different method if requested)
- as it happens BezTriple.s and BPoint.s are basically available to
be removed, just need to rewrite editipo code that still does
background calc of screen coordinates
- MVert.{xs,ys} are still around because they are abused in some places
for other info (not sure if this is safe actually, since they are
short's and the mvert limit went up).
And did I mention this commit is comes out to -305 lines? Well it does.
DLM to share data from DerivedMesh (reduces some copying/memory allocation)
- added displistmesh_copyShared function to copy a DLM but not duplicate any
internal data
- changed crease drawing to use DerivedMesh functions... this means varying
edge width style of creases had to go, I replaced by using varying color to
show crease weight instead. Don't think this is a big loss since the subsurf
result gives you a much better indication of the crease weight anyway.
- bug fix in mirror modifier, didn't copy edge creases from editmesh correctly
- remove python access to Optimal and Subsurf flags (they don't
work this way anymore, I suppose need to replace with python
access to modifiers but not going to do right now).
- removed interface access to OPTIMAL mode, needs to be rethough...
this means at the moment subsurfs outside editmode always draw
and render all edges
Blender.NMesh.GetRawFromObject through a displist conversion method as used by
Blender when converting them in the UI.
Notes: Objects with only edges (3D curves/polyline without bevel) do not have
normals, so they are all initialised to (1, 0, 0) on conversion
Converting from meta objects only work on the "mother ball". That is,
the object with the lower base name.
Example: "meta" for all the "meta.*" objects.
Meshes extracted from curve based objects (Font/2D filled curves)
contain both the filled surfaces and the outlines of the shapes.
Materials are taken from the object's material list. Material handling
in NMesh is incorrect anyway, as it always uses the materials from the
mesh, ignoring the setting in ob->colbits.
This patch also makes the include order a little clearer.
A couple of warnings have been fixed by using better types:
- Using char instead of short when parsing color values.
The "constructor" expects and uses char anyway.
- Explicit casting to short when storing normals back in mvert.
- Changing constant doubles to floats with "f" to make compiler happy.
The only warning left regards NMFace.flag which is stored as a short but is
used to fill in TFace.flag which is a char. I didn't want to change the
object's structure so I left it like that. I didn't add an explicit cast when
putting it back in TFace so that the warning can remind us that there might be
something to change there.
- Jean-Michel Soler: 1) paths import -- ai module; 2) tex2uvbaker;
- Campbell Barton: obj exporter.
BPython:
- Campbell also provided a patch (+docs) to add shader data access to Blender.Material.
Again, thanks guys and sorry for the long delay.
with MSVS 8)
- broke mesh_create_shadedColors out of shadeDispList, used to
build vertex colors for mesh in vpaint as well (also fixed
bug where they were not initialized correctly for subsurfs)
- added modifier_copyData and modifier_findByType functions
- change editmode modifiers to only calculate if Realtime and
Editmode bits are both set, makes more sense for copying
modifiers
- update object_copy to correctly copy modifiers
- removed duplicate redefinition of ME_ attributes in python,
this is a horrible idea, why was it done in the first place?
- update armature auto vertex group code to check for subsurf
in modifier stack
- fixed flip_subdivision to work with move to modifier stack
- added copymenu_modifiers, can copy all modifiers or just
data from first modifier of a certain type (not sure how
to deal with multiple modifiers of same type... not
a big issue though I think)
- PoseMode: Wkey menu, "Flip Left/Right Names". On selected bones, it flips
the L/R tags in names, and calls the proper code to rename everything
that's related (constraint targets, bone-childs, etc).
- PoseMode: Shift+S snapmenu: snap cursor to selected now works
- Outliner: select bones now correctly sets 'active' flag for bones,
updating the UI as well.
Also made sure you cannot select hidden bones in outliner.
- 3DWindow: in PoseMode/EditMode draws name of active Bone too (with user
option "Draw active object name" set.
- Added the new Armature/PoseMode options in View3D pulldowns.
Cleanup:
- moved Pose code from editaction.c to poseobject.c
- removed BSE_editaction.h and BSE_editaction_types.h, moved contents of
it to BIF_editaction.h. One include per C file should be fine. :)
I know the src/ structure would require more elaborated includes, but
we don't have that now...
