At the moment, 3dhistog only does the #0 sub-brick (or the one indicated by the -dind option). I suppose that could be modified.

One solution without modification of the program would be to catenate all the sub-bricks together into one big sub-brick, 3dhistog THAT dataset, then delete it as being kind of useless by itself. Kind of klugy, but it works. Here is a csh script to do this on a sample dataset:

#!/bin/csh
set dset = v1_time+orig
@ n = `3dnvals $dset` - 1
set lll = ( $dset'[0]' )
foreach q ( `count -dig 1 1 $n` )
set lll = ( $lll:q $dset'['$q']' )
end
3dZcat -prefix hhh $lll:q
3dhistog -nbin 0 hhh+orig > hhh.out
/bin/rm -f hhh.out

The result is in file hhh.out, which you could plot with something like 1dplot 'hhh.out[1]'

Hope this helps -- it's not very efficient for longish datasets (the sample above has 68 sub-bricks, which was pretty fast, but if you had a 1000 sub-bricks, that would be a horse of a different color, which does not exist and has an infinite number of legs).

-- bob cox