- Blur nodes didn't accept RGB buffers (only RGBA or single channel ones)
- Mix node had maximum for 'mix' on 1.0, for passes edit that could become
more, made it 5!
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.
Cleaned-up drawing code of keyframes. Code redundancy has been
reduced, and there should be slight performance gains (less looping,
no sorting of keys needed, possibly less memory usage).
Patch by Matthew Plough: This adds an option to the Track To constraint to use the target's Z axis for Up axis reference instead of the global Z axis.
Off by default for backward compat.
Doing NULL pointer paranoia check in time.
It's completely pointless calling the function with null pointer,
but who knows.
At least it does not crash then.
although it has a lot of noise. Not to mention our bad string code gives
a load of warnings.
I've reviewed specifically:
- file reading/write
- dna and library code
- node system
- entire render module
Done a couple of files in src/ too, seemed to be nice errors.
Actually no real bug, but unfinished work in Composite:
The ZCombine node only accepted images, no value inputs for sockets.
Now only the first (top) input socket should be an image, the other ones
can use constant color or values too. Nice way to split an image in
foreground and background.
Also: added an output socket for the resulting Z value.
Do note that zcombone is aliased, no AA or sample info is available.
That's for another time.
Another bug from the tracker, reported by Mike Stramba.
A duplicated Py_XDECREF in the wrong place made assigning
arrays from Vector objects not work.
Also, fixed nasty bug in C API of idproperties (the function
to look up a property from a group was broken). Fixed a memory
leak too.
In addition, made "del group['property']" delete properties
from group; previously this would just crash (or at least it
should have). Added a small addition to the example in the
epydocs for IDGroup.
didn't keep in mind the pose channels could change order...
Note the disabled code in readfile.c; will be finished later. It's for
debugging library dependencies.
More referenced library issues: if an Object loses the Armature (because
for example the linked file removed it) the Object still had the Pose and
posemode flag set, crashing Blender.
The first incarnation assumed that proxies were local objects per
definition. Unfortunately that makes it impossible to - for example -
reference-link an entire Scene with proxies, to be used as a special
character set.
This commit makes the proxy implementation also a bit more clear.
Related work: the scene-sets were not executed fully or correctly for
the dependency graph. That happens now (in 3d view) as well.
Although we will delay the main Rigid Body user interface after 2.43 release early 2007, I need some constraint data/UI to make progress with COLLADA Physics.
Added RigidBody constraint UI
LR_ALTKEY+LR_CTRLKEY+LR_SHIFTKEY+ P will bake rigidbody
Contribution by RCRuiz, Ramon Carlos.
Fixed the bug of "print type(property)" crashed blender;
just needed to add PyType_Ready init code for all ID property
python types; I made a function IDProp_Init_Types and put
code to call it in types_initAll().
Also added GPL/BL headers to idproperty files.
Removed all limitations from render code for maximum threads. The only
define for this now is in BLI_threads.h, and currently set to 8.
Note that each thread renders an entire tile, and also allocates the
buffers for the tiles, so; more threads might work better with smaller
tiles.
IMPORTANT: node system won't work yet with more than 2 threads! So, don't
try material nodes or compositing with over 2 threads. That I'll commit
later today.
What does work (should work :) is AO and soft shadow now.
This bug was caused by broken's recent patch to make the Curve modifier
determine its axis from the modifier panel, rather than the object's track
axis. In the case where a Curve modifier is applied by parenting the object
to the curve, the curve axis needs to be copied from the object's track axis
in modifiers_getVirtualModifierList(). This commit adds the necessary copy.
Vertex/Face/Frame duplication now draws using OpenGL display lists. Makes
drawing go much faster (2-5 times, depending on size of duplicated object).
This system uses boundbox checks too, so outside of view it draws faster.
Note for face duplication: I've fixe a bug for incorrect alignment when
the parent was rotated when a parenting happened, the 'inverse parent
correction matrix' then messed up alignment. For face duplication it now
works OK, but for vertex-dupli not... need a way to fix this backwards
compatible.
center, rotated according to face edges, and optionally with scale of the
face.
http://www.blender3d.org/cms/Face_Duplicator.828.0.html
Also: bugfix in undo/redo for relinking data to the UI. It was not checking
for Library data, accidentally linking non-library data with same names.
