Skip to Content

confuse about dose-filter and restore noise power

2 replies [Last post]
Joined: 30 Aug 2017

Hi all:

I am now trying unblur_openmp_7_17_15.exe and sum_movie_openmp_7_17_15.exe for motion correction. After process my image movie stack (test_movie.mrc as input),I can get four sum images:

#1:sum image without dose-filter from unblur: test_movie_ubl_sum.mrc

#2:sum image with dose-filter from unblur: test_movie_ubl_DF.mrc

#3:dose-filter image from sum_movie: test_movie_ubl_sum_DF.mrc

#4:restore-noise-power image from sum_movie after dose-filter: test_movie_ubl_sum_RNP.mrc

my understanding is that #1 was used for CTF estimation and #3 was used for particle picking and classification due to high contrast after dose filtering.I extract selected particles from #4 images for final refinement.

Then, I compared the FFT from the #2 and #4 and find nothing different, so my question is what is the difference between #2 and #4 ?

Thanks in advance,


Mengjie Liu

Joined: 16 Jul 2010
Hi, There is no difference


There is no difference between 2 and 4. In your unblur examples you are always restoring noise power (this is the default, you can however set it to no in the expert options). Sum movie doesn't do anything that unblur doesn't do, it just doesn't repeat the alignment, and therefore can save you time. This means, that there is no point running unblur more than once, just run it once, then run sum movie to generate the other states that you want.

E.g in your scheme, you could do the following :-

#1 - run unblur, no dose filter. Use this for CTF estimation
#2 - run sum_movie (give shifts from #1), dose filter, no noise restore - Use this for particle picking.
#3 - run sum_movie (give shifts from #1), dose filter, noise restore - use this for processing.



Joined: 30 Aug 2017
Thanks Tim

Hi Tim,

Thanks for you reply.I will not do dose filter in Unblur except alignment for next case for saving my computer resource and time.

It seems unblur work better than other motioncorr software for my case this time. Thanks again.