Apparently python allows to create Key blocks without proper *from pointer
back to the owner of the Key. The Key unfortunately wasn't mean to be a
dynamic block linkable to any Mesh...
This patch sets the from pointer on any call to do_mesh_key, which is safe
to do anyway. I made a note to get rid of that convention once, but at
least now we got non-crashing blenders. :)
(In bugreport noted that we need to check Python code for it too!)
Strips with a repeat fraction (like 3.51) should also "hold" on the
fraction of .51 when indicated so.
Patch provided by Roland Hess, but cleaned up (exisiting) mess a bit too.
Once an Object Ipo position was inserted, the object couldn't be moved when
a particle system was tied to it. Needed code to copy object settings,
before the particle animation was done.
thus directories in the path are created using RE_make_existing_file
in fluidsim.c)
- changed default output to be '//fluidsimdata/...'
- hopefully fixed bugs #3466 and #3470 (orco texture coords caused
problems when the fluidsim mesh was bigger than the original one,
one no other deformations were used - new deformedVerts are now
allocated in this case in DerivedMesh.c)
- Curves without bevel or extrude didn't calculate the Z value for
softbody when the "3d" option for curves was off.
- Typo in softbody.c, for switch of // comments to /* */
in the game engine again.
All files in source/gameengine/ I've committed below have a comment like
/* XXX note .... */
where something needs to be done. This is pretty straightforward for Erwin,
who has been notified in person too about this. :)
in other layers (or hidden with local view).
In my search for the absolute minimum of recalculations, changes are only
flushed when they're visible. On changing layers, the tags then are just
set again (for everything that potentially moves) to ensure proper state.
However, it didn't work proper for linked Mesh objects that changed in
editmode, the Derivedmesh callback then accessed memory out of bounds.
The current dependency code was more designed for animation systems...
updating display data should work too, but might need some more tests!
(Thanks Andrea for clear error sample!)
- raytracing code now uses cylinder-line intersect for strands. It used
to intersect with the screen-aligned strand quads, giving too many
misses for tracing.
Note; mirror for hair is still not well supported!
- added in Materials a choice for whether a material is Traceable for
raytracing or whether it is part of Shadow buffers. This way you can
exclude hair strands from raytracing, but still get shadowbuffers for it.
replace the original one. so modifiers now work with them
(apply modifier, or edit mode still work on original mesh).
this should fix the three fluidsim bugs in the tracker.
- fixed stupid makesdna problem (writing "char string[160+80]" isnt
a good idea :)
- changed GUI a bit, now displays an estimate of the required memory,
there's still a problem with redrawing (currently relies on a call
to the derived mesh generation)
- the fluidsim struct changed to store the bounding box,
and the current loaded surface mesh
- temporary simulation files are now removed if env. var.
BLENDER_DELETEELBEEMFILES is not set or zero
- fluidsimSettingsFree now gets properly called when freeing an object
To enable dynamic tiff support.
I had to fix some of the logic in the fileselect box for icons,
I also expanded the patch to look in various default locations for
a dynamic libtiff.so/libtiff.dll
and look at the env variable BF_TIFF_LIB if it can't find it automatically.
If unable to load the library it prints a message about setting BF_TIFF_LIB
to the console.
I haven't been able to test it on a lot of platforms but hopefully it
will just work ;) I added the files to scons but have not had a chance to
test that as well.
Kent
set whilst the actual time was 1 frame after the strip. Appeared to be
a rounding error that didnt show in OSX.
Previously I added a threshold, to make sure strips are included when the
current frame is exactly on the end. That threshold now is smaller, and
I also made the fmod() to be done only on repeating strips.
Too strict checking in depsgraph code missed a dependency case; this
specifically when objects have own motion (ipo) and are being deformed
by a parent.
This worked, unless the parent also had own motion (ipo or constraint)...
Also: the DAG_add_relation() added multiple relationships for object pairs,
now they "Or" the relationship description flag. Makes it a tad faster
again :)
Object duplicators didn't draw when not selected. Was one 'else' to many in
the code!
Also noticed that duplicators couldn't be transformed when they have Ipos,
was an improper call to do_ipo in duplicator code... also made sure the
transformation info in the object is stored/restored before duplicating.
This adds Radiance HDR image file support. So now at least we can save
the 'fbuf' (4x32 bits float colors) in Blender.
It doesn't change anything for internal support in imbuf for floa colors,
so when reading .hdr files it still converts it to 32 bits RGBA.
As an extra I've added that saving images with F3 now also adds the
optional extension, when the F10 "Extensions" option is set.
One important note; I don't know the proper license for the code, it was
provided without... will await feedback from Alfredo about it. For now
I've added the standard Blender GPL header.
This fixes an error in the striding system, which was by default correcting
path position backwards in time, which could give errors when a stride bone
moves in the beginning of a path faster than the path moves. (Can you
follow that? cool!)
The patch checks for this case, it corrects by default forwards in time,
unless we're at the end of a path. As bonus this fix will also ensure the
character stays on the path closer.
had a shadow in memory there was an issue ..
yes, spring calculation should care for scaling in object matrix
so here it is
note: for meshes the bp array is abused for caching -->
only 1 matrix multiplication compared to 2 per spring
(Non deformed) lenghts. Code for lattice (by Jens Ole) was already there
btw, was replaced accidentally by Daniel during his modifier craze :)
ALso added comments in code to clarify the softbody initialize flow, which
happens in 2 stages now. Goal weights and springs are read from the
original data, the vertex coordinates from the modifier data.
marker wasn't on the first or last possible position. Caused by clipping.
As bonus; added Cardinal interpolation option too, which is just that
little bit different! (Cardinal goes through the controlpoints, bspline not)
curves. On sharp 'peaking' curves the handle was calculated
using both X and Y distance. This could result in overshooting.
New code only evaluates the X distance, resulting in much more moderate
sized handles.
Thanks Gabio for the demo file!
- bug: posemode, bones were drawing names and axes even when hidden
- bug: using softbody guides actually worked on themselves, causing
an infinite loop
- feature: when a pose/bone is completely locked for transform, a grab
will change into rotate by default.
Well, it already worked a bit, but without weight options or edge
stiffness. You now can set the weights using the "Properties" Panel in
the 3D Window (allows multiple selections too) or with Wkey in Edit Mode.
Bezier curves have this too.
NOTE: Lattice SoftBody Goal created yesterday won't work anymore!
I've had to recode weight support for Nurbs Points, using a new weight
variable... this because the existing W variable was in use for Nurbs
already. Also Lattices have this new Weight variable, so the code is nice
uniform. Sorry for the artists who already created complex Lattices... :)
NOTE2: Surface Objects don't support edge stiffness yet
NOTE3: I've removed ancient screen coordinates from the Bezier struct,
which makes - even with added weight and padding - the struct smaller!
Demo file:
http://download.blender.org/demo/test/2.40/softbody_curve_lattice.blend
Animating detailed clothes with softbody becomes messy, so now we'll
try it this way. :)
It simply uses the W (weight) value, as already available in each Lattice
Point. Only had to make it editable;
- NKey panel
- or press W in editmode
Further there's a minimalistic W button in the softbody Panel!
Vertex-parenting a forcefield to a softbody caused eternal loop :)
It is partially because of the timing system (=weak) but actually also
because derivedmesh calls could also not build meshes all the time!