When linking a referenced (Library) Image to a texture face, it wasn't
tagged to become saved in file. So you lost data.
Also fixed: tooltip for the "Li" icons was wrong.
freed. Windows nicely crashes then.
Also: added human readable string in fileglobal, so you can check the
subversion in a file manually too. Rule:
- Find GLOB chunk (first chunk after REND now)
- skip sizeof BHead, 20 bytes in 32bits, or 24 in 64 bits
- there's 4 characters showing subversion, right now it reads " 1".
Next to the release code, a subversion number is written in the file
now. This is in the chunk GLOB, which is now in beginning of file.
Subversions can be used to have finer control over do_versions(), so you
don't have to wait for a release to patch stuff nicely. We can also
increase these subversions regularly to denote important changes.
If a subversion is not 0, it's being printed in the header, so a user can
also verify the state. (We might even tag cvs for it?).
Next to this, a minimum version and subversion number are written too.
From now on, if you change files in a way a past binary cannot read this
nicely anymore, you should set these minima to the current version.
This was especially added for the new mesh layers, which will not work
for older binaries once we make 2.43.
Ancient bug: texteditor input was limited to "isprint()" characters, the
default non-accented simple ascii set. I've removed that.
Now we still have a conflict with hotkey handling, so ALT+character input
won't work. But, for keyboards that have special character keys from itself,
this patch will allow typing them in Text now.
Array Modifier type "path length" didn't force a path calculation on load.
Note that this is still improper handling in Blender... an exception that
has to move to the depsgraph.
library .blend files anymore, making it a load faster to use.
- Fixed ancient annoyance; samples were not properly freed, giving a lot
"Error Totblock" when using sound.
This fix also involves removing an ancient NaN hack, which treated the
samples as fake Library data in the Main database.
But still, the Blender Sound and Sample code is horrible... :/
nice little 'debug' feature
SHIFT_ESC breaks baking without throwing away the baked keys
so you can check the solution calculated so far
baked sb replay checks for NULL pointers in keys -> motion will stop there
even saving and loading to file works here,
but i would not expect it to do so
humm .. may be i'll add a dirty flag for that
Crash using Shape Widget Wizard script, was an error in customdata copy.
The script still throws a python error though, but that seems unrelated
to this crash.
Added a callback instance_init() so that any particular instance of a
texture plugin can initialize data. Updated the clouds2 and tile example
plugins to have a (dummy) call back.
- New perspective projection capability. If a camera is used as the
projection object, the modifier detects whether to do perspective or
orthographic projection based on the camera type. If any other object
type is used as the projection object, orthographic projection is used.
- Orthographic projection actually works properly now.
- The projected UVs are scaled and offset so that the image is centred in
the projecting camera's view.
- AspX and AspY inputs have been added to control the aspect ratio of the
projected UVs.
Also:
- I have added the Mat4MulVec3Project() function to BLI_arithb.h; this
function converts a 3-dimensional vector to homogeneous coordinates
(4-dimensional, with the 4th dimension set to 1), multiplies it with the
given matrix, then projects it back to 3 dimensions by dividing through
with the 4th dimension. This is useful when using projection matrices.
Random seeding is still not perfect in render, especially lack of good
thread support still.
- VectorBlur node was calling seed for each exec, causing other nodes to
get fixed random too.
- added seed in non-OSA main loop for render
- use BLI_srandom, is better than BLI_srand
CurvePaths: if the path has vector handles, or when it's a Poly curve, the
amount of interpolated points on the path was too limited (6 actually).
Now the 'resolu' of the curve defines the amount of interpolated points.
This enables motionpaths or deforming with sharp corners.
2006/11/19 00:07:32 CET
Fix for bug #5250: inaccurate conversion between edit and pose mode bones.
Two very bad bugs:
- replacing atan() with atan2() should also remove the M_PI correction!
This is the equivalent:
angle=atan(x/y); if(y<0) angle+=M_PI;
angle= atan2(x, y);
- the new NormalizedVecAngle2() call was negating an input vector, causing
calling code to screw up. All arithb.c calls should not alter input.
Yesterday's commit from Brecht broke armature editmode, damaging armatures
in a way you cannot rescue. This rewinds the main changes. After commit
I'll try to find the error...
