›

Multiscale Gradient Correction - Powerful new tool.

Roger Smith
in PixInsight
Hi Adam,
I love this tool as it takes the drudgery our of placing points as in DBE, and still getting bad results. In December I tried using DBE on an image (Flaming Star with Tadpole Nebula). But one strong curved (nasty) gradient went through the Tadpole Nebula. After several attempts the result was not satisfactory. My New Year's gift was MGC, and your video series on it. Thanks! Now using MGC the results are much better, but I am still adjusting the settings.

I discovered that I can do the SpectrometricFluxCalculation on my image, then use StarXterminator, and then run MGC on the starless image. I can see the effect much better without the 12,000+ stars. 

Your new videos are really spot on! I appreciate your showing the PixelMath to help see the background model blurring, and the affects of the MGC controls.
 
I do have some questions:
How are the MARSimages collected/made? Did they have to have any gradient correction?
Are there some actual sky gradients, and the MGC goal is to correct our images to match the MARS images gradients?
Is the following true?: By contrast, DBE is trying to make the all the background to one level. 

Thanks,
     Roger

Comments

  • BlockHeadBlockHead
    Hi Roger,

    1. You will need to look at the official forum for the details of the MARS data.
    2. MARS data is certainly corrected by some method...but as far as we are concerned the MARS data/references are considered free of gradients- and the difference between the MARS data and our own captures gradients in our own images. It is a differencing... not a matching.

    -the Blockhead
  • Roger Smith
    Hi Adam,
    After watching your MGC videos and trying multiple different MGC settings, I could get an improved image, but not really good. 

    I finally settled on first applying MGC with 1024 gradient scale to overall correct my image.
    Then apply MGC again with reduced gradient scale of 256 to fix the more localized bumps remaining in my gradient. For both passes above, I adjusted structure separation and model smoothness to not over correct the desired object.

    Trying to do everything in one pass of MGC would not give an optimal solution for my data. 

    IMO, MGC is far superior to DBE when gradients are within nebulosity, but DBE may be enough simple gradients without a lot of nebulosity. 

    Best regards,
        Roger
  • Dean Wierman
    I am sure glad I looked over here.  The CN challenge this month is Witch Head Nebula and it is pretty challenging here Washington due to being low + my skill level, but I managed to get some subs with the Anitla Triband RGB on a mono camera  (LUM) so far.  I was struggling to get much of the nebulosity and trying different exposures...and then thought I'd check on this forum and this was it.  Way way way better on post wbpp at 120s on the Rasa 8 with the ZWO 2600MM.  - I am not exactly sure where all the gradient was coming from, Rigel and/or the moon, but this process was magic.

    Thanks Adam for all the content,  and Roger for the thread.
  • Dean Wierman
    I have a follow up question.  With the fullish moon in Bortle 6, the Anitla Triband RGB and this new process, it appears to get meaningful data for Luminance.

    With the moon so bright, I am assuming it is a bad time to collect RGB and I should just wait until less moon and hopefully clear skies in the next few weeks.  Does this new process change that logic for RGB?  I have the Baader
    2" CMOS Optimized LRGB Filter Set | High Point Scientific

    Thanks for your thoughts.  It is looking positive for clear skies for the next 7 days and I am trying to sort out if I should focus on other targets and not waste my time on Witch head RGB until there is less moon.

    Dean

  • BlockHeadBlockHead
    No, MultiScale Gradient correction does not mitigate the primary ill-effects of the bright moonlight.
    The increased brightness of the sky will always add (photon) noise to your data.. making it harder to achieve the contrast the dark skies provide. NB imaging in Ha is least affected...but still is affected nonetheless. 
    You would agree with me that you need to wait for the sun to set before taking images... they same is true to a lesser extent with respect to the moon. RGB with the moon up is a bluish noisy mess.
    -the Blockhead
  • Dean Wierman
    thank you.  What do you think about the attached for LUM.  It has been collected in the days leading up to full moon with the triband rgb. It is from 141 subs at 120s on the Rasa 8.  Collection for me here on witchhead is from 1845 to 2230 so the moon was not right in the way.


    image
    witchhead.jpg
    901 x 628 - 58K
  • jeffrey merrilljeffrey merrill
    Hi:  very much appreciate the MSGC..not sure why there's background music on the first intro video..it's very distracting and I was trying to focus on what you were saying..  so one vote/plea here to jettison the music background ..your videos are quite good without it

    jeff
  • BlockHeadBlockHead
    edited January 15
    Hi Dean,

    Your image has significant artifacts. The brighter round thing with the dark top right is not a real feature of the sky. It could be due to either flat field errors or...that crazy moon. 

    -the Blockhead
Sign In or Register to comment.