fix a serious bug in collision mode (rigid bodies should have been initialized as kinematic, to prevent any movement, because active rigid bodies are used because of better performance) This never happened to me since I didn't test intersecting objects.. my fault.

thanks again to Gael Rosset for finding a case where this problem showed clearly.

scripts/addons/cam/collision.py

@@ -74,8 +74,9 @@ def prepareBulletCollision(o):
 			bpy.ops.object.convert(target='MESH', keep_original=False)
 
 		collisionob=bpy.context.active_object
-		bpy.ops.rigidbody.object_add(type='ACTIVE')
+		bpy.ops.rigidbody.object_add(type='ACTIVE')#using active instead of passive because of performance.TODO: check if this works also with 4axis...
 		collisionob.rigid_body.collision_shape = 'MESH'
+		collisionob.rigid_body.kinematic=True#268 and 70 
 		collisionob.rigid_body.collision_margin = o.skin*BULLET_SCALE
 		bpy.ops.transform.resize(value=(BULLET_SCALE, BULLET_SCALE, BULLET_SCALE), constraint_axis=(False, False, False), constraint_orientation='GLOBAL', mirror=False, proportional='DISABLED', proportional_edit_falloff='SMOOTH', proportional_size=1, snap=False, snap_target='CLOSEST', snap_point=(0, 0, 0), snap_align=False, snap_normal=(0, 0, 0), texture_space=False, release_confirm=False)
 		collisionob.location=collisionob.location*BULLET_SCALE

scripts/addons/cam/image_utils.py

@@ -148,7 +148,7 @@ def numpytoimage(a,iname):
 	a=a.reshape(d)
 	a=a.repeat(4)
 	a[3::4]=1
-	#i.pixels=a
+	#i.pixels=a this was 50 percent slower...
 	i.pixels[:]=a[:]#this gives big speedup!	
 	print('\ntime '+str(time.time()-t))
 	return i