Overfitting after Homogeneous Refinement of Membrane-bound protein


I am processing a Cryo-EM dataset of a membrane-bound protein and am fairly new to cryoSPARC. I am getting good alignment in my 2D classes using a box size of 176 pixels, where the box size is about 20 A larger than the particle diamter.

Ab-initio and refinement ran properly, and I am getting a map of about 4.4 A resolution.

In order to improve the resolution, I tried increasing the box size from 176 pixels to 216. The 2D classes now incorporate more of the membrane, and upon homogeneous refinement, I obtained a map of 4.08 A.

Upon inspection, I found symptoms of overfitting, with the entire structure having spike-like artefacts.

The dynamic mask is perfect as per my previous reconstruction at 176 pixel, so I am unable to figure out what is going wrong. I would appreciate any help or experience shared regarding this problem.