|
|
@@ -1 +1,11 @@
|
|
|
WIP
|
|
|
+
|
|
|
+NOTE to self and other: This probably isn't gonna work, 32bit fixed point has
|
|
|
+showed to be too low precision for a 3D physics engine (even though it is enough
|
|
|
+for 3D rendering). The issue is that we need to represent both very high and
|
|
|
+very low energies so we can allocate about 9 bits to the fractional part of the
|
|
|
+number, however this is too little to deal with slow rotations and low energies;
|
|
|
+it happens too often that for example objects balance on their edges because
|
|
|
+their energy gets quite small and rounded to zero. The project probably needs to
|
|
|
+be rewritten from scratch with more precision (still wouldn't use floating point
|
|
|
+but rather some tiny custon fixed point library).
|
Miloslav Ciz