My original post was going to be a detailed request for an overhaul of the texture extraction system.
In writing it I realized there are likely many valid workflows for texture extraction and the most useful request would be for the following:
A complete, and pythonic, python API as well as pyside support for building QT UIs. Additionally I would request that the development team 'dog-food' the Python API and by re-writing the existing texture extraction tools and whatever else might make sense (like a alpha manager or something). This would be a great time to think about what the lowest level programming interface to the texture extraction really should be.
Minor texture extraction improvement:
Our artists can spend a significant amount of time finding and painting out artifacts introduced by the raytrace extraction process. While we have some tools to aid in the process I believe this could be fixed at it's source.
In my mind the easiest way to handle the spikes would be to detect if any pixel samples 'miss' geometry and set the pixel's alpha to black then use the same post-process dilation technique used for borders to fill the holes (letting the user know what you've done). I'm sure there are other solutions as well.
