I found the problem - something got broken re: the filling of parent / child grid information in the move from 2.X to 3.0. I'm checking in a fix shortly. 

On Thu, May 1, 2014 at 12:31 PM, Andrew Myers <atmyers2@gmail.com> wrote:
Hi Folks,

I'm seeing a strange issue with Chombo data in yt-3.0. This is before any recent changes I made refactoring the frontend. Attached are two images of the same slice through the same dataset, one with yt-3.0 and one with yt-2.6. The image looks fine in yt-2.6, but there are clearly artifacts at the level boundaries for yt-3.0. What's strange is that, if I loop over all the grids and dump out the density, I get *exactly* the same answers for 3.0 and 2.6. It's almost like there is some kind of additive blending thing going on for the 3.0 slice plots, though.

I'm a bit at a loss for where to look for this. Has anyone seen something like this before?