| Age | Commit message (Collapse) | Author |
|---|
|
With large steps in hair editing strokes the solution to distance-constrained
hair displacement becomes inaccurate, leading to a gradual lengthening of strands.
To counteract this the length constraints can enforce a velocity relative to the
constant target length of each segment. This results in an overall constant
length, even when the constraint solution is inaccurate for nonlinear constraints.
Note that the result would become much better with backward Euler integration,
currently there is a noticable 1-step lag between strokes and the distance fix.
|
|
This is not really nice yet, but just serves as a test for internal forces.
|
|
|
|
|
|
|
|
|
|
This happens when cmath.h is included after math.h in cpp code.
Kudos to Sergey for pointing this out.
|
|
|
|
movie strips
Now we free sequencer cache and close all unneeded FFmpeg handles when rendering.
This is the same logic as image sequence memory freeding.
|
|
Reported by @panzergame in D1414.
`glDisable` calls with bad enum argument `GL_TEXTURE_2D_MULTISAMPLE` that came from this line:
`tex->target = (n == 1) ? GL_TEXTURE_1D : (samples ? GL_TEXTURE_2D_MULTISAMPLE : GL_TEXTURE_2D);`
Reviewers: brecht
Reviewed By: brecht
Subscribers: AlexKowel, yurikovelenov, panzergame
Differential Revision: https://developer.blender.org/D2145
|
|
|
|
The first character of the ID type was used to tag IDs for updates which
is weak since different IDs can have the same first character (for
example meshes, materials and metaballs), causing unnecessary updates of
unrelated IDs.
Now we use a unique index per ID type to tag for updates, unifying IDs
arrays indexing along the way.
Reviewers: sergey, mont29
Differential Revision: https://developer.blender.org/D2139
|
|
Depth buffer values are used by the viewport pan and zoom code to
adjust response scaling factors between mouse and viewport movement.
Letting smoke write to the buffer confuses it and causes the camera
to get stuck and move very slowly inside smoke domains, because it
thinks it is very close to an object.
|
|
The old function name was not clear enough
|
|
If the user enable polygon draw and press ESC before drawing something,
the last stroke must not be moved back.
|
|
Wasn't copying back local view bytes to object after changing layer.
Steps to reproduce were:
* Split 3D view in default startup.blend
* Enter local view in one of the 3D views
* Move default cube to different layer *in the other 3D view* (the one that's not in local view)
* Try transforming object from 3D View that's in local view (should lag)
|
|
The polygons must move the stroke to back only when polygon is complete
|
|
For artist point of view is very useful to have an option to draw by
default the new strokes on back of all strokes in the layer.
|
|
During code review a field was renamed, but one line was missing.
|
|
Do not clear continuous drawing. This code was related to a previous
test and must be removed.
|
|
From source code history, it seems this behavior is a relic of a very
old limitation when the same object couldn't be both a Softbody and a
Collision object. In those times if somehow both modifiers were added
to an object, Softbody auto-disabled itself and buttons were hidden.
Now however there is no problem having both modifiers on an object,
so there is no point hiding the buttons. The same exact buttons are
available on the physics tab in any case.
|
|
As reported in T48928, From Dupli UV is not supported for grid
distribution, and running the normal code as usual simply produces
nonsense data, because fuv is used to hold orco instead of
interpolation factors for uv, and num is zero.
Since support won't be added in 2.78, just stop outputting nonsense.
|
|
This uses the Lagrange multiplier method for correcting the "velocity" of
vertex displacements as described in
"Linear-Time Dynamics using Lagrange Multipliers" (Baraff, 1996)
Future additions would include stiffness forces for creating more natural
hair shapes, and handling of collisions (deflection).
|
|
|
|
|
|
|
|
|
|
Convention for properties callback names is to prefix them with a version of the RNA struct name.
|
|
All in all, this patch adds an Alembic importer, an Alembic exporter,
and a new CacheFile data block which, for now, wraps around an Alembic
archive. This data block is made available through a new modifier ("Mesh
Sequence Cache") as well as a new constraint ("Transform Cache") to
somewhat properly support respectively geometric and transformation data
streaming from alembic caches.
A more in-depth documentation is to be found on the wiki, as well as a
guide to compile alembic: https://wiki.blender.org/index.php/
User:Kevindietrich/AlembicBasicIo.
Many thanks to everyone involved in this little project, and huge shout
out to "cgstrive" for the thorough testings with Maya, 3ds Max, Houdini
and Realflow as well as @fjuhec, @jensverwiebe and @jasperge for the
custom builds and compile fixes.
