Some software enables overlay of a batch of images so that you simulate the
classical crude phi slicing and get visible patterns. I think we tried with
d*Trek.

Jan


On Tue, Apr 28, 2015 at 9:20 AM, Graeme Winter <[email protected]>
wrote:

> Looking at PAD images is something we have had to get used to at Diamond,
> and sometimes it takes some tweaking to get a really good idea of what the
> images actually look like. This is a challenge if you measure the data
> properly with fine slicing & low dose...
>
> One thing which really helps is to have the positive image rather than the
> negative i.e. white spots on a dark background. The eye is much better at
> picking these out. There are also tweaks which can be done within image
> viewers to make spots more visible for example showing the maximum rather
> than average of an n x n zoomed out image. ADXV white on black background +
> 100% scale generally works well. Summing the images using e.g. XDS
> merge2cbf or dials.merge_cbf can make them easier to interpret,
> particularly if you're used to looking at 1 degree CCD images.
>
> Within dials (http://dials.sf.net) we have an image viewer,
> dials.image_viewer, which also has the option to calculate local mean,
> variance, dispersion and so on so you can see "what the spot finding sees"
> - this can make a massive difference. For example:
>
> Default image
>
>
> http://www.ccp4.ac.uk/xia/screenshots/Screen%20Shot%202015-04-28%20at%2008.01.34.png
>
> Tweaked to "inverse" i.e. right way around, brightness turned down:
>
>
> http://www.ccp4.ac.uk/xia/screenshots/Screen%20Shot%202015-04-28%20at%2008.01.52.png
>
> - much better - but then the dispersion map thresholded a la spot finding:
>
>
> http://www.ccp4.ac.uk/xia/screenshots/Screen%20Shot%202015-04-28%20at%2008.02.06.png
>
> This really shows the images well... also allows you to tweak the spot
> finding settings nicely. This is a particular problem with good crystals
> from a good beamline as the spots can often be 1 or 2 pixels with many many
> counts on a background of very small (often 0) counts.
>
> A final challenge with these detectors is that they *actually count
> photons* i.e. the statistics are the same as counting cars going past the
> school* - where on a CCD if you can *see* a spot there is a spot there, as
> it's climbed out through the readout noise etc, on a PAD you need to look
> at the numbers and by and large put more faith in the processing software
> than your eyes, as people are great at finding patterns where no patterns
> are there to be found.
>
> Flip side is yes, the images are harder to interpret but these detectors
> do a nice job of actually recording diffraction!
>
> best wishes Graeme
>
>
>
> -------------------------------------
> *other elementary examples exist.
>
>
>
> On Tue, Apr 28, 2015 at 3:08 AM Jens Kaiser <[email protected]> wrote:
>
>> SSRL Bluice opens the image in adxv upon double click in the diffraction
>> window.
>>
>> HTH,
>>
>> Jens
>>
>> On Mon, 2015-04-27 at 16:57 -0700, Bernhard Rupp (Hofkristallrat a.D.)
>> wrote:
>> > Thanks - particularly great if we had these images/option available to
>> look
>> > at
>> > in real time during data collection, w/o first having to download the
>> raw
>> > data (not
>> > really feasible during remote data collection). I don't think the ESRF
>> > online data base has the option, but other beam lines may?
>> >
>> > Thx, BR
>> >
>> > -----Original Message-----
>> > From: James Holton [mailto:[email protected]]
>> > Sent: Monday, April 27, 2015 4:05 PM
>> > To: [email protected]; [email protected]
>> > Subject: Re: [ccp4bb] PAD images
>> >
>> >
>> > In the ADXV viewer:
>> > http://www.scripps.edu/tainer/arvai/adxv.html
>> >
>> > Go to Edit:Settings and click on the "Small Spots" radio button. This
>> solves
>> > most of the "I can't interpret the spots" problems you describe.
>> >
>> > -James Holton
>> > MAD Scientist
>> >
>> > On 4/27/2015 3:31 PM, Bernhard Rupp (Hofkristallrat a.D.) wrote:
>> > > Hi Fellows,
>> > >
>> > > I wonder whether it's just me and my eyesight failing (or excessive
>> > > internal lubrication)....
>> > >
>> > > It seems that the art of looking at diffraction patterns and being
>> > > able to tell a lot about modulation, superstructures, extinctions,
>> > > etc. becomes kind of useless old fart stuff when dealing with PAD
>> > > images.  I can't for my life see interpretable patterns on frames
>> > > where the beamline autoprocessing delivers actual data sets. The
>> > > absence of a point spread function etc that gave interpretable
>> > > film-like images on IPs or CCDs, seems to be the reason.
>> > >
>> > > A PAD pixel with 1000000 counts looks like one with 100 when viewed
>> > > with the low dynamic range of the displays compared to the huge
>> > > dynamic range of the detector.
>> > >
>> > > Is there somewhere in the process a humanly unusable composite image
>> > > with a point spread that allows visual pre-processing, inspection, and
>> > > interpretation despite a low dynamic display range?
>> > >
>> > > Looking at the hklview or similar after processing is pointless (no
>> > > pun intended), because the stuff I might be interested in is already
>> > > processed away.
>> > >
>> > > Some humanly interpretable raw data images would be quite useful...
>> > >
>> > > Best regards,  BR
>> > > -----------------------------------------------------------------
>> > > Bernhard Rupp
>> > > 001 (925) 209-7429
>> > > +43 (676) 571-0536
>> > > [email protected]
>> > > http://www.ruppweb.org/
>> > > -----------------------------------------------------------------
>> > > The man who follows the crowd will get no further than the crowd.
>> > > The man who walks alone will find himself in places where no one has
>> > > been before.
>> > > -----------------------------------------------------------------
>>
>


-- 
Jan Dohnalek, Ph.D
Institute of Biotechnology
Institute of Macromolecular Chemistry
Academy of Sciences of the Czech Republic
Heyrovskeho nam. 2
16206 Praha 6
Czech Republic

Tel: +420 296 809 340
       +420 226 201 571
Fax: +420 296 809 410

Reply via email to