- added decimate modifier & removed old decimate interface
(currently lacks warning about destroying data, and there needs
to be a way for modifiers to return errors back to the interface)
- allow applyModifier to return NULL to indicate error
- unfortunately new decimate modifier means it does not know exact
number of faces in mesh (other modifiers may come before) and so
instead interface uses a percentage. if people need exact face
count slider then I will have to think of some hack to fit this
in. note that it does display the output face count so its possible
to tweak the pct to get what you want regardless.
- removed python Wave object
If you are bored now how much easier it is to implement something
like decimate as a modifier. Very few changes to interface, very
few entry points.
to normalised coordinate (convention in blender, helps with
halo)
- removed vertexnormals(), vertexnormals_mesh()
- removed CTX_NO_NOR_RECALC (always assume already calculated)
- change NMesh.c to call mesh_calc_normals
- chance load_editMesh to call mesh_calc_normals after done
converting instead of using editmesh normals
- update recalc_editnormals to also calc vertex normals (whats
4 more adds and a sqrt among friends)
Its hard to believe, but it just might be the case that there
are only two places mesh normals are calculated now (renderer
and kernel)
and this is better left to user (whee this was a fun commit! so
much deleting!)
- removed mesh_calculate_vertex_normals (replaced by mesh_calc_normals)
- adds a new point class
* point/ vector math (p + v = p, p - p = v, etc.)
* points can be transformed by matrices/quats
* wraps 'place vector' type vectors that have no magnitude
- wrapped toXXX() methods work correctly
* toXXX() will NOT wrap data (this is due to the fact that wrapped data cannot be converted)
* added a 'wrapped' attribute to mathutils classes to determine wether the object is accessing python or blender data
- added the ability to negate vectors/points with "-vec"
* deprecated vector.negate()
- added the ability to shorhand inverse matrices with "~mat" (tilde)
- conversion between vector/point with toXXX() methods
to write one that is based on geometry (and not just vertex position)
- added editmode versions of modifier deform/apply calls and flag
to tag modifiers that support editmode
- added isFinalCalc param to applyModifier, basically a switch to let
subsurf know if it is calc'ng orco or not (so it can deal with cache
appropriately). This is kinda hacky and perhaps I can come up with
a better solution (its also a waste to do a complete subdivide just
to get vertex locations).
- changed ccgsubsurf to not preallocate hash's to be approximately correct
size... this was probably not a big performance savings but means that
the order of faces returned by the iterator can vary after the first
call, this messes up orco calculation so dropped for time being.
- minor bug fix, meshes with only key didn't get vertex normals correctly
calc'd
- updated editmesh derivedmesh to support auxiliary locations
- changed mesh_calc_modifiers to alloc deformVerts on demand
- added editmesh_calc_modifiers for calculating editmesh cage and final
derivedmesh's
- bug fix, update shadedisplist to always calc colors (even if totvert==0)
- changed load_editMesh and make_edge to build me->medge even if totedge==0
(incremental subsurf checks this)
todo: add drawFacesTex for ccgderivedmesh
So, modifiers in editmode are back (which means auto-mirror
in edit mode works now) although still not finished. Currently
no cage is computed, the cage is always the base mesh (in
other words, Optimal edge style editing is off), and the final
mesh currently includes all modifiers that work in edit mode
(including lattice and curve). At some point there will be toggles
for which modifiers affect the final/cage editmode derivedmesh's.
Also, very nice new feature is that incremental subsurf in object
mode returns a ccgderivedmesh object instead of copying to a new
displistmesh. This can make a *huge* speed difference, and is very
nice for working with deformed armatures (esp. with only small
per frame changes).
an object) but this is not going to work... I can't remember the reason
I did it this way in the first place either! oops! regardless, switch
to all mesh_ derived accessors taking object argument. there is still
a bug in render orco calculation though. (hunt hunt)
- removed python files that should have been ditched in previous commit
based on time change. would be nice if dep graph could handle this.
- made dep check if modifiers need update on time change
- fix render crash (access null)
- added new Build Effect modifier type. compared to old one works as
a full member of modifier system, means can apply subsurf, etc on
it, reorder, what have you. and it is all nice and self contained.
