Works for all item types, and has much less functions and parameters than the drawable transform API because it uses the new context states. Untested and not finished!
