Author Topic: Image Integration question: about computing image statistics  (Read 3883 times)

Offline whwang

  • Member
  • *
  • Posts: 62
Hi,

I am using PI to integrate several Nikon D800 images.  During the processing, I found that
most of the time is spend on computing image statistics, and it takes very long.  However, I had
told PI not to do any normalization, not to evaluate noise, not to apply weights, and not to
reject pixels.  In such a case, why it is still necessary to compute image statistics?  I suppose
all PI needs to do is to just read the images and integrate them.  The statistics part really takes
long time.  Is there any way to bypass this step?

Cheers,
Wei-Hao

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4729
Re: Image Integration question: about computing image statistics
« Reply #1 on: 2014 October 09 19:10:28 »
this is one of those things. uncheck "use file cache" and it won't do the statistical computations...

rob

Offline jerryyyyy

  • PixInsight Old Hand
  • ****
  • Posts: 425
    • Astrobin Images
Re: Image Integration question: about computing image statistics
« Reply #2 on: 2014 October 10 09:04:20 »
Welcome to the group!  That makes at least two Takahashi 180 EDs here...

PS  Under the Scripts there is a Benchmark... you may want to check how fast your system is overall. 

PPS  I know those files will be huge. 
« Last Edit: 2014 October 10 11:56:45 by jerryyyyy »
Takahashi 180ED
Astrophysics Mach1
SBIG STT-8300M and Nikon D800
PixInsight Maxim DL 6 CCDComander TheSkyX FocusMax

Offline whwang

  • Member
  • *
  • Posts: 62
Re: Image Integration question: about computing image statistics
« Reply #3 on: 2014 October 10 14:06:57 »
Hi,

Thanks for the answers.  The benchmark seems OK.  I will disable the file cache next
time to see if it improves.

Cheers,
Wei-Hao

Offline jerryyyyy

  • PixInsight Old Hand
  • ****
  • Posts: 425
    • Astrobin Images
Re: Image Integration question: about computing image statistics
« Reply #4 on: 2014 October 11 12:46:36 »
Yes, I bet the problem or slowness is due to the size of the files....  You can also watch what it is doing in the Performance Tab under Task Manager.  It seems to be pretty good at parallel processing.  I am running the SubframeSelector script as I write this and it is using all 8 cores.  Nothing is pegged. 
Takahashi 180ED
Astrophysics Mach1
SBIG STT-8300M and Nikon D800
PixInsight Maxim DL 6 CCDComander TheSkyX FocusMax

Offline whwang

  • Member
  • *
  • Posts: 62
Re: Image Integration question: about computing image statistics
« Reply #5 on: 2014 October 14 17:29:38 »
Hi,

I would like to follow this up.  This time, I disabled the use of file cache, and disabled everything else I can.
(Please see the attached image for my setup.)  PI still spends lot of time on computing image statistics and
MRS noise evaluation.  This is quite annoying.  All I need is a simple stacked image for evaluation.  I really
don't want to spend many minutes on the statistics analyses that I don't need.

Any clue on how to prevent the lengthy statistics analyses?

Cheers,
Wei-Hao

Offline pfile

  • PTeam Member
  • PixInsight Jedi Grand Master
  • ********
  • Posts: 4729
Re: Image Integration question: about computing image statistics
« Reply #6 on: 2014 October 14 17:56:11 »
Wei-Hao - something must have changed in this version of PI/ImageIntegration because that tickbox used to turn off all the statistics computations.

rob

Offline whwang

  • Member
  • *
  • Posts: 62
Re: Image Integration question: about computing image statistics
« Reply #7 on: 2014 October 14 22:39:04 »
 :'(