Skip to content
This repository

HTTPS clone URL

Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
branch: master

Jan 28, 2011

  1. clojure support version 0.0.1

    Jon Vlachoyiannis authored
  2. clojure support version 0.0.1

    Jon Romero authored

Jan 06, 2011

  1. Marcos Paulo Berteli Slomp

    fixed the memory-leak detection include headers to prevent compilatio…

    …n errors in debug mode on MSVC prior to 2010.
    
    several cleanups and additions regarding thread synchronization. prepared source code for a more advanced form of libusb_handle_events() that will consume less CPU (not ready yet, code is still using the old processing loop).
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  2. Marcos Paulo Berteli Slomp

    a few extra comments for MSVC 2008 and prior projects.

    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  3. Marcos Paulo Berteli Slomp

    changed the PKTS_PER_XFER and NUM_XFERS (from 16/16 to 32/8, respecti…

    …vely) for WIN32 platform builds.
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  4. Marcos Paulo Berteli Slomp

    added helper function 'libusbemu_clear_transfer()' to facilitate 'lib…

    …usbemu.cpp' tasks
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed

Jan 04, 2011

  1. Juan Carlos del Valle

    Functionalities added to the client and server. Some AS3 library impr…

    …ovements.
    
    Signed-off-by: Juan Carlos del Valle <jc.ekinox@gmail.com> (imekinox)
    imekinox authored
  2. Juan Carlos del Valle

    AS3 client library was completelly redefined. It wraps now almost an…

    …y functionality of freenect driver.
    
        Server was changed to use a single socket and a client-server communication protocol was defined too.
        Reduced server cpu usage and tested against memory leaks.
        Working in windows and osx (linux should work too)
    
        Signed-off-by: Juan Carlos del Valle <jc.ekinox@gmail.com> (imekinox)
    imekinox authored

Dec 29, 2010

  1. Brandyn White

    Initial commit of matlab wrapper (minally tested, need feedback)

    Signed-off-by: Brandyn A. White <bwhite@dappervision.com>
    bwhite authored qdot committed
  2. Brandyn White

    Cleaned up the demos to reflect decisions made from the forum/irc dis…

    …cussions.
    
    1. Abstracted the frame conversion code to frame_convert.py.  This will prevent the massive changes that we have been seeing as all of the duplicative code is in there now.  This makes optimization and normalization experiments cleaner to test out.
    2. Removed demo_ipython and demo_kill_async as they are mostly duplicates of the other demos
    3. Made the "multi" demo default to using all kinects at once instead of one at a time
    4. Change the default normalization to make better use of the 8 bit range.
    
    Signed-off-by: Brandyn A. White <bwhite@dappervision.com>
    bwhite authored qdot committed

Dec 28, 2010

  1. Brandyn White

    Cleaned up demos as per this post

    http://groups.google.com/group/openkinect/browse_thread/thread/c72b3ade917152de
    
    1. Every demo that runs forever has a way to kill (which is printed when you start the program)
    2. The normalization is now consistent (uses truncation which amiller and I believe is a nicer visualization than normalization)
    3. Updated readme to include these changes
    4. Added a new demo that sweeps through the depth image displaying slices of depth at a time
    5. Added more visibility to the raw_tilt_state struct in freenect.pyx
    6. Removed "../include/libfreenect.hpp" from src/CMake* as it doesn't exist there anymore
    
    Signed-off-by: Brandyn A. White <bwhite@dappervision.com>
    bwhite authored
  2. Uses bit shift vs. adaptative normalization

    Adaptative normalization to the current frame extrema is ok
    for visualizing but turns data meaningless for any other purpose.
    So in order to keep the example useful for all those purpose
    the dept are just shifted to maximize the dynamic range of the
    received 11 bits over the 16 representation bits.
    
    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  3. demo_cv_async: clean up a debug variable

    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  4. python async example exits on esc and avoids GC

    python wrapper example demo_cv_async now avoids temporary
    reallocations on video bgr reordering that provoqued GC stalls.
    
    Also adds gracefull exit on ESC by using the Kill signal.
    
    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  5. demo_cv_sync_multi: separate views for each device

    Create a different pair of windows for each device
    
    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  6. Drew Fisher

    cameras.c: fix a typo that results in the use of the wrong size buffe…

    …r when requesting packed 10-bit depth data.
    
