tracker/pt: declare floating-point type size in one place

We want double precision for POSIT. It's best for the type to be set in
ope place without the need to go over everything while switching it back
and forth during tests.

Machine epsilon for float is very small as per
<https://en.wikipedia.org/wiki/Machine_epsilon>. Also see the absurdly
high epsilon of 1e-4 of POSIT that we've had. With floats, making the
epsilon lower resulted in change deltas flushing to zero. This typically
led to the translation Z value being very unstable in PT.

After the epsilon and data type size changes the Z value is stable.

1 files changed, 22 insertions, 6 deletions

diff --git a/tracker-pt/ftnoir_tracker_pt_settings.h b/tracker-pt/ftnoir_tracker_pt_settings.h
index e4bfa371..830bf92e 100644
--- a/tracker-pt/ftnoir_tracker_pt_settings.h
+++ b/tracker-pt/ftnoir_tracker_pt_settings.h
@@ -6,8 +6,26 @@
  * copyright notice and this permission notice appear in all copies.
  */
 
-#ifndef FTNOIR_TRACKER_PT_SETTINGS_H
-#define FTNOIR_TRACKER_PT_SETTINGS_H
+#pragma once
+
+#include <limits>
+#include <opencv2/core.hpp>
+
+struct pt_types
+{
+    using f = double;
+
+    static constexpr f eps = std::numeric_limits<f>::epsilon();
+    static constexpr f pi = f(3.14159265358979323846);
+
+    template<int n> using vec = cv::Vec<f, n>;
+    using vec2 = vec<2>;
+    using vec3 = vec<3>;
+
+    template<int y, int x> using mat = cv::Matx<f, y, x>;
+    using mat33 = mat<3, 3>;
+    using mat22 = mat<2, 2>;
+};
 
 #include "opentrack-compat/options.hpp"
 using namespace options;
@@ -28,9 +46,9 @@ struct settings_pt : opts
 
     value<int> clip_ty, clip_tz, clip_by, clip_bz;
     value<int> active_model_panel, cap_x, cap_y, cap_z;
-    
+
     value<int> fov;
-    
+
     value<bool> dynamic_pose;
     value<int> init_phase_timeout;
     value<bool> auto_threshold;
@@ -67,5 +85,3 @@ struct settings_pt : opts
         auto_threshold(b, "automatic-threshold", false)
     {}
 };
-
-#endif //FTNOIR_TRACKER_PT_SETTINGS_H