With large (and fairly sparse) particles such as RyRs, I often find that I have strong Thon rings at low to medium resolution (presumambly from the protein), and then nice Thon rings at ~3-4 Å (from the solvent). This is particularly evident when processing using movie frames as input. (see example attached).
CTFFIND does a pretty good job of fitting both, but I wonder in such a case whether there might be some benefit in using discontinous resolution ranges for the CTF fit - e.g. 20-6Å for the protein, and 4-3 Å for the solvent?