- no camera mode: only return transform and zoom, but not modify it. User can decide how to combine result with other camera moving features
- more accurate mode where fingers stay on the same world position, so something like strategic zoom is possible. Currently all movement is based on viewport coordinates only