    Thanks to David García Garzón, who pointed out the error.
    
    Signed-off-by: Drew Fisher <drew.m.fisher@gmail.com> (zarvox)
    zarvox authored qdot committed

Dec 27, 2010

  1. Juan Carlos del Valle

    Usign new c_sync definition for motor and accelerometers. Fixed CMake…

    …Lists little error.
    
    Signed-off-by: Juan Carlos del Valle <jc.ekinox@gmail.com> (imekinox)
    imekinox authored qdot committed
  2. updates to libfreenect.hpp didnt compile on linux gcc

    Signed-off-by: Kai Ritterbusch <kai.ritterbusch@gmx.de>
    Kai R authored qdot committed
  3. Marcos Paulo Berteli Slomp

    added QuickThread::Sleep() and QuickThread::Yield()

    added QuickThread::Myself() to get a QuickThread object that operates on the calling thread.
    
    libusbemu has no platform-specific dependencies anymore: system-specific wrappers are provided through libusbemu_threads.h
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  4. Marcos Paulo Berteli Slomp

    renamed EXPORT to FREENECTAPI to avoid name collisions with client code.

    removed '#include <libusb10/libusb.h>' dependency for WIN32 client code (freenect_usb_context is actually masqueraded as void type, and because of that, 'core.c' required an additional pointer cast)
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  5. Marcos Paulo Berteli Slomp

    added support for the sleep() function to be called under a Win32 env…

    …ironment
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  6. Marcos Paulo Berteli Slomp

    few extra comments, nothing really important...

    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  7. Marcos Paulo Berteli Slomp

    re-added guard to allow compilation under MinGW-gcc

    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  8. Marcos Paulo Berteli Slomp

    using namespace-defined libusbemu::MIN instead of __min non-standard …

    …function or MIN macro
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  9. ESC exits gracefully from cv based sync examples

    Detecting the esc key to exit and calling synch_stop to
    switch off the IR projector.
    
    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  10. Avoiding depth overflow in ipython example

    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  11. No reallocations and GC stalls in python examples

    Calling as_type creates a temporary np matrix that is
    allocated and deallocated all the time. This makes
    the performance to degenerate after a short time.
    
    This was solved by preallocating a matrix for the
    bgr->rgb inversion and using directly the depth map.
    
    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed
  12. Python: Avoiding 8bits overflow of 11bits depth

    In python wrapper exemples for sync interface,
    11bit depth data was mapped to 8bits without shift
    and overflowed.
    Now displaying a 16 bit depth image and shifting
    all the depth pixels 5 bits left to maximize the
    dynamic range of the image.
    
    Signed-off-by: David García Garzón <david.garcia@barcelonamedia.org>
    David García Garzón authored qdot committed

Dec 26, 2010

  1. Kyle Machulis

    Hopefully the end of the windows platform fix changes

    Signed-off-by: Kyle Machulis <kyle@nonpolynomial.com>
    qdot authored
  2. Kyle Machulis

    Even more fixes from rebase issues with libusbemu.

    Signed-off-by: Kyle Machulis <kyle@nonpolynomial.com>
    qdot authored
  3. Kyle Machulis

    Removed more unneeded code left in due to bad rebase.

    Signed-off-by: Kyle Machulis <kyle@nonpolynomial.com>
    qdot authored
  4. Kyle Machulis

    Fixed accidental commit of unresolved conflicts

    Signed-off-by: Kyle Machulis <kyle@nonpolynomial.com>
    qdot authored
  5. Marcos Paulo Berteli Slomp

    improved _WIN32 guard that exports the freenect API: it is now possib…

    …le to build the library in Windows with a Win32 port of gcc without being forced to compile C files as C++ files.
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  6. Marcos Paulo Berteli Slomp

    lots of new useful information, mainly on the motivation and concept …

    …behind libusbemu.
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
  7. Marcos Paulo Berteli Slomp

    code is now using libusbemu thread back-end.

    added 'libusb_internal.h' to better separate pieces of code and prevent name collisions through namespace directives. libusb_handle_events() do not use a global mutex variable anymore: every context now has its own mutex.
    
    Signed-off-by: Marcos Paulo Berteli Slomp <mslomp@gmail.com>
    slomp authored qdot committed
Something went wrong with that request. Please try again.