- Split display options into two panels:
* Display, which contains all general display options related on editor itself
* Marker Display, which contains options which makes sense for markers themselves
and not used so often.
- Remove Show GPencil option which doesn't actually make sense because strokes might
be disabled from GPencil panel
- Removed Show Pyramid option and show pyramid for KLT tracker as default.
- Added hotkey to toggle Show Disabled (Alt-D)
- Added hotkey to toggle Show Pattern (Alt-S)
Changes in interface and hotkeys are done by Sebastian Koenig.
Thanks for the patch!
Own changes in interface were related on Aspect Ratio. It doesn't make
much sense for tracking (camera pixel aspect is what you'll want to use,
but for masking it would make much more sense because when you just
want to create a mask for footage you don't actually have camera).
Just removing from interface seems not so nice for now.
Merged from soc-2011-tomato rev43872
Initial idea was to perform bilinear filtering for displaying proxied frame
to make it looking a bit smoother. It was done but it was also discovered
that using such kind of texture buffers helps on some crappy videocards
when playing $k footage.
Currently check for NPOT support is disabled, so use this option with care.
- Split display options into two panels:
* Display, which contains all general display options related on editor itself
* Marker Display, which contains options which makes sense for markers themselves
and not used so often.
- Remove Show GPencil option which doesn't actually make sense because strokes might
be disabled from GPencil panel
- Removed Show Pyramid option and show pyramid for KLT tracker as default.
- Added hotkey to toggle Show Disabled (Alt-D)
- Added hotkey to toggle Show Pattern (Alt-S)
Changes in interface and hotkeys are done by Sebastian Koenig.
Thanks for the patch!
Own changes in interface were related on Aspect Ratio. It doesn't make
much sense for tracking (camera pixel aspect is what you'll want to use,
but for masking it would make much more sense because when you just
want to create a mask for footage you don't actually have camera).
Just removing from interface seems not so nice for now.
- Renamed graph_jump_to_current_frame to graph_center_current_frame
which makes more sense.
- Curve view now can be locked to time cursor (Lock to Time Cursor in
Display panel or L button in curve view). Not sure if offset from
locked position will make much sense here.
- Added hotkey for solving -- Shift-S.
- Fixed tooltip displaying for track sequence forwards in clip editor
- Corrected detection of 8 tracks so it wouldn't count tracks disabled
on keyframes.
- Scale track preview to actual track widget size instead of scaling the
whole preview image with given zoom ratio, so no extra memory needed to
store zoomed margin would be used.
- Track's statistics text will fit pattern position instead of search if
marker is disabled on current frame.
- Fixed toggle selection operator if selected track is hidden due to
"Hide Disabled" policy.
This commit implements basic clipboard support for movie tracking data
int clip editor. Used own implementation of clipboard like it's done
for sequencer.
Ideally it needed to be switched to more general clipboard system, but
currently this system is designed for text data only and it need to
be re-designed itself. But this feature is quite useful since object
tracking is implemented, so it should be OK to live with such own
implementation for a while.
Added slider to define scale of object solution which is used to define
"depth" of object relative to camera position. This slider effects on all
"users" of object solution such as bundles display, constrained objects and so.
Added new operator called "Set Solution Scale" to set real scale for object
solution based on real distance between two bundles reconstructed for this object.
New slider and operator can be found on "Object" panel in toolbox when in
reconstruction mode and active tracking object isn't a camera.
This commit implements basis stuff needed for object tracking,
use case isn't perfect now, interface also should be cleaned a bit.
- Added list of objects to be tracked. Default there's only one object called
"Camera" which is used for solving camera motion. Other objects can be added
and each of them will have it;s own list of tracks. Only one object can be used
for camera solving at this moment.
- Added new constraint called "Object Tracking" which makes oriented object be
moving in the save way as solved object motion.
- Scene orientation tools can be used for orienting object to bundles.
- All tools which works with list of tracks or reconstruction data now
gets that lists from active editing object.
- All objects and their tracking data are available via python api.
libmv. Since both KLT and Hybrid rely on ESM underneath, KLT and Hybrid now
have a minimum correlation setting to match. With this fix, track failures
should get detected quicker, with the issue that sometimes the tracker will
give up too easily. That is fixable by reducing the required correlation (in
the track properties).
Comment from Keir's commit:
Add a new hybrid region tracker for motion tracking to libmv, and
add it as an option (under "Hybrid") in the tracking settings. The
region tracker is a combination of brute force tracking for coarse
alignment, then refinement with the ESM/KLT algorithm already in
libmv that gives excellent subpixel precision (typically 1/50'th
of a pixel)
This also adds a new "brute force" region tracker which does a
brute force search through every pixel position in the destination
for the pattern in the first frame. It leverages SSE if available,
similar to the SAD tracker, to do this quickly. Currently it does
some unnecessary conversions to/from floating point that will get
fixed later.
