Summary:
Made objects update happening from multiple threads. It is a task-based
scheduling system which uses current dependency graph for spawning new
tasks. This means threading happens on object level, but the system is
flexible enough for higher granularity.
Technical details:
- Uses task scheduler which was recently committed to trunk
(that one which Brecht ported from Cycles).
- Added two utility functions to dependency graph:
* DAG_threaded_update_begin, which is called to initialize threaded
objects update. It will also schedule root DAG node to the queue,
hence starting evaluation process.
Initialization will calculate how much parents are to be evaluation
before current DAG node can be scheduled. This value is used by task
threads for faster detecting which nodes might be scheduled.
* DAG_threaded_update_handle_node_updated which is called from task
thread function when node was fully handled.
This function decreases num_pending_parents of node children and
schedules children with zero valency.
As it might have become clear, task thread receives DAG nodes and
decides which callback to call for it.
Currently only BKE_object_handle_update is called for object nodes.
In the future it'll call node->callback() from Ali's new DAG.
- This required adding some workarounds to the render pipeline.
Mainly to stop using get_object_dm() from modifiers' apply callback.
Such a call was only a workaround for dependency graph glitch when
rendering scene with, say, boolean modifiers before displaying
this scene.
Such change moves workaround from one place to another, so overall
hackentropy remains the same.
- Added paradigm of EvaluaitonContext. Currently it's more like just a
more reliable replacement for G.is_rendering which fails in some
circumstances.
Future idea of this context is to also store all the local data needed
for objects evaluation such as local time, Copy-on-Write data and so.
There're two types of EvaluationContext:
* Context used for viewport updated and owned by Main. In the future
this context might be easily moved to Window or Screen to allo
per-window/per-screen local time.
* Context used by render engines to evaluate objects for render purposes.
Render engine is an owner of this context.
This context is passed to all object update routines.
Reviewers: brecht, campbellbarton
Reviewed By: brecht
CC: lukastoenne
Differential Revision: https://developer.blender.org/D94
That ended up in tricky code trying to mimic depsgraph
branch behavior API-wise preserving texspace and bound
box calculation compatible with previous releases.
So for now bring cu->disp back to the trunk but keep
texpsace and boundbox APIs the same as in the branch.
This keeps texpsapce and boundbox behavior fully compatible
with previous releases and still makes API the same as
for meshes.
previous commit was incorrect, the face flipping depended on the orientation of the curve.
fix by passing the bevel direction to the fill function so we can have a reliable front/back.
This also gives some speedup for all curve filling since we can avoid calculating the normal since its already known.
modifiers with text/curve/surface objects.
This is because the code here did not yet support the distinction between
generating the tesselated mesh at viewport/render resolution and generating a
separate displist/derivedmesh for render that is not stored in the object but
created and freed on the fly.
note: BPY_class_validate() could come in handy later if we need to check classes for properties/functions but for now there is no point in keeping it in.
Fix#21187: 2.5svn26947 - particles + meta sphere = crash in rendering
Use separated displists for mballs in view3d and render stuff.
Do not recalculate displist for view3d while rendering - mball.c
uses several global variables which shouldn't be accepted from
parallel threads.
unrelated changes that ended up being more trouble to commit separate...
- removed BLI_split_dirfile(), was nasty, occasionaly modifying the source string, it could create directories and used the $CWD in some cases. was only used in 2 places in filesel.c, if this gives problems can address without bringing back this function.
renamed BLI_split_dirfile_basic --> BLI_split_dirfile
- view3d_operator_needs_opengl was being called for offscreen render when it wasnt needed.
Used approach with creating DerivedMesh for curves whet they've got such modifiers.
Available modifiers are: array, edge split, mirror, solidify, subsurf.
* Missing changes to release/windows/installer
* Sequencer fixes in source/blender/src/seqaudio.c dont apply to 2.5
* brechts fix for #18855 r20763 wasnt merged, does this apply to 2.5?
Think global, act local!
The old favorite G.scene gone! Man... that took almost 2 days.
Also removed G.curscreen and G.edbo.
Not everything could get solved; here's some notes.
- modifiers now store current scene in ModifierData. This is not
meant for permanent, but it can probably stick there until we
cleaned the anim system and depsgraph to cope better with
timing issues.
- Game engine G.scene should become an argument for staring it.
Didn't solve this yet.
- Texture nodes should get scene cfra, but the current implementation
is too tightly wrapped to do it easily.
From the anti-globalization department:
G.obedit terminated!
Wherever possible, use CTX_data_edit_object(C) to get this
now. It's stored in scene now, and the screen context has
it defined.
So, editmode mesh is back! :)
At the moment only TABkey works and mouse select, 1 vertex at a
time. More will follow of course.
Note for the devs:
- G.editMesh has been removed, be careful with old code.
