LED flats

Discussion in 'General' started by David23, Jun 26, 2019.

  1. David23

    David23 Standard User

    Joined:
    Jan 9, 2018
    Messages:
    7
    In the MaxIm DL manual, page 138 acquiring Flat-Fields it states that to calibrate the camera do not use fluorescent or LED light. If you are acquiring a flat-field with an optical tube and camera, do the same restrictions apply. Online there appear to be many devices for taking flat-fields that use LED lights.
     
  2. Doug

    Doug Staff Member

    Joined:
    Sep 25, 2014
    Messages:
    6,059
    The issue with LED and fluorescent flats is the spectral response. Fluorescent tubes generate a series of emission lines at discrete wavelengths rather than a continuous emission like an incandescent bulb. White LEDs are usually built using a blue or UV LED with a fluorescent phosphor that converts the blue light to a wide band of spectral lines. This is worse because it has the fluorescent spectrum and a huge bright peak in the blue or near UV.

    The problem is that these are a poor match for the illumination spectrum you will see from the night sky. All sensors have varying spectral sensitivity, and their flat-field characteristics vary with wavelength.

    This is more important for some sensors than for others. I especially wouldn't recommend it with a backside illuminated sensor. Unfortunately the only sure way to find out if it will work for your instrument is to try it. This is why we don't recommend it. It may or may not work okay... you just have to try it.
     
  3. David23

    David23 Standard User

    Joined:
    Jan 9, 2018
    Messages:
    7
    I am using a DSLR camera, and using a LED panel appears to be working. With your comments in mind, I will do more testing.

    Thanks
     
  4. Colin Haig

    Colin Haig Staff Member

    Joined:
    Oct 27, 2014
    Messages:
    2,365
    David, there's two kinds of "calibration" at work here.
    You can use an LED-based panel for flats (I have a Spike-a-flat unit), and "calibrate" the dust donuts and vignetting out of your images.
    However, it is very difficult to "calibrate" colour balance as photographers think of it with most LED panels, due to the reasons Doug mentions.
     
  5. Doug

    Doug Staff Member

    Joined:
    Sep 25, 2014
    Messages:
    6,059
    It's not just that though. All sensors have variation in sensitivity over the spectral response; for example, most sensors are less sensitive in blue. However there is some pixel-to-pixel variation in spectral response. This can be exacerbated by strong spectral lines. It's not something that is specified in the sensor datasheet, so it is difficult to know which sensors have this issue without trying it. I think this issue may be less common with modern sensors than it used to be.

    Back-illuminated CCD sensors can have another issue: internal reflections (within the chip) causing "fringing", which is basically Newton Rings - same thing you see when you spill a little oil on a wet driveway. Tiny thickness differences across the sensor will cause them to appear in the presence of a strong spectral line. Newton Rings will show up if you have spectral lines; in fact, the night sky has some lines (both natural sky glow and of course light pollution) that can cause the rings and they're very hard to deal with. The only thing that works at all well in that case is to use sky flats.
     
  6. Doug

    Doug Staff Member

    Joined:
    Sep 25, 2014
    Messages:
    6,059
    Here's an image showing fringing in a back-illuminated sensor:

    [​IMG]
     
  7. David23

    David23 Standard User

    Joined:
    Jan 9, 2018
    Messages:
    7
    Currently, I am trying to astrometry with the DSLR camera, and the LED panel appears to be working fine. I know that I am not using the software to its full potential, but I would like to know what type of activities would require the use of calibrated colour balanced flats.

    Thanks
     
  8. Colin Haig

    Colin Haig Staff Member

    Joined:
    Oct 27, 2014
    Messages:
    2,365
    Sure, astronomical imaging is a pretty broad field.
    For astrometry or occultations, colour balance isn't as important if you are measuring positions or timing events.
    Knowing the behaviour of the different RGGB sensitivities can impact light curves for occultations a bit.

    For variable star work and photometry, where you're measuring brightness, sensor behaviour against calibrated (known) stars, is critically important.
    Then, if you move on to astro-photography, it gets very important for esthetic reasons (eg does it look right).

    Here's a nice example where Brian, a DSLR user has done a really good job colour balancing their image:
    https://forum.diffractionlimited.com/threads/m-17.4088/#post-23106
    Stars around M17 are golden-yellow, the nebula itself is a nice pale red.
    This user has basic equipment, and there are some issues with distortion in the corners at the edge of the field, but the centre of the field is really well done.

    With DSLR's you've got a RGGB Bayer Colour matrix, versus a monochrome sensor with colour filters, where the flats help with dust donuts on the individual filters as well as the scope and CCD camera.

    By way of comparison:
    Here's one of Robert's early DSLR images when he was just getting started (he's progressed quite a way since):
    https://forum.diffractionlimited.co...60mmfl-95xpwrbaaderfringekillersum1-jpg.4403/
    Here's a pro image of the same object done by Tony Hallas with a monochrome CCD and filters: ( Tony is a guru of imaging):
    https://forum.diffractionlimited.com/threads/the-whale-by-tony-hallas.5971/

    Here's another example:
    https://forum.diffractionlimited.co...atory-in-colorados-mountains.5892/#post-31216

    Carry on - you're doing good stuff.
     
  9. Brian Atteridge

    Brian Atteridge Cyanogen Customer

    Joined:
    Jan 23, 2017
    Messages:
    23
    Location:
    Lewisville, TX
    Just read this post and wanted to throw a few notes in about my image that might help you. I use MaximDL when I do color balancing, but I do it manually. I imagine that I could try to do it automatically, but I have found it is easy to do manually and doesn't take a lot of time. My DSLR camera has had the IR filter removed and I use an astrophotography IR filter when shooting so the images I take are almost always skewed in the red spectrum. What I do is use the leveling tool to balance it out by adjusting the histogram to make the red, green and blue spectrum align. I use a manual adjustment of the slider for each color separately to bring them into alignment so that once done the three histograms look almost identical and align closely. I am uploading an image (Example.jpg) that shows you what it looks like before I do this. You'll notice on this raw image the red is slightly to the right of the rest and the image range is really limited so the total image is very red. If you bring the left Red indicator to the left edge of the histogram curve and the right one far enough to the right to get most of the data (too far right and it under exposes) then slide the middle up the inner edge of the histogram like I should in the second image I uploaded (Example2.jpg) you will see the resulting imager is now more normal. Do this for each color individually and if it doesn't look exactly right on the first try, apply the results and do it again till it does. This is where some of the automatic functions in MaximDL might do this quicker but i haven't tried any of them yet. Of course, you still need to process flats, but I usually do this first then do the leveling. Otherwise it doesn’t look as good. My flat box is a multi-LED one with several sheets of paper covering it and using the LED's to back-light it and is then slipped over the optical tube and clamped in place. I made it myself. Frankly it does work pretty good simply for photographs, however for doing the photometry it might not work as good since it is LED based.

    And Colin, Thank you for the compliment! :) I am actually proud of that one too thought i wish i had more images of it to tease out more details. Plan to get some next chance i have.
     

    Attached Files:

Share This Page