A mechanism for processing video data is disclosed. Video decoder initialization information (DII) is determined. The DII includes data describing a maximum level (MaxLevel) to which all coded video sequences (CVSs) of the bitstream conform and a maximum format that comprises one or more parameters. The MaxLevel indicates a first maximum DPB size needed for decoding the bitstream. The maximum format indicates a second maximum DPB size, denoted as DPB memory for as maximum format (DpbMemory4MaxFormat), determined by simultaneously setting all parameters of the maximum format to largest values. The DII further comprises a first indication of whether the DpbMemory4MaxFormat exceeds the first maximum DPB size. A conversion is performed between a visual media data and a bitstream based on the indication.