And about colour tracking: I see the problem… in my opinion this is something related to jit.findbounds object, not your patch (the same issue was discussed in the topic I mentioned).
But we can create several workarounds. Simplest trick is to split image into 3 components (RGB) and process them independently (see attached patch). It’s not very robust solution, I have something better in mind (based on solution from mentioned topic) and I send you patch based on this better architecture today later.
Jul 5, 2013 at 1:21am #255011