Wow, that sounds usable. I know it's anecdotal but how did you find the quality of the output, and can you compare it to any closed source model?