I selected 5 ISP images (same field, different bands) to calculate the zeropoint.

These were the 5 images, all 5 second exposures (typical for ISP):

isp/20100225/o5252i0361o03/o5252i0361o03.chip01.fits

isp/20100225/o5252i0361o04/o5252i0361o04.chip01.fits

isp/20100225/o5252i0361o05/o5252i0361o05.chip01.fits

isp/20100225/o5252i0361o06/o5252i0361o06.chip01.fits

isp/20100225/o5252i0361o07/o5252i0361o07.chip01.fits

I then used dvo, and the bright star catalog (catdir.synth.bright), to find some bright stars that I could match (by eye) to the images. I found 10 stars (the same 10) in g, i, and r band, only 9 (in z), and only 3 in y (I found 5 more to get a better estimate).

psPhot was used to get instrument mags for these stars, and the zero point was found by taking the median of (catmag - instmag - 2.5log10(exp_time))

These are the ISP zeropoints I calculated:

band	ZP	stddev
g	19.3	0.3
r	19.1	0.3
i	18.4	0.2
z	17.2	0.1
y	14.0	0.8

Here are plots of ZP vs instrument magnitude, for the various bands: