Disentangled usage of HLT and offline one pad clusters