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.
Systematically adding some custom id to template_list using default UI_UL_list class, this one is commoly used more than once in an area, yielding collision issues if they do not have a custom id...
(Did not add those when I created that module, because I did not thought we would actually need them in usual UI code, but turned out I was wrong).
Also made some optimizations in those py gettext funcs, when i18n is disabled at build time, no need to do pyobject -> cstring -> pyobject conversions!.
It introduces a new (py-extendable and registrable) RNA type, UIList (roughly similar to Panel one), which currently contains only "standard" list's scroll pos and size (but may be expended to include e.g. some filtering data, etc.). This now makes lists completely independent from Panels!
This UIList has a draw_item callback which allows to customize items' drawing from python, that all addons can now use. Incidentally, this also greatly simplifies the C code of this widget, as we do not code any "special case" here anymore!
To make all this work, other changes were also necessary:
* Now all buttons (uiBut struct) have a 'custom_data' void pointer, used currently to store the uiList struct associated with a given uiLayoutListBox.
* DynamicPaintSurface now exposes a new bool, use_color_preview (readonly), saying whether that surface has some 3D view preview data or not.
* UILayout class has now four new (static) functions, to get the actual icon of any RNA object (important e.g. with materials or textures), and to get an enum item's UI name, description and icon.
* UILayout's label() func now takes an optional 'icon_value' integer parameter, which if not zero will override the 'icon' one (mandatory to use "custom" icons as generated for material/texture/... previews).
Note: not sure whether we should add that one to all UILayout's prop funcs?
Note: will update addons using template list asap.
This fixes some "regressions" introduced in rev50781 which lead to much
worse solution in some cases. Now it's possible to bring old behavior back.
Perhaps it's more like temporal solution for time being smarter solution is
found. But finding such a solution isn't so fast, so let's bring manual
control over reprojection usage.
But anyway, imo it's now nice to have a structure which could be used to
pass different settings to the solver.
- Fix for copy default settings from active track operator
- Add meaningful tracking presets
API changes:
- Added parameter exact to Marker.find_frame, so now it's
possible to get estimated marker
- Added Marker.pattern_bound_box to get pattern's bound box
* Code cleanup, removed unneeded code.
* Style cleanup, don't break lines to early
(unless marked as pep8-80 or pep8-120 compliant)
* Keep 1 line after layout declaration empty.
In contrast to start_frame (which affects on where footage actually
starts to play and also affects on all data associated with a clip
such as motion tracking, reconstruction and so on) this slider only
affects on a way how frame number is mapping to a filename, without
touching any kind of tracking data.
The formula is:
file_name = clip_file_name + frame_offset - (start_frame - 1)
- Display track's reprojection error in dopesheet
- Make sure track is selected when clicking on dopesheet channel
- Attempt to make headers a bit cleaner without long labels which
doesn't actually make sense.
It was a bit confusing to synchronize settings used in pre-calculated
dopesheet channels which was storing in tracking data with settings
used for display which is in space data.
This was initially done by converting one flags to other and checking
if space's settings matches pre-calculated one, but that had several
issues if two different dopesheet are using different settings:
- Channels would be re-calculated on every redraw for each of spaces
- Dopesheet operators could fail due to the could be using channels
calculated for other space.
That was also quite nasty code checking if requested settings matches
pre-calculated one.
Added option to use Grease Pencil datablock as a mask for pattern
when doing motion tracking. Option could be found in Tracking Settings
panel.
All strokes would be rasterized separately from each other and every
stroke is treating as a closed spline.
Also added option to apply a mask on track preview which is situated
just after B/B/W channel button under track preview.
===========================================
Major list of changes done in tomato branch:
- Add a planar tracking implementation to libmv
This adds a new planar tracking implementation to libmv. The
tracker is based on Ceres[1], the new nonlinear minimizer that
myself and Sameer released from Google as open source. Since
the motion model is more involved, the interface is
different than the RegionTracker interface used previously
in Blender.
The start of a C API in libmv-capi.{cpp,h} is also included.
- Migrate from pat_{min,max} for markers to 4 corners representation
Convert markers in the movie clip editor / 2D tracker from using
pat_min and pat_max notation to using the a more general, 4-corner
representation.
There is still considerable porting work to do; in particular
sliding from preview widget does not work correct for rotated
markers.