- removed old Build effect, old files convert to new style on load
- couldn't help myself, added a randomize feature to build effect
- removed Python BuildEffect support
- removed mesh_deform (merge into mesh_modifier)
- switch python lattice_apply function to use object_apply_deform,
this isn't exactly equivalent but the python system shouldn't
have been calling that deep into the kernel anyway.
New feature: Modifier stack
- added Object.modifiers (list of ModifierData elements)
- added DNA_modifier_types.h
o contains type definition for the file data for the various
modifier types
- added BKE_modifier.h
o contains modifierType_get_info (access to modifier type registry)
o structs and defines for runtime modifier usage
- updated mesh_calc_modifiers to evaluate modifier stack (note that
for the time being it also evaluates the old style modifiers so files
should load and work as normal).
- add file handling modifier code (todo: don't replicate on object copy)
- add modifier stack UI code (lives in object panel)
Only real new feature at the moment is that you can apply lattices and
curves *after* a subdivision surface which was never possible before.
Todo:
- DEP graph updating does not work correctly yet, so you generally have
to tab cycle to see results.
- editmode calculation does not use modifier stack.
- bug fixes (there must be a few in there somewhere)
- internal methods for mathutils class need to incref before returning a pointer to self
* memory allocation errors (freeing bad pointers) were appearing after repeatedly calling internal methods on the same object due to ref counts.
- changed mesh_modifier, sbObjectStep, object_deform to take vertexCo
argument instead of operating on mesh
- fixed bug where a derived mesh would not be returned in editmode
- removed object_wave, replaced by init_wave_deform and calc_wave_deform
- moved cached DerivedMesh to Object, not Mesh... fixes heisenbugs
with linked objects
builds new DerivedMesh... caching can come later)
- split DerivedMesh returning functions into editmesh and mesh groups
- got rid of DL_NORS displist type (get built on fly for mesh when
needed)
- got rid of Mesh.disp (yay!)
- started to punch DerivedMesh returning functions into shape to introduce
modifier stack
- Mostly this cleans up the #includes and header files in the python project.
- Warning fixes are mostly casting issues and misc fixes. General warning clean up.
- #include Python.h MUST come as the first include to avoid the POSIX redefine warning in the unix makefiles
- fno-strict-aliasing flag added to makefile to fix a unavoidable type punning warning in types.c
- appropriate callers of makeDispList replaced with depgraph calls
- unappropriate places just killed... small chance this gives some
errors in corner cases if dep graph isn't notified (example, font
family displists) but these can be tracked down as they show up.
- still a large number of callers of makeDispListCurveTypes, but
makeDispListMesh has just a few.
This adds two new method to the BPY Object type:
- makeParentDeform
- makeParentVertex
Both are based on makeParent. In fact, I splitted off the parenting
part of makeParent into an internal function so that all can reuse it. I
also added DEP_graph refresh flag that where missing.
makeParentDeform sets the parent type accordingly for correct
deformation (armature and curve) and checks if the parent and childs are of the
correct type for deformation.
makeParentVertex takes as additional mandatory argument a tuple of
indices to tell the parent which vertex to use. It does some object type
checking too.
Also included documentation. :)
Test file is in the patch there: http://projects.blender.org/tracker/?func=detail&atid=127&aid=2833&group_id=9
Note to Ton: When doing vertex parenting, the dashed parent line is only updated correctly if the objects are on screen when calling the function. If they aren't, the line is wrong and is only corrected when another recalc is called on the parent or child.
which calculates texspace on demand if need be.
- removed almost all calls to tex_space_mesh
There may be a few corner cases where this goes wrong (meshes with vertex
keys) but these should get ironed out by coming modifier system.
- switch all mesh boundbox access to go through mesh_get_bb
- switch object_handle_update to call mesh_changed instead of making
the displist data immediately (delayed calculation)
still a makeDispList that dispatches to the appropriate one.
makeDispList is on the way out and this makes it easier to track down
exactly which places use makedispList and for what types of objects.
- switch calls to makeDispList to appropriate more specific function (if
the object type is known by caller).
- added mesh_changed function that invalidates cached mesh data (but does
not rebuild, mesh data gets rebuilt on access). Most old calls to
makeDispListMesh use this instead now.
