As far as I know the CIPIC HRTF database is not diffuse field compensated. So in order to get a "correct" frequency response I should apply some eq. Or was that taken care of?
Have you read their paper? According to the paper, they were "free-field compensated". Measurements were made 1 meter from the ear, so if you want to simulate larger distances, you'll need additional delays, EQ, etc. As long as your additional processing is linear, you should be able to apply it to the SphericalPanner output (not necessary to process each impulse response).
Also when I want to use my own HRTF database how would I need to arrange the file so it works?
If it's your own recordings (or if they are in the public domain), we'd be happy to pre-process them, put them into the format expected by Spherical Panner, and include them in a new subfolder of the Ears files.