Merge pull request #158 from floe/registration

add info about intrinsic structure by @sh0

Author: floe

examples/protonect/include/libfreenect2/protocol/response.h

@@ -161,8 +161,35 @@ LIBFREENECT2_PACK(struct RgbCameraParamsResponse
   // unknown, always seen as 1 so far
   uint8_t table_id;
 
-  // this block contains at least some color camera intrinsic params
-  float intrinsics[25];
+  // color -> depth mapping parameters
+  float color_f;
+  float color_cx;
+  float color_cy;
+
+  float shift_d;
+  float shift_m;
+
+  float mx_x3y0; // xxx
+  float mx_x0y3; // yyy
+  float mx_x2y1; // xxy
+  float mx_x1y2; // yyx
+  float mx_x2y0; // xx
+  float mx_x0y2; // yy
+  float mx_x1y1; // xy
+  float mx_x1y0; // x
+  float mx_x0y1; // y
+  float mx_x0y0; // 1
+
+  float my_x3y0; // xxx
+  float my_x0y3; // yyy
+  float my_x2y1; // xxy
+  float my_x1y2; // yyx
+  float my_x2y0; // xx
+  float my_x0y2; // yy
+  float my_x1y1; // xy
+  float my_x1y0; // x
+  float my_x0y1; // y
+  float my_x0y0; // 1
 
   // perhaps related to xtable/ztable in the deconvolution code.
   // data seems to be arranged into two tables of 28*23, which

examples/protonect/src/libfreenect2.cpp

@@ -475,10 +475,10 @@ void Freenect2DeviceImpl::start()
   command_tx_.execute(ReadRgbCameraParametersCommand(nextCommandSeq()), result);
   RgbCameraParamsResponse *rgb_p = reinterpret_cast<RgbCameraParamsResponse *>(result.data);
 
-  rgb_camera_params_.fx = rgb_p->intrinsics[0];
-  rgb_camera_params_.fy = rgb_p->intrinsics[0];
-  rgb_camera_params_.cx = rgb_p->intrinsics[1];
-  rgb_camera_params_.cy = rgb_p->intrinsics[2];
+  rgb_camera_params_.fx = rgb_p->color_f;
+  rgb_camera_params_.fy = rgb_p->color_f;
+  rgb_camera_params_.cx = rgb_p->color_cx;
+  rgb_camera_params_.cy = rgb_p->color_cy;
 
   command_tx_.execute(ReadStatus0x090000Command(nextCommandSeq()), result);
   std::cout << "[Freenect2DeviceImpl] ReadStatus0x090000 response" << std::endl;