- also included is some fixes for preprocessor inclues and some clean up of the previous commit
-rewrite and bugfixes
----------------------------------
Here's my changelog:
-fixed Rand() so that it doesn't seed everytime and should generate better random numbers
- changed a few error return types to something more appropriate
- clean up of uninitialized variables & removal of unneccessary objects
- NMesh returns wrapped vectors now
- World returns wrapped matrices now
- Object.getEuler() and Object.getBoundingBox() return Wrapped data when data is present
- Object.getMatrix() returns wrapped data if it's worldspace, 'localspace' returns a new matrix
- Vector, Euler, Mat, Quat, call all now internally wrap object without destroying internal datablocks
- Removed memory allocation (unneeded) from all methods
- Vector's resize methods are only applicable to new vectors not wrapped data.
- Matrix(), Quat(), Euler(), Vector() now accepts ANY sequence list, including tuples, list, or a self object to copy - matrices accept multiple sequences
- Fixed Slerp() so that it now works correctly values are clamped between 0 and 1
- Euler.rotate does internal rotation now
- Slice assignment now works better for all types
- Vector * Vector and Quat * Quat are defined and return the DOT product
- Mat * Vec and Vec * Mat are defined now
- Moved #includes to .c file from headers. Also fixed prototypes in mathutils
- Added new helper functions for incref'ing to genutils
- Major cleanup of header files includes - include Mathutils.h for access to math types
- matrix.toQuat() and .toEuler() now fixed take appropriate matrix sizes
- Matrix() with no parameters now returns an identity matrix by default not a zero matrix
- printf() now prints with 6 digits instead of 4
- printf() now prints output with object descriptor
- Matrices now support [x][y] assignment (e.g. matrix[x][y] = 5.4)
- Matrix[index] = value now expectes a sequence not an integer. This will now set a ROW of the matrix through a sequence. index cannot go above the row size of the matrix.
- slice operations on matrices work with sequences now (rows of the matrix) example: mymatrix[0:2] returns a list of 2 wrapped vectors with access to the matrix data.
- slice assignment will no longer modify the data if the assignment operation fails
- fixed error in matrix * scalar multiplication
- euler.toMatrix(), toQuat() no longer causes "creep" from repeated use
- Wrapped data will generate wrapped objects when toEuler(), toQuat(), toMatrix() is used
- Quats can be created with angle/axis, axis/angle
- 4x4 matrices can be multiplied by 3D vectors (by popular demand :))
- vec *quat / quat * vec is now defined
- vec.magnitude alias for vec.length
- all self, internal methods return a pointer to self now so you can do print vector.internalmethod() or vector.internalmethod().nextmethod() (no more print matrix.inverse() returning 'none')
- these methods have been deprecated (still functioning but suggested to use the corrected functionality):
* CopyVec() - replaced by Vector() functionality
* CopyMat() - replaced by Matrix() functionality
* CopyQuat() - replace by Quaternion() functionality
* CopyEuler() - replaced by Euler() functionality
* RotateEuler() - replaced by Euler.rotate() funtionality
* MatMultVec() - replaced by matrix * vector
* VecMultMat() - replaced by vector * matrix
- New struct containers references to python object data or internally allocated blender data for wrapping
* Explaination here: math structs now function as a 'simple wrapper' or a 'py_object' - data that is created on the fly will now be a 'py_object' with its memory managed by python
* otherwise if the data is returned by blender's G.main then the math object is a 'simple wrapper' and data can be accessed directly from the struct just like other python objects.
- Jean-Michel Soler: paths import -- ai and svg modules;
- Jean-Baptiste PERIN: bvh to armatures (note: should not work until we re-wrap armatures in bpython);
- Campbell Barton: obj importer.
Thanks guys, excuse me the delay.
- tiny doc update.
This fixes the active object in place when orbiting the view.
Choppy 15fps demo can be seen there: http://www.elysiun.com/~theeth/bf/around_active.html
Image Memory Grabage Collection
This adds memory handling to the image code. An image is tagged each time it is used.
During a collection cycle (frequency of cycles is user defined), if an image is older
than a user defined limit, its buffer gets deallocated. This also applies to gl memory buffers.
Images that are loading in GL memory needs to go through two time outs before being fully deallocated: the first time out deallocated the gl memorry, the second the buffer in ram.
Notes:
Image buffer loaded from python gets tagged as permanent upon load. That tag is removed when python stops using the image.
