search

LBEM-CH / focus

High Throughput Electron Microscopy Image Processing Software
http://www.focus-em.org
GNU General Public License v2.0
35 stars 14 forks source link

Wiener before CTF shows wrong Thon rings #106

Closed scherers closed 9 years ago

scherers commented 9 years ago

When looking at the simulated Thon rings of the newly added CTF-before-Unbending option I observe this potential aliasing effect as discussed before. Now I processed an image with a high defocus and got the attached CTF in tiles. For me this looks like a bug as the somehow the simulated power-spectrum is 30% too small!? I had a look at the code doing the CTF-correction in tiles and I did not find APIX, which is required for CTF correction. We have a pixel size of 1.34A for the dataset. This could explain the 30% error we get by implicitly setting APIX=1 instead of 1.34. Does somebody agree?

screenshot from 2014-12-08 16 01 54

sthennin commented 9 years ago

Hi, The pixel size is calculated, from the MAGNIFICATION and STEPSIZE-DIGITIZER. This happens in the code in line 289 of 2dx_ctfcor.for. Henning.

sthennin commented 9 years ago

I just realize that the artifact problem is on the left side, which is the simulated CTF. So, this is not from the FFT calculation. I think it is a Moire effect that shows up when the Thon ring density gets close or beyond the density of pixels. The question is, if we shouldn't limit the CTF correction to the area up to which the Thon ring spacing is still higher than the pixel spacing. But in reality, this may not matter.

However, we also saw another problem in the FFTs of the image, where physical (not simulated) Thon rings mirrored back into the FFT from the Nyquist edge. Do you find an example for that?

sthennin commented 9 years ago

the new CTF correction in stripes now works with FFTW and correctly addresses reciprocal space.