can I render out the zbuffer image that poser uses for the depth of field, so I can use it as a mask in photoshop and do the blurring myself?

I don't think there's a direct way to get it.
One trick you might try to get around Poser's rather jagged depth blur:
add a couple of cyclorama or skydome props located at appropriate distances,
each nearly transparent.   The farthest parts of your background would be
seen through two cycloramas, the middle parts would be seen through one.

My python page
My ShareCG freebies

I was thinking about it. Black background with volumetric light directly at camera  position pointing at subject.Light fades @ farthest object. what do you think?

wallm, ya know that the depth cue function produces a render with an alpha channel?
maybe do that, and invert it in photoshop or something, to get DOF fx.

I seem to recall couple of threads with depth of field discussion:
http://www.renderosity.com/mod/forumpro/showthread.php?message_id=2998822&ebot_calc_page#message_2998822

Hi, my namez: "NO, Bad Kitteh, NO!"  Whaz yurs?
BadKittehCo Store  BadKittehCo Freebies and product support

The biggest problem you'll run into is that you only have 256 levels of grey to play with.
For a scene with a lot of depth this leads to poor results.

Poser doesn't use a z buffer image for depth of field. If you want to refine the DoF effect, you can increase the Pixel samples value on the FireFly tab of the Render Settings dialog (please increase the value carefully though, it increase the actual number of samples taken exponentially; 6-8 should do).

Attached Link: http://www.stewreo.de/poser/dof_tutorial.html

Stewer has a method for faking a depth render - at the link. 

Attached Link: Using DepthCue for Depth of Focus

old school DOF technique from '99

Try using Gaussian blur in your paint program, changing the value at the right increments. I use this method in use with the clone tool from a copied image onto the original.