summaryrefslogtreecommitdiffhomepage
AgeCommit message (Collapse)Author
2015-12-18tracker/pt: nix unusedStanislaw Halik
2015-12-18tracker/pt: don't fill mask on frameStanislaw Halik
Saves few % of cpu load
2015-12-18pose-widget: also bilinear interpolation of alpha valueStanislaw Halik
2015-12-18ui: adjust marginopentrack-2.3-rc21p5Stanislaw Halik
2015-12-18ui: make more compactStanislaw Halik
2015-12-18glwidget: use transparent octopus backgroundStanislaw Halik
2015-12-18api/mat: fix typos/breakageStanislaw Halik
2015-12-17api/joy: refresh only manually on certain eventsStanislaw Halik
Refresh joylist when new listener arrives, and when the joy singleton just gets created. Enumerating joys all the time causes high CPU usage for some of the users. Issue: #279 Backtrace-by: @exulant Reported-by: @aka2k
2015-12-17pt: histogram more granular 6 -> 8Stanislaw Halik
256 is divisible by 8, also less cpu usage
2015-12-17cmake/api: link with strmiids.lib on win32Stanislaw Halik
It worked by accident since cmake/compat already links to it.
2015-12-08tracker/pt: reduce auto thresholding histogram bucket sizeStanislaw Halik
Previously it was too slow to 640x480@75.
2015-12-06api/keys: prevent idempotent keysopentrack-2.3-rc21p4Stanislaw Halik
2015-12-06api/joy: move from headerStanislaw Halik
2015-12-06api/joy: prevent idempotent keypressed passed to receiverStanislaw Halik
2015-12-06compat/options: get rid of std::string usageStanislaw Halik
It's pointless to use it here.
2015-12-06compat/options: move from headerStanislaw Halik
2015-12-06gui/settings: set parent, otherwise not modalStanislaw Halik
2015-12-06gui/settings: don't forget to show a modal dialog before executingStanislaw Halik
2015-12-06gui/main: don't raise a new window, it's enough to set visibleStanislaw Halik
2015-12-06api/joy: speed up poll_axis pathStanislaw Halik
We don't really need to poll for joysticks in tracker/joy.
2015-12-06api/joy: nix static, now that we're not a singletonStanislaw Halik
2015-12-06tracker/joy: adapt to non-singleton joy workerStanislaw Halik
2015-12-06joystick: no longer singleton, use fake window handleStanislaw Halik
We can create arbitrary amount of dinput handles, given they're passed unique window handles.
2015-12-06api/keys: use a fake window for DirectInput handleStanislaw Halik
2015-12-06gui/keys: allow for pausing global keystrokes for options dialogStanislaw Halik
Toggling tracking while prompting for toggle tracking key causes a deadlock somewhere.
2015-12-06api/keys: nix tautological #ifdefStanislaw Halik
2015-12-06contrib/aruco: oops, right extensionStanislaw Halik
2015-12-06contrib/aruco: use @frost555's marker imageStanislaw Halik
2015-12-06api/camera-names: move to compat/Stanislaw Halik
2015-12-06api/camera: move from headerStanislaw Halik
There's no need to have a copy in each module.
2015-12-06api/keys: initialize hresultStanislaw Halik
2015-12-06api/keys: fix buildStanislaw Halik
2015-12-06compat/shm: fix buildStanislaw Halik
2015-12-06api, main: implement global shortcuts for controlling trackingStanislaw Halik
Issue: #252 That involves moving stuff around to get rid of circular dependencies. You need to bind keys to shortcuts again this once.
2015-12-06tracker/aruco, tracker/pt: sleep before releasing cameraStanislaw Halik
Really fast toggling tracking crashed with my PS3 Eye.
2015-12-05gui: add shortcuts for starting/toggling trackingStanislaw Halik
2015-12-05api/shortcuts: no longer keep shortcuts hardcodedStanislaw Halik
2015-12-05api/keys: fix multiple shortcut receiversStanislaw Halik
We can't depend on std::vector's address of array subscript to remain constant, duh.
2015-12-04aruco: smaller search windowopentrack-2.3-rc21p3Stanislaw Halik
Suitable for 30 Hz camera modes and higher. Requested-by: @frost555 Issue: #273
2015-12-03aruco: use a single standard box sizeStanislaw Halik
Higher box sizes use more CPU due to the need to convolve a lot. It looks fine with both high and low exposure on both Logitech C525 and PS3 Eye webcams. Issue: #273
2015-12-03aruco: use only Otsu thresholding, never adaptiveStanislaw Halik
Detection rate stays as good, likely better as before. @mursey reports in #274 that non-Otsu case eats way more CPU.
2015-12-03api/joystick: refresh more often, it's cheap enough cpu-wiseStanislaw Halik
2015-12-03api/joy: there's no joy_handle == nullptr case, don't check for itStanislaw Halik
2015-12-03compat/shm: move header inclusion after guardStanislaw Halik
2015-12-03api, compat: clean up verbose fprintf logspamStanislaw Halik
2015-12-03api/joy: fix raceStanislaw Halik
erase from iterator referenced the wrong 'joys' variable
2015-12-03accela: don't clamp max speed to just few degreesStanislaw Halik
2015-11-29aruco: rework timeout logicopentrack-2.3-rc21p2Stanislaw Halik
- cycle otsu twice before selecting new box size don't cycle otsu each frame. - slower backoff from timeout value before new box size on occasional detections need at least 5 successes for each failure to not exceed. - select more sensible timeout per box size of .35 seconds. this is enough also with 30 Hz webcams, but too little for 15 Hz modes.
2015-11-29api/joystick: drop locking, add commentStanislaw Halik
2015-11-29api/joystick: oops, access the right 'joys' variableStanislaw Halik