- EditMesh now is property of Mesh itself
Although it means unlimited editmodes, for migration purposes we
better stick to 1 "obedit" per scene, which is in Context too
- G.obedit will get removed soon, so use CTX_data_edit_object(C)
Or if you can't, just scene->obedit for now
- Also removed the CTX_data_edit_mesh(), this has no meaning
anymore. EditMesh is not context senstitive anymore, only the
edit-object for time being is.
- Martin: I've already tucked some EditMesh pointer in T and
removed all G.editMesh there.
- the number of segments was always 1 too many on cyclic curves.
- [#17739] - normals were not being calculated when rendering curves.
Replaced macro DL_SURFINDEX with a function. it that assumes variable names and could break from the loop that called it.
=========
Merge of the famous particle patch by Janne Karhu, a full rewrite
of the Blender particle system. This includes:
- Emitter, Hair and Reactor particle types.
- Newtonian, Keyed and Boids physics.
- Various particle visualisation and rendering types.
- Vertex group and texture control for various properties.
- Interpolated child particles from parents.
- Hair editing with combing, growing, cutting, .. .
- Explode modifier.
- Harmonic, Magnetic fields, and multiple falloff types.
.. and lots of other things, some more info is here:
http://wiki.blender.org/index.php/BlenderDev/Particles_Rewritehttp://wiki.blender.org/index.php/BlenderDev/Particles_Rewrite_Doc
The new particle system cannot be backwards compatible. Old particle
systems are being converted to the new system, but will require
tweaking to get them looking the same as before.
Point Cache
===========
The new system to replace manual baking, based on automatic caching
on disk. This is currently used by softbodies and the particle system.
See the Cache API section on:
http://wiki.blender.org/index.php/BlenderDev/PhysicsSprint
Documentation
=============
These new features still need good docs for the release logs, help
for this is appreciated.
This commit upgrades the modifier stack to only calculate the data which is
needed, either by modifiers further down the stack or by other functions at
the end of the stack (e.g. drawing functions).
This speeds up modifier stack recalculation, especially where vertex
groups and UV coordinates are concerned. For example, a mesh with an Armature
modifier followed by a Subsurf modifier would previously have required the
Subsurf modifier to interpolate all the vertex groups in the mesh, slowing
down modifier calculations considerably. With this update, vertex group data
is not propagated beyond the Armature modifier, so calculations are faster.
Note that this depends on the order of modifiers in the stack. If the Armature
and Subsurf modifiers were swapped in the above example, the Subsurf modifier
would have to interpolate vertex groups, as they are needed by the Armature
modifier.
- Curve/Nurbs/Font/MBall now all draw arrays.
- had to flip abgr to rgba in shaded drawing
- Mesh drawing can't be easily done; the indices for faces are not in
in one chunk. Also need a way to gether trias/quads, per material.
Speedup results are mixed. Something between 2-4 times. Especially for
text it seems to help.
All data layers, including MVert/MEdge/MFace, are now managed as custom
data layers. The pointers like Mesh.mvert, Mesh.dvert or Mesh.mcol are
still used of course, but allocating, copying or freeing these arrays
should be done through the CustomData API.
Work in progress documentation on this is here:
http://mediawiki.blender.org/index.php/BlenderDev/BlenderArchitecture/CustomData
Replaced TFace by MTFace:
This is the same struct, except that it does not contain color, that now
always stays separated in MCol. This was not a good design decision to
begin with, and it is needed for adding multiple color layers later. Note
that this does mean older Blender versions will not be able to read UV
coordinates from the next release, due to an SDNA limitation.
Removed DispListMesh:
This now fully replaced by DerivedMesh. To provide access to arrays of
vertices, edges and faces, like DispListMesh does. The semantics of the
DerivedMesh.getVertArray() and similar functions were changed to return
a pointer to an array if one exists, or otherwise allocate a temporary
one. On releasing the DerivedMesh, this temporary array will be removed
automatically.
Removed ssDM and meshDM DerivedMesh backends:
The ssDM backend was for DispListMesh, so that became obsolete automatically.
The meshDM backend was replaced by the custom data backend, that now figures
out which layers need to be modified, and only duplicates those.
This changes code in many places, and overall removes 2514 lines of code.
So, there's a good chance this might break some stuff, although I've been
testing it for a few days now. The good news is, adding multiple color and
uv layers should now become easy.
- Shaded drawmode is back (shift+z).
Note it still only uses orco texture; but lighting/shading is using
the internal render module entirely.
- "Make Sticky" option back.
(Also fix in sticky texture render, was wrong scaled)
-> Any Group Duplicate now can get local timing and local NLA override. This
enables to control the entire animation system of the Group.
Two methods for this have been implemented.
1) The quick way: just give the duplicator a "Startframe" offset.
