summaryrefslogtreecommitdiffhomepage
AgeCommit message (Collapse)Author
2017-04-18tracker/pt: make calibrated cross yellowStanislaw Halik
Otherwise it's not visible enough with red-only image.
2017-04-18compat/{simple-mat,euler}: unbreak calibrationStanislaw Halik
No idea what regressed.
2017-04-18cmake: remove skip-translation-update cache varStanislaw Halik
Better way is to checkout .ts files from the current branch after running `make i18n'.
2017-04-18tracker/pt: replace original blob center search with meanshiftStanislaw Halik
The functions are almost identical so why not. I removed several bits: - weighting by squared pixel value is bad. weight by pixel value instead. - making the ROI twice as big doesn't make sense and makes for misdetected blobs. remove it. - switch radius coefficient to something doing less iterations. - sprinkle some __restrict pointer qualifier. - cv::floodfill invocation had some hardcoded flag value. - point radius circle and the bullseye line length weren't adjusted by scaling ratio. while the circle fitted the radius tightly, it was now clutter, so I removed it, leaving only the properly-scaled bullseye. - brightness had to go sadly since it's not accumulated anymore.
2017-04-18logic/tracker: fix option not used in logicStanislaw Halik
2017-04-18tracker/pt: fix tab order in dialogStanislaw Halik
2017-04-18pose-widget: also compute alpha in the loopStanislaw Halik
2017-04-18pose-widget: check area for invalid denomStanislaw Halik
The denom check is a magic number otherwise.
2017-04-17pose-widget: iterate over BGRStanislaw Halik
2017-04-17pose-widget: interpolate next in world coordinate spaceStanislaw Halik
The next value was chosen in view space.
2017-04-17few widgets: use repaint() for no event loop overheadStanislaw Halik
The Qt event loop overhead formed a decent amount of total CPU usage. `repaint()' is a direct call.
2017-04-17cv/video-widget, pose-widget: use premultiplied argb32 surfaceStanislaw Halik
2017-04-17api/plugin: unref inner enumStanislaw Halik
2017-04-17tracker/hatire: include right headerStanislaw Halik
2017-04-17add Intel IPP licensing informationStanislaw Halik
2017-04-17remove stale headtracker licenseStanislaw Halik
2017-04-17filter/accela: revert -180->180 fixStanislaw Halik
Does more harm than good. For 360' inertial devices, recommend other or no filter at all. Issue: #600
2017-04-14tracker/steamvr: use double promotion for all arithmeticStanislaw Halik
The computation order isn't defined so parts might only get promoted to double after arithmetic. Promote all parts manually before arithmetic.
2017-04-14logic/tracker.cpp: explicit set to zeros with custom centerStanislaw Halik
cf. https://github.com/opentrack/opentrack/issues/352#issuecomment-293885962
2017-04-14tracker/steamvr: don't reset seated pose for non-HMDStanislaw Halik
cf. https://github.com/opentrack/opentrack/issues/352#issuecomment-293885962
2017-04-13tracker/steamvr: use normal center for controllersStanislaw Halik
Issue: #352
2017-04-13tracker/steamvr: use doubles for intermediate computationStanislaw Halik
2017-04-13logic/tracker: fix typoStanislaw Halik
2017-04-13logic/tracker: call center() before data()Stanislaw Halik
For real this time. Issue: #352
2017-04-13main: call set-library-paths only after qapplication existsStanislaw Halik
Otherwise accesses qt install prefix. Issue: #598
2017-04-13gui/options: fix default tabStanislaw Halik
2017-04-13gui/options: allow for up to +- 500 cm custom pose translationStanislaw Halik
Related-to: #352
2017-04-13logic/tracker: zero at the right pipeline stageStanislaw Halik
2017-04-12cmake/mingw-w64: mess it up againStanislaw Halik
2017-04-12logic/tracker: no need for timeBeginPeriodStanislaw Halik
There's jitter of up to 3 ms. Increasing timer resolution pessimizes overall performance. For Falcon BMS it costs about a dozen frames.
2017-04-12gui/options: adjust dynamic neck limitStanislaw Halik
2017-04-12matrix -> euler: prevent gimbal lock multiple solutions harderStanislaw Halik
2017-04-12pose-widget: fix some breakage, forgot what it wasStanislaw Halik
2017-04-12gui/mapping: allow for up to 100 cm translationStanislaw Halik
Defaults to 30 cm as before. Changes X snap value for larger max value. Related-to: #352
2017-04-12cv/calibrator: allow for experimental roll calibrationStanislaw Halik
It mostly works from my testing.
2017-04-12compat/util: add few attribute macrosStanislaw Halik
2017-04-12compat/simple-mat: simplify SFINAEStanislaw Halik
variable templates rock ass.
2017-04-12cmake: don't barf out on nonexistent files in lreleaseStanislaw Halik
2017-04-12compat/simple-mat: remove useless unitStanislaw Halik
2017-04-12compat/euler: add incomplete & untested quat codeStanislaw Halik
2017-04-10tracker/rift-140: rename unitStanislaw Halik
2017-04-09tracker/steamvr: add copyrightStanislaw Halik
Issue: #352
2017-04-09Merge pull request #597 from antcodd/steamvr-fixStanisław Halik
tracker/steamvr: Fix rotation issues
2017-04-10tracker/steamvr: Fix rotation issuesAnthony Coddington
Use well known matrix to euler decomposition formula, fixing some incorrect accesses. May need gimbal lock avoidance. Don't use opentrack centering as ResetSeatedZeroPose preserves calibrated real world up (Y) vector. Note: controllers are centered to HMD seated position. fixes most remaining issues in #352
2017-04-07rename several protocols from abbreviated namesStanislaw Halik
2017-04-07proto/flightgear: dialog fixesStanislaw Halik
- add "help" button opening the right contrib directory with readme in it - use layouts in the dialog Issue: #590
2017-04-07proto/flightgear: few issuesStanislaw Halik
- move fgtypes.h contents into module header - fix several inverted DOF - construct a 32-bit ipv4 address from nibblets directly - switch to no alignment for the datagram, was 2 bytes - default to 127.0.0.1 as the target address Issue: #590
2017-04-07cmake: remove PROJECT() scope from subdirectoriesStanislaw Halik
cf. https://bugreports.qt.io/browse/QTCREATORBUG-17955
2017-04-07tracker/steamvr: make staticStanislaw Halik
2017-04-06tracker/steamvr: return more precise radian resultStanislaw Halik
While atan2's input is limited to single-float precision, extend result precision to double-float precision. The result's in radians. Why didn't gradians get popular?