Reviewers: sergey, campbellbarton, mont29
Reviewed By: sergey, campbellbarton, mont29
Differential Revision: https://developer.blender.org/D2060
|
|
There were two issues here. One is that the fix done originally for this
bug only checks for colliding with the same face as the single preceeding
hit. If the particle hits an edge or vertex of the collider, it in fact
hits two or more faces, so the loop ends up cycling between first two
of them and reaches the max collision limit.
The fix is to disable the collider for the sim step once a permeability
roll succeeds, by adding it to a skip list. Skipping just one face causes
some particles to bounce at odd angles in case of partial permeability.
The second problem was that the collider bounced back a small percentage
of particles, and the cause seemed to be that the code was set to flip
the velocity if the particle was just past the collider but still within
collision distance. Inverting both values causes a half permeable collider
to stop particles, so it seems that this if branch shouldn't bounce at all.
Test file: {F327322}
Reviewers: lukastoenne, brecht
Reviewed By: brecht
Subscribers: brecht, #physics
Maniphest Tasks: T26658
Differential Revision: https://developer.blender.org/D2120
|
|
Prints error on startup.
|
|
rna_GPencilBrush_name_set() was trying to use a mere bGPDbrush as a complete ToolSettings,
was doomed to fail...
|
|
* "Flip direction" -> "Flip Direction"
* "Show drawing direction" -> "Show Directions"
* "Grease Pencil Curves" -> "Brush Curves"
(I was considering "Brush Response Curves" instead, but that seemed like too much
of a mouthful)
* "X" for removing a palette. The UI there was more similar to a standard datablock
selector, so it should use the "+X" combo instead of "+-" combo for consistency.
(Note though, presets tend to use "+-" instead - e.g. see the Render Settings)
|
|
(Ctrl-Shift-J is for "Join and Copy")
|
|
Previously, it would insert "invisible" points after the endpoints of the strokes,
so that they wouldn't appear to be joined, but that behaviour could also get quite
confusing as you wouldn't be sure whether the strokes were really joined or not.
To keep the previous behaviour, simply enable the "Leave Gaps" option on the operator
after running it. This setting will get saved between runs of the operator.
|
|
Previous check to skip non-linkable IDs resulted in not clearing those skipped ID's newid member,
wich lead to try to remap it later in code.
|
|
|
|
Group membership testing for including/excluding feature lines was not
accounting for object names possibly further qualified by library file
paths.
Also fixed a few potential (but unlikely) references of uninitialized
variables.
A big thank to Bastien Montagne for the insight on the cause of the
problem and how to fix it.
|
|
|
|
There were actually two issues, one in recent changes and one existing... forever?
I) id_make_local() would never proceed over indirectly linked data, which is required in specific 'append' context.
II) BKE_image_copy() was not setting id->newid pointer of newly copied ID.
About II: don't really know why image copy does its own cooking instead of using generaic ID copy helpers.
Think this should be changed, but that would be after 2.78 now.
|
|
rotation (tentatively).
This method is more straightforward and easier to predict when testing constraint solving.
|
|
Drawing both text and the wave onto a sound strip makes both hard to read,
which is a concrete issue for Hjalti at the moment. This was the simplest
fix I could think of to give him control over what he sees.
|
|
This is a good point to change this as grease-pencil-v2 branch was just merged, so I hope merge conflicts with other branches are minimal.
|
|
|
|
on OS X.
|
|
Now all consistent and using term "properties region" instead of "properties panel". Ideally we had a more generic operator for all those.
Fixes T49006.
|
|
Improve current Grease Pencil in order to get a better 2D animation tool.
More info in WIKI pages: https://wiki.blender.org/index.php/User:Antoniov
Reviewed By: Severin, aligorith, campbellbarton
Patch by @antoniov, with edits by @Severin.
Differential Revision: https://developer.blender.org/D2115
|
|
display on OS X.
|
|
Ignore that lights are not included in local view, always render them if they are on the correct layers. BI and Cycles do this as well.
|
|
When the colliders are joined, each cell that touches them only gets
velocity contribution once. When there are multiple objects, velocities
are summed, which causes some cells to get 2x, 3x or more than the actual
object velocity.
Fix this by using the average velocity of all colliders touching a cell.
Reviewers: miikah, lukastoenne
Reviewed By: lukastoenne
Subscribers: dafassi, scorpion81, #physics
Maniphest Tasks: T43782
Differential Revision: https://developer.blender.org/D2112
|