I might have missed some tagging spots, especially in the rendering pipeline. Someone with more knowledge about this code should check to be careful.
Tagging is done on every access, for rendering, this will probably be a performance hit. A scheme should be developped to only tag when the rendering is completed.
Collecting is called in draw_object, most likely not the best place to do it.
Safe from undo, since using undo deallocates memory anyway (like when loading a blend file with one currently opened)
Userpref DNA changes:
I've changed a couple of flagging variables from short to int. Some because they needed more space, others to keep SDNA happy.
Info window changes:
I've grouped a couple of buttons in aligned blocks and changed the color of mutually exclusive options to make them clearer.
Matt didn't do any changes on that in tuhopuu, so hopefully I'm not stepping on anyone's feet with this.
Also changed double constants into floats with f in a couple of places (mostly space.c) to make compiler happier.
Aim was to get a total refresh of the animation system. This
is needed because;
- we need to upgrade it with 21st century features
- current code is spaghetti/hack combo, and hides good design
- it should become lag-free with using dependency graphs
A full log, with complete code API/structure/design explanation
will follow, that's a load of work... so here below the list with
hot changes;
- The entire object update system (matrices, geometry) is now
centralized. Calls to where_is_object and makeDispList are
forbidden, instead we tag objects 'changed' and let the
depgraph code sort it out
- Removed all old "Ika" code
- Depgraph is aware of all relationships, including meta balls,
constraints, bevelcurve, and so on.
- Made depgraph aware of relation types and layers, to do smart
flushing of 'changed' events. Nothing gets calculated too often!
- Transform uses depgraph to detect changes
- On frame-advance, depgraph flushes animated changes
Armatures;
Almost all armature related code has been fully built from scratch.
It now reveils the original design much better, with a very clean
implementation, lag free without even calculating each Bone more than
once. Result is quite a speedup yes!
Important to note is;
1) Armature is data containing the 'rest position'
2) Pose is the changes of rest position, and always on object level.
That way more Objects can use same Pose. Also constraints are in Pose
3) Actions only contain the Ipos to change values in Poses.
- Bones draw unrotated now
- Drawing bones speedup enormously (10-20 times)
- Bone selecting in EditMode, selection state is saved for PoseMode,
and vice-versa
- Undo in editmode
- Bone renaming does vertexgroups, constraints, posechannels, actions,
for all users of Armature in entire file
- Added Bone renaming in NKey panel
- Nkey PoseMode shows eulers now
- EditMode and PoseMode now have 'active' bone too (last clicked)
- Parenting in EditMode' CTRL+P, ALT+P, with nice options!
- Pose is added in Outliner now, with showing that constraints are in
the Pose, not Armature
- Disconnected IK solving from constraints. It's a separate phase now,
on top of the full Pose calculations
- Pose itself has a dependency graph too, so evaluation order is lag free.
TODO NOW;
- Rotating in Posemode has incorrect inverse transform (Martin will fix)
- Python Bone/Armature/Pose API disabled... needs full recode too
(wait for my doc!)
- Game engine will need upgrade too
- Depgraph code needs revision, cleanup, can be much faster!
(But, compliments for Jean-Luc, it works like a charm!)
- IK changed, it now doesnt use previous position to advance to next
position anymore. That system looks nice (no flips) but is not well
suited for NLA and background render.
TODO LATER;
We now can do loadsa new nifty features as well; like:
- Kill PoseMode (can be option for armatures itself)
- Make B-Bones (Bezier, Bspline, like for spines)
- Move all silly button level edit to 3d window (like CTRL+I = add
IK)
- Much better & informative drawing
- Fix action/nla editors
- Put all ipos in Actions (object, mesh key, lamp color)
- Add hooks
- Null bones
- Much more advanced constraints...
Bugfixes;
- OGL render (view3d header) had wrong first frame on anim render
- Ipo 'recording' mode had wrong playback speed
- Vertex-key mode now sticks to show 'active key', until frame change
-Ton-
no executable code.
Martin noticed many of our bpy instance variables were
incorrectly marked as class variables in the doc. This
commit essentially changes the title of sections of the doc
from Class Variables to Instance Variables. Now that we are
adding class or module dictionaries for constants, etc. this is
a distinction worth making. Plus it is right.
