The second image is interesting, you are certainly getting lots of 'relief' detail on the subject, but you have the typical problem of (very in this case) uneven field illumination which is a characteristic of mismatched DIC combinations and oblique illumination. As Pau says it's not clear if this is true DIC or an extreme example of the form of oblique illumination that Graham Matthews called VOILA that Pau references.
Yes the gradient is pronounced in these images. Its definitely caused by the dic prisms but its greatly increased in post processing by using contrast, curves or levels adjustment. So far the system is producing 3d relief and optical sectioning. I still have to work on increasing contrast... After some research I believe this is a copepod. Does anyone know?
Edit: Its a quick fix in post to correct the gradient. Here are the updated versions