Masking glycans for 3D variability analysis

I am very interested in running 3D variability analysis on my dataset. I have read the documentation in the guide and ran ab initio with K=1 followed by homogenous refine to create a consensus refinement of 405000 particles. Heterogenous refine results on this particle stack suggested different discrete conformations (opening and closing of the active site and hinging at the linker between two domains) but I want to test 3DVA to see if some discrete conformations may have been missed and also visualize continuous flexibility.

The protein is small (~140kDa) and heavily glycosylated (~10 N-linked glycans) and I am worried that 3DVA would mostly capture glycan flexibility. Since the glycans would move when the underlying protein moves I don’t know how to mask them out for 3DVA. Also, the guide recommends dilating the mask to ensure that the entire protein can be contained in all different conformations but doing so would also incorporate more glycan density.
Could anyone please provide me with some guidance? I would greatly appreciate it!

Further to my earlier message - if I am expecting large variation such as active site closure within a domain or motion at a hinge point between domains, would this generally be more variable than variability within glycan structures so that the former would be detected in the first few variable components? Perhaps large-scale glycan variability would only arise when the underlying protein around that N-linked asparagine undergoes conformational change?