All other areas should be ported to new representation:
* Added support of sliding individual corners. LMB slide + Ctrl
would scale the whole pattern
* S would scale the whole marker, S-S would scale pattern only
* Added support of marker's rotation which is currently rotates
only patterns around their centers or all markers around median,
Rotation or other non-translation/scaling transformation of search
area doesn't make sense.
* Track Preview widget would display transformed pattern which
libmv actually operates with.
- "Efficient Second-order Minimization" for the planar tracker
This implements the "Efficient Second-order Minimization"
scheme, as supported by the existing translation tracker.
This increases the amount of per-iteration work, but
decreases the number of iterations required to converge and
also increases the size of the basin of attraction for the
optimization.
- Remove the use of the legacy RegionTracker API from Blender,
and replaces it with the new TrackRegion API. This also
adds several features to the planar tracker in libmv:
* Do a brute-force initialization of tracking similar to "Hybrid"
mode in the stable release, but using all floats. This is slower
but more accurate. It is still necessary to evaluate if the
performance loss is worth it. In particular, this change is
necessary to support high bit depth imagery.
* Add support for masks over the search window. This is a step
towards supporting user-defined tracker masks. The tracker masks
will make it easy for users to make a mask for e.g. a ball.
Not exposed into interface yet/
* Add Pearson product moment correlation coefficient checking (aka
"Correlation" in the UI. This causes tracking failure if the
tracked patch is not linearly related to the template.
* Add support for warping a few points in addition to the supplied
points. This is useful because the tracking code deliberately
does not expose the underlying warp representation. Instead,
warps are specified in an aparametric way via the correspondences.
- Replace the old style tracker configuration panel with the
new planar tracking panel. From a users perspective, this means:
* The old "tracking algorithm" picker is gone. There is only 1
algorithm now. We may revisit this later, but I would much
prefer to have only 1 algorithm. So far no optimization work
has been done so the speed is not there yet.
* There is now a dropdown to select the motion model. Choices:
* Translation
* Translation, rotation
* Translation, scale
* Translation, rotation, scale
* Affine
* Perspective
* The old "Hybrid" mode is gone; instead there is a toggle to
enable or disable translation-only tracker initialization. This
is the equivalent of the hyrbid mode before, but rewritten to work
with the new planar tracking modes.
* The pyramid levels setting is gone. At a future date, the planar
tracker will decide to use pyramids or not automatically. The
pyramid setting was ultimately a mistake; with the brute force
initialization it is unnecessary.
- Add light-normalized tracking
Added the ability to normalize patterns by their average value while
tracking, to make them invariant to global illumination changes.
Additional details could be found at wiki page [2]
[1] http://code.google.com/p/ceres-solver
[2] http://wiki.blender.org/index.php/Dev:Ref/Release_Notes/2.64/Motion_Tracker
One side of change is related on making code easier to follow, due it started
being quite messy because of all in-lined mode/view checks. Now there's a bit
of code duplication, but it's much easier to see what's going on there.
Another side of patch is related on re-arranging elements in header in a way
that follows rule "depending elements are placed after elements they depends on".
This might be a bit against mostly-used-based elements placement, but now it's
much easier to figure out where to add new option. Also it fits better other
blender's areas such as image editor header, i.e.
Now it's indicates at which scene frame number movie clip starts playing back.
This this setting is still belongs to clip datavlock and used by all users of
clip such as movie compositor nodes, constraints and so.
After long discussion and thoughts about this it was decided that this would
match image's current behavior (which initially seen a bit crappy), but that's
actually allows:
- Keep semantics of start frame in image and clip datablocks in sync
- Allows to support features like support of loading image sequences
with crappy numbers in suffix which doesn't fit long int.
- Allows to eliminate extra boolean checkbox to control such kind of offset.
Hopefully from pipeline POV it wouldn't hurt because idea of having this things
implemented in original way was working only if sequence before processing
started naming form 001.
Number of start frame in opened image sequence used to be distinguished automatically
in a way that file name used on open would be displayed at scene frame #1.
But sometimes it's useful to have it manually configurable (like in cases when you're
processing image sequence and replacing clip's filepath to postprocessed image sequence
and want new clip to show at the same frame range as it was rendered from).
Added Custom Start Frame flag to movie clip (could be accessed from Footage panel in
clip editor) and Start Frame which means number of frame from sequence which would
be displayed at scene frame #1.
For example if you've got clip pointing to file render_00100.png and Start Frame of 100
this file would be displayed at scene frame #1, if Start Frame is 1 then this image
would be displayed at scene frame #100,
