Newbie Question - PixInsight Calibration

Hi,
not sure if this is the discussion area mentioned in the fasttrack training videos. In any case, I've completed all lessons up to and including Image Integration. Although I'm not on the latest PI-version, I was able to apply the parameters as shown and create the final files for integration. I have one question or problem though. During calibration, PI converts my OSC images, taken with a Nikon D850, to grayscale. Why? I am able to complete the process but the final image has lost a lot of colors that are contained in the original RAW-files. I don't have this issue with the images downloaded for the course, but very much with my own images. DSS gives me much better colors - and it's free! Not sure what to do about it in PI. Any help appreciated.

thanks
Matt

Comments

  • edited October 2021
    Hi Matt,

    Hmm.. a couple of things...

    1. PixInsight does *not* convert your OSC images to grayscale during calibration. Not if you load the images as PURE RAW (as shown in FastTrack Training). The images should be grayscale to begin with... and *then* after calibration the images are debayered showing color. All of this happens automatically in WBPP assuming CFA is chosen as well as the correct matrix pattern. The images from your camera are natively grayscale images! When you see a color picture at the back of your camera's screen or on your computer- it has been debayered for you.

    2. Regarding DSS- I would suggest not making comparative statements such as "DSS gives me much better colors and free..." ... it simply isn't useful. DSS and other programs, to my knowledge, do not offer as large a parameter space to make adjustments. You may indeed get output from DSS without having to configure as much as you do in PI and see something like "better colors." But, and this is important, this does not mean that the DSS output is optimal. Once configured (with the right parameters), Pixinsight will give you comparable and, and in many cases that matter, optimal results once it is configured properly. 

    Let me be clear.  You could say that EXACT same thing about my videos. It is an easy argument to say that "I can learn image processing/PixInsight much easier with YouTube videos... - and they are free!" compared to the effort you need to do to watch the many instructional videos I create. The value I am striving to impart is an understanding of the processes themselves as well as problem solving tactics, creative techniques... etc. This is why I made FastTrack Training. It gives people a *real* entry into the world/style of my instruction. It is verbose and longer than most- but it comes with benefits that I think are lacking in other "comparable" offerings.

    -the Blockhead
  • thanks Adam for the quick reply. I fully agree with your comments about the videos, that's why I signed up here; being fully aware that if things are not fully understood and/or done wrongly at the beginning they are hard or impossible to rectify later on.

    As to my problems: I did press PURE RAW in the format explorer window as instructed. I also created the COSMETIC CORRECTION window as instructed, then started WBPP and loaded my lights, darks and bias (I don't have any flats, but since I was using a 500mm Nikon lens, not a telescope, I will crop a lot eventually so vignetting is not a big issue). Everything proceeds ok and eventually it creates 120 files with the _c_cc_d postfix in the debayered subfolder. But AFTER the files have been written, and that is strange, it continues and the log shows that it does this for every debayered image:

    Computing descriptors for image 120 of 120

    G:/Travel Photos/2021/New Brunswick/639ND850 - NGC7635/PI/calibrated/light/debayered/_DSC6997_c_cc_d.xisf

    Loading image: w=7104 h=4728 n=3 RGB Float32

    11 image properties

    43 FITS keyword(s) extracted.


    ConvertToGrayscale: Processing view: Image01

    In-place color space conversion: RGB -> Grayscale: done

    150.621 ms


    StarDetector: Processing view: Image01

    19833 star(s) found


    DynamicPSF: Global context

    1 view(s)

    19833 star(s)

    19662 PSF fittings

    6.232 s

    ------------------------

    FWHM : 2.253 px

    Eccentricity : 0.716

    SNR : 5.078

    ------------------------


    My question - is this supposed to happen (including grayscaling) or is there an issue with my setup? And another question - is the PROCESS ImageCalibration (rather than the script) good for anything? It seems it does not do any debayering or allows to load calibration files.

    Thanks and much appreciated!!
    Matt

  • This appears to be the part of the post processing sequence that looks at images to determine a reference for StarAlignment or weighting. This grayscale conversion is only done in memory- and does not affect the debayered images. 

    Are your debayered images in color as expected? 
    How about the registered images? (I am assuming you selected tihs...)
    And finally the integrated image.. what does it look like.


    I think some screenshots would be helpful. You need to show the problem and examples of the things it cannot be. There isn't anything in WBPP that converts from a color image to a grayscale one in terms of output written to disk. Some of the screenshots should show the images zoomed in 100% or more.

    For the bulk of pre-processing typically done there is no significant difference/benefit to manually calibrating. 

    -the Blockhead


  • Could be the quality of the images themselves of course, basically I just want to be sure that I'm on the right track. 2 screenshots are attached, the first one the raw file, showing a bit of pink-purple around M16, the second one is the debayered version of the same image. Mosaic pattern was set to AUTO, RGGB gives me similar results, but the others don't work at all. I followed your videos for the rest of the settings.

    Thanks again for all your help.

    4595RAW.jpg
    623 x 830 - 84K
    4595debayered.jpg
    623 x 830 - 71K
  • Yes, the "green" image looks correct. This is what an uncorrected color image (from OSC) will look like. If you display with an unlinked STF... you will likely have what you intend. Unfortunately your stars are out of focus...but that is a different issue. 

    -theBlockhead

Sign In or Register to comment.