The hybrid tracker glues the two trackers (brute & ESM) together
to get an overall better tracker. The algorithm is simple:
1. Track from frame 1 to frame 2 with the brute force tracker.
This tries every possible pixel position for the pattern from
frame 1 in frame 2. The position with the smallest
sum-of-absolute-differences is chosen. By definition, this
position is only accurate up to 1 pixel or so.
2. Using the result from 1, initialize a track with ESM. This does
a least-squares fit with subpixel precision.
3. If the ESM shift was more than 2 pixels, report failure.
4. If the ESM track shifted less than 2 pixels, then the track is
good and we're done. The rationale here is that if the
refinement stage shifts more than 1 pixel, then the brute force
result likely found some random position that's not a good fit.
svn command used: svn merge -r 42375:42376 -r 42377:42379 ^/branches/soc-2011-tomato
This commit implements:
- Configurable settings for newly creating tracks
Now it's possible to set tracking algorithm and it's settings for
all newly creating tracks including manual tracks creation and
tracks creation by "Detect Features" operator.
- Moves margin, frames limit and adjust frame inside per-track
settings.
Was request from Francois for this.
- Adjust Frames replaced with menu called Pattern Match where it's
possible to choose between matching pattern from keyframe frame
or from previously tracked frame.
Didn't see somebody used adjust frames values differ from 0 and 1,
and this menu should make things more clear here/
This operator does needed changes to
- 3D viewport
- Scene settings
- World settings
- Compositor
- Scene objects
in a way scene becomes ready to be composited into footage.
Known issue: preview doesn't work "out-of-box" after running this script,
selecting View node and hitting Tab helps. Not sure it can be solved
in nicer way at this moment.
- Changed some names so now people who aren't really familiar with
motion tracking can understand what they exactly means
- Also cleaned up and rephraded some descriptions
- Changed behavior of operator which creates empty for 2d tracks:
now it operates on all selected tracks rather than active track only
- Added checkbox to enable/disable rotation stabilization
- Move tracking-related constraints to own section in list
Currently there are only two constraints, so can look a bit odd,
but it'll be other constraints like "Object Solver" and so.
- Move motion-tracking parameters from 3D viewport Display panel
to it's own panel.
- Get rid of "Bundle" in 3d viewport. It's quite obvious that it's
a 3D representation of tracks is used in 3D viewport and it shouldn't
be so confusing for artists now.
- Also get rid of "Bundle" in Follow Track constraint.
Old files can change a bit because of changes in DNA.
- Also get rid of "Bundles" in operator which creates vertices cloud
from 3D position of tracks.
- Rename "Principal Point" to "Optical Center" in the interface.
- Add support for refining the camera's intrinsic parameters
during a solve. Currently, refining supports only the following
combinations of intrinsic parameters:
f
f, cx, cy
f, cx, cy, k1, k2
f, k1
f, k1, k2
This is not the same as autocalibration, since the user must
still make a reasonable initial guess about the focal length and
other parameters, whereas true autocalibration would eliminate
the need for the user specify intrinsic parameters at all.
However, the solver works well with only rough guesses for the
focal length, so perhaps full autocalibation is not that
important.
Adding support for the last two combinations, (f, k1) and (f,
k1, k2) required changes to the library libmv depends on for
bundle adjustment, SSBA. These changes should get ported
upstream not just to libmv but to SSBA as well.
- Improved the region of convergence for bundle adjustment by
increasing the number of Levenberg-Marquardt iterations from 50
to 500. This way, the solver is able to crawl out of the bad
local minima it gets stuck in when changing from, for example,
bundling k1 and k2 to just k1 and resetting k2 to 0.
- Add several new region tracker implementations. A region tracker
is a libmv concept, which refers to tracking a template image
pattern through frames. The impact to end users is that tracking
should "just work better". I am reserving a more detailed
writeup, and maybe a paper, for later.
- Other libmv tweaks, such as detecting that a tracker is headed
outside of the image bounds.
This includes several changes made directly to the libmv extern
code rather expecting to get those changes through normal libmv
channels, because I, the libmv BDFL, decided it was faster to work
on libmv directly in Blender, then later reverse-port the libmv
changes from Blender back into libmv trunk. The interesting part
is that I added a full Levenberg-Marquardt loop to the region
tracking code, which should lead to a more stable solutions. I
also added a hacky implementation of "Efficient Second-Order
Minimization" for tracking, which works nicely. A more detailed
quantitative evaluation will follow.
Original patch by Keir, cleaned a bit by myself.
- Lock to selection and center to selection will now work fine with undistorted rendering
- Do not display pyramid for disabled tracks
- Corrected fix for wrong correlation_min property name
