Such kind of image sequences wouldn't be displayed still
and supporting them would take some additional time, but
for now it'll be no memory leak on attempt opening such
images.
There're some features planned which would
require rigid registration, but this code
would need to be re-done anyway to use new
minimizer and solving some issues with ICP
algorithm there.
Several major things are done in this commit:
- First of all, logic of modal solver was changed.
We do not rely on only minimizer to take care of
guessing rotation for frame, but we're using
analytical rotation computation for point clouds
to obtain initial rotation.
Then this rotation is being refined using Ceres
minimizer and now instead of minimizing average
distance between points of point of two clouds,
minimization of reprojection error of point
cloud onto frame happens.
This gives quite a bit of precision improvement.
- Second bigger improvement here is using bundle
adjustment for a result of first step when we're
only estimating rotation between neighbor images
and reprojecting markers.
This averages error across the image sequence
avoiding error accumulation. Also, this will
tweak bundles themselves a bit for better match.
- And last bigger improvement here is support of
camera intrinsics refirenment.
This allowed to significantly improve solution
for real-life footage and results after such
refining are much more usable than it were before.
Thanks to Keir for the help and code review.
This information is useful, but in cases when you, say,
working on a bundler it's annoying to scroll all the
information up.
Now behavior would be:
- running `./blender --debug-libmv` will print all the
debug messages
- running `./blender --debug-libmv --verbose 0` will
print only debug messages from solvers, recosntruction
and so, but will bypass final reprojection bunch of
messages
- running `./blender --debug-lib,v --verbose 1` will
include final reprojection messages.
This shall not lead to any functional changes, just
avoids radial distortion code duplicated in camera
intrinsics and bundling code.
For fancier bundle adjustment supprting different
distortion models this is not actually enough and
would need to make some bigger changes, but this
changes makes code a bit easier to maintain already.
Alt+Mouse Wheel events depended on the button under the mouse being active, else it would change the current frame.
Button re-activation wasn't 100% reliable since it depends on fake mousemove events.
Now keep button active when its executed if the mouse stays over it.
in the UI
This is the second part of the fixes for [#32492], making it easier to identify
which part of a driver (i.e. which of its targets) is causing problems
A number of additional/related changes needed to be made:
* Red-alert status for layouts is now propagated down to child layouts when they
are created. This is needed as otherwise some of the templates used in the Graph
Editor driver settings won't actually get the red-alert status flushed down to
them. Also, note that this status needs to be set before any widgets are added
to the layout, or else the settings aren't applied when the relevant widgets get
created.
* "Single Property" RNA-Paths resulting in out of bounds array access will now
trigger an error status and appropriate warnings
TODO:
* The error tagging doesn't get applied immediately after variables are created,
or their types changed
* There was also some other weirdness involved when a higher-value flag (1<<10)
was used for this setting, which needs further attention
(so one can find more about the joystick class without having to search joystick)
code untested by the way. Sorry but I can no longer build sphinx docs here. I will try to fix that later.
* Replace magic numbers with enum-defines
* Remove superfluous return; statements (used as last statements in a few
"static void" methods)
* Remove some old unused/commented out stuff
This commit splits editarmature.c and poseobject.c into several files, such that
certain types of functionality are (mostly) self-contained within particular
files (instead of being mixed in with other functionality in a large file).
In particular, this was done so that:
1) Armature EditMode tools are now in the armature_*.c files only, and Pose Mode
tools in pose_*.c files only.
- In one or two cases, this hasn't been possible as the two modes rely on
much of the same shared infrastructure.
2) The "clear loc/rot/scale" operators and pose show/hide are no longer housed
in editarmature.c
3) Selection operators, Transform operators, structural (add/delete) operators,
and supporting utilities for the modes (enter/exit/roll-calculations) are not
all interleaved in an ad-hoc manner
Notes:
* I've tried to ensure that the history of the new files has been maintained by
doing
svn copy {editarmature.c/poseobject.c} {armature_*.c/pose_*.c}
Unfortunately, this means that the diffs are a bit messy.
* There should be no functional/logic changes here. Just code moving around and
cosmetic comment tweaks where needed.
* #includes have largely been untouched for now, but could be cleaned up later
* CMake changes untested, but should work in theory.
Fix for 2.66a
Some editors in Blender (Like Graph) have "maxzoom" set to 0, which caused the
MMB zoom sensitivity to be minimal. That used to be much more responsive, restored
it to how it worked before.
Most notable difference from now on will be that all py is handled from current blender's resource dirs, no more from source dir. Better for consistency, and avoid e.g. cycles' addon to be checked twice...
Two example files that crashed texture nodes.
- On delete texture nodes, it should free the exec cache (because this cache
stores the node pointer.
- On redo, nodes can exist can exist without typeinfo set. Exec (free) code
was not checking for that. Don't ask me why this happens... tex nodes are weird.
Main purpose of this is to be more compatible with older
versions of blender (before alpha cleanup) where sequencer
used to display premultiplied image on an straight opengl
viewport.
Now sequencer preview would behave closer to image editor
However adding Alpha and R|G|B displays is not so simple
because sequencer is using 2D textures. Would be nice to
implement this options as well, but this is not so much
important IMO.
This hall fix
- #34453: VSE: Subtract function does not work properly
TODO: Make RGBA display default for our startup.blend
- linking groups into a scene now uses the objects original layers (some users have the layers set to useful values, so overwriting isnt so nice).
- dropping objects into the 3d view would make them active but not selected, a valid but confusing state, since most tools activate and select now objects.
This fixes the placement code of new files added to the sequencer timeline.
The old code tried to guess the strip position from the current mouse
pointer position.
Annoying effect: if you add a new strip using the menu, especially if the
file editor pops up, the strip ends up in nowheres land (most likely around
track 40, frame -200).
New behaviour: strips are always placed at cfra, which is the
sequencer equivalent to the 3D cursor (and that's where new objects in
3D editing end up).
Bonus feature: we try our best to guess the right track by finding the
nearest strip by type.
The patch was inspired by
[#32766] VSE: Add Strip on Current Frame
Thanks to venomgfx for the idea!
Also made it theraded linear solver, seems it makes
sense for iterative schur with inner iterations
enabled.
Use OpenMO's max therads called from bundler code
to detect how many threads to use. Could be changed
in a way that number of threads is passing in options
from blender side in the future.
Also removed redundant V3D definition from compiler's
flags.
