This dialog allows you to set parameters controlling CTF fitting to FFTs; they all affect the fitting when clicking in an FFT, and some also affect fitting with the Ctffind and Ctfplotter script commands. The dialog can be kept open while getting images and fitting to them. If an FFT is being displayed with a CTF fit drawn on it, then the program will redo the fit whenever you change most values in this dialog.
Use Ctfplotter for: Fit on Click
Select this option to use Ctfplotter instead Ctffind for CTF fitting when clicking in an FFT. Ctfplotter gives ~10-30% less variable results for astigmatism, on average, with typical images, and takes somewhat longer to run. It shows significant advantages with weaker CTF signal and tilted images. If you find that you need to turn off astigmatism fitting with Ctffind, try Ctfplotter.
Use Ctfplotter for: Fits for tuning
Select this option to use Ctfplotter instead Ctffind for CTF fitting when calibrating or correcting astigmatism or when correcting coma. Tests at 200 kV with a cross-line gratig showed that Ctfplotter was ~30% more accurate for coma correction, but no better for astigmatism correction. Ctfplotter may not work correctly in this context with images having high astigmatism (>= 1 micron), but then Ctfplotter may not either.
Use slower search, needed for high astigmatism (Ctffind only)
Select this option to do a slower 2-D initial exhaustive search instead of the default 1-D search with a rotational averaged spectrum. This search is needed if the astigmatism is large (for example, if analyzing images with beam-tilt). Although it should not be needed for ordinary amounts of astigmatism, fitting is apparently more reliable with it on, so the 2-D search is recommended unless you find it too slow. The 2-D search is parallelized so its speed will depend heavily on the number of computer cores available. Including phase in the search will slow it down considerably.
Compute resolution to which Thon rings fit (Ctffind only)
Select this option do the extra computations needed to determine the resolution to which the Thon rings can be fit, or 0 not to. These computations will increase the fitting time by 10-20%, which may be noticeable.
Draw rings to computed resolution (Ctffind) or to end of fitting range (Ctfplotter)
Select this option to draw half-rings on the image either out to the resolution determined by these extra computations if they were done with Ctffind, or to the end of the fitting range used with Ctfplotter, which is based on a local correlation between the power spectrum and fitted CTF curve.
Custom value of highest resolution to fit (Ctffind only)
Enter a limiting value to the maximum resolution to which to fit the amplitude spectrum, in Angstroms. A smaller value fits farther out in frequency space. The minimum allowed value is 3 unless you enter 0 to revert to the default. The line below this text box shows the default value that will be used if the entry here is 0. This default may be set in the SerialEM properties file. If there is no property setting, the default is 5 if the voltage on program startup is above 120 KV, which is the same as the default maximum resolution in Ctffind4, and 10 otherwise. SerialEM will choose a maximum resolution higher than this value when the defocus is high and the minimum resolution is a high number as well. However, with a non-FEG microscope or other situation where Thon rings may never approach this resolution, a higher value may be needed here or in the properties file. If you click in an FFT with the middle mouse button, the status bar will show the resolution ('Period') at that frequency.
The value set here applies when fitting with the Ctffind script command.
Fixed phase shift when not finding phase
Enter a fixed phase shift imposed by a phase plate (a positive value in degrees). This phase shift will be taken into account when circles are drawn on a live FFT for the fixed defocus defined by the Set Defocus for Circles command, when a defocus is reported after left-clicking on an FFT without CTF fitting turned on, and when doing CTF fitting by clicking in the FFT if phase is not being searched for. If a non-zero shift has been entered, the defocus in the status bar will be followed by ': PP'.
Lowest resolution to fit when fitting with phase
Enter a limiting value to the minimum resolution in Angstroms for fitting when phase is non-zero, i.e., when searching for phase or using a fixed phase shift. A smaller value starts the fit farther from the origin in frequency space. This limit applies either when a constant phase shift is assumed or when the phase is being found in a fit with the 'Ctffind' script command. This limit provides a maximum value for the minimum resolution and can be used to exclude the region before the first zero, or even the first zero from fitting, since the spectrum here may be distorted due to a gradual cut-on in the phase shift at low frequencies. However, the minimum resolution may end up being higher than this limit since it is also constrained to be least 1.5 times the maximum resolution used for a particular fit. Enter 0 for no limit. The standard value used by some phase plate experts seems to be 20 A.
The value set here applies when fitting with the Ctffind script command.
Find phase with search limits
Select this option to find the phase shift as well as defocus and astigmatism. An initial exhaustive coarse search for the best fit will be done with a step size of 6 degrees, followed by a conjugate gradient minimization. In the text boxes, set the minimum and maximum phase of the initial search, in degrees. This search will be slower and may be unreliable with weaker data unless the following option is used to fix the astigmatism at 0.
Omit astigmatism when searching for phase
Select this option to fix astigmatism at zero when searching for phase. This will reduce the dimensionality of searches from 4 to 2, increasing the speed and reliability of the fitting.