2) Advanced: in the NLA Editor you can add ActionStrips to the duplicator
to override NLA/action of any Grouped Object.
For "Group NLA" to work, an ActionStrip needs to know which Object in a
group it controls. On adding a strip, the code checks if an Action was
already used by an Object in the Group, and assigns it automatic to that
Object.
You can also set this in the Nkey "Properties" panel for the strip.
Change in NLA: the SHIFT+A "Add strip" command now always adds strips to
the active Object. (It used to check where mouse was). This allows to add
NLA strips to Objects that didn't have actions/nla yet.
Important note: In Blender, duplicates are fully procedural and generated
on the fly for each redraw. This means that redraw speed equals to stepping
through frames, when using animated Duplicated Groups.
-> Recoded entire duplicator system
The old method was antique and clumsy, using globals and full temporal
copies of Object. The new system is nicer in control, faster, and since it
doesn't use temporal object copies anymore, it works better with Derived
Mesh and DisplayList and rendering.
By centralizing the code for duplicating, more options can be easier added.
Features to note:
- Duplicates now draw selected/unselected based on its Duplicator setting.
- Same goes for the drawtype (wire, solid, selection outline, etc)
- Duplicated Groups can be normally selected too
Bonus goodie: SHIFT+A (Toolbox) now has entry "Add group" too, with a
listing of all groups, allowing to add Group instances immediate.
-> Library System
- SHIFT+F4 data browse now shows the entire path for linked data
- Outliner draws Library Icons to denote linked data
- Outliner operation added: "Make Local" for library data.
- Outliner now also draws Groups in regular view, allowing to unlink too.
-> Fixes
- depsgraph missed signal update for bone-parented Objects
- on reading file, the entire database was tagged to "recalc" fully,
causing unnecessary slowdown on reading.
Might have missed stuff... :)
morning)
- fun for the whole family, boolean mesh modifier... doesn't work
with layered modifiers yet (just uses base mesh), although may
god have mercy on your soul if you want to run boolean on a
subsurf anyway
- added displistmesh_add_edges
This exposes a bug in boolean, apparently the output is somehow
random (hash on alloc'd pointer value perhaps) which is sortof
lame.
It also makes more apparent the desire for some level of control
over dep graph evaluation during editmode (at the moment dep
graph is reevaluated for a mesh object in editmode, but since
mesh changes are on editmesh other objects don't really see
any change, so it is a wasted recalc).
- new feature, twiddled with lattice resizing to try to maintain
existing vertex changes... much nicer than just resetting the
lattice if you decide you need more detail in the lattice.
- modifiers work with lattices now. yes, that does mean you
can make a chain of lattices effecting each other 8 miles
long.
- some cleanup of softbody code, was rather splintered and call
path was twisted and confusing. reworked main object step
routine to do things in a more obvious and consistent manner
and without duplicate code
- added ob->softflag OB_SB_RESET instead of sbObjectReset
call
- modifier changes reset softbody now
- moved curve_getVertexCos/curve_applyVertexCos into curve.c
- update curve modifier eval to work with virtual modifiers
- update modifier apply to work with curves/surfs
- update make parent to also recalc object data
NOTE: Although you can turn SB on for curve/font objects at the
moment it doesn't really work because they call deform in
multiple steps and this confuses SB. Not sure how to deal with
atm.
- removed {lattice,curve}_modifier functions
- changed render code to use displist for curve rendering
instead of making its own. required adding a bevelSplitFlag
field to DispList. I also fixed the bevel face splitting
which did not work correctly in many situations.
- changed so all curve data creation happens in makeDispListCurveTypes,
includes making bevel list and filling polys
- changed render code to use displist for surface rendering
- removed Curve.orco variable, built as needed now
- removed stupid BLI_setScanFill* functions... why use a function
argument when you can use a global and two functions! Why indeed.
(this fixed crash when reloading a file with filled curves and
toggling editmode)
- bug fix, setting curve width!=1 disabled simple bevel for no
apparent reason
- cleaned up lots and lots of curve/displist code (fun example:
"if(dl->type==DL_INDEX3 || dl->type==DL_INDEX3)"). Hmmm!
- switched almost all lattice calls to go through lattice_deform_verts,
only exception left is particles
- added DBG_show_shared_render_faces function in render, just
helps to visualize which verts are shared while testing (no
user interface).
- renamed some curve bevel buttons and rewrote tooltips to be
more obvious
- made CU_FAST work without dupfontbase hack
Also by the way I wrote down some notes on how curve code
works, nothing spiffy but it is at:
http://wiki.blender.org/bin/view.pl/Blenderdev/CurveNotes
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
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)
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
- convert MeshDerivedMesh to calculate new vertex normals and such
on initialize, means copy free conversion to DispListMesh
- replace vertex access through function by direct access fo
MeshDerivedMesh
- shadeDispList was not getting correct orco's
