The translation applied when transitioning from downsampled to full resolution montages is off -- this is more pronounced when looking at montages after parameter sweeps where different translation factors can shift the original montages to distinct positions in the volume.