Due to streamer limitations (really Linux kernel versions vs. DAC) I've downsampled a lot of DSD to 96/24 and honestly I cannot tell the difference. So my belief is that differences in DSD may be due to mastering differences.
There was a study done very early in the DSD era showing that the EQ was significantly different for DSD vs. Redbook. I do not know how much applies today.
There was a study done very early in the DSD era showing that the EQ was significantly different for DSD vs. Redbook. I do not know how much applies today.