Problem Wiping iTelescope Data

[whixson]

I have about 3.5 hours of LRGB data of M101 from iTelescope T21. When doing a LRGB combine and AutoDev I get what looks like a good image in the center but milky white all around. It is not calibrated with Flats as the calibration files are very old (over 6 months) and including Flats made it worse. When using Wipe, it darkens the corners but leaves a broad area of mottling around M101 and I can't seem to clean it up I uploaded the 4 masters to

https://1drv.ms/u/s!Aod6X19w9TMDgYoAvDP ... g?e=6RONuf

[admin]

Hi,

I have to say this is - on the face of it - really rather disappointing data from the iTelescope service.

In spite of the a very uneven background (some smudges, vertical sensor lines having different biases) I didn't have too much trouble with Wipe cleaning up the background however;

StarTools_130.png (849.93 KiB) Viewed 6112 times

(please note that the automatic AutoDev in Wipe is still meant to show your dataset in the "worst" possible light in a sense, in order to help you see any remaining imperfections or areas that you should keep an eye on. By no means should you feel you need to emulate that same stretch after running Wipe).

AutoDev following turned out OK;

StarTools_131.png (425.77 KiB) Viewed 6112 times

The biggest issue, by far, is the severe walking noise resulting from no bias calibration and (apparently) no dithering. Is this what you mean by mottling (e.g. the diagonal "stripy" look around M101)?

NewComposite.jpg (183 KiB) Viewed 6112 times

The walking noise is a form of correlated/pattern noise. Patterns and non-randomness (in other words "detail") is something that StarTools - in every nook and cranny of the code - is built for to preserve.

Believe it or not, I actually have done some research to try to model this sort of walking noise (a normal person/developer wouldn't even bother - probably rightfully so ), but it is very hard to model (the lines are not really lines but "curve" differently depending on the field of view). So, for now, I'm afraid bias and/or (preferably at least) dithering is the way to go to fix this sort of issue.

Any trouble, questions, let me know!

[whixson]

Wow Ivo that's amazing! WIsh you had stored your results on my server . I too have been disappointed with iTelescope data lately. I used matching dark masters so I didn't think I needed biases, but I can try that. One problem is the calibration files are 6 months old! Dr. Sasse said they are doing new cal frames 'soon' but this isn't the first telescope I've used that had this problem.

I was surprised to see the walking noise as I thought that was a problem associated with CMOS cameras.I do normally dither just forgot this time.

Thanks again! I'll try again

Wayne

[admin]

Wow Ivo that's amazing! WIsh you had stored your results on my server .

It should be super easy to replicate yourself though - it's just a standard workflow with more-tweaks-than-usual (however with a super-charged new Decon module digging out detail - release soon!)

[whixson]

Thanks, I’ll take another run at it! Looking forward to the Decon update. I will say I noticed a tremendous speed increase with the current version (AVX2)!

Wayne

[whixson]

Ok got a lot better results! First I used the Cosmetic Correction Cold Column feature in AstroPixelProcessor Agressive setting there removed a lot of those vertical defects. I also used master biases from iTelecope, and was actually surprised they made a difference even though I used matching darks.

Once in ST Wipe and AutoDev, I played with a combination of different ROIs, Fine Detail settings, and Outside ROI influence settings to get a dark smoother background. I think I ended up with an ROI that included the galaxy but not all the extensions, Fine Detail of 5 pixels, and 10% Outside ROI influence before going on to contrast etc.

Thanks for the encouragement in pursuing solutions to processing challenging data. I’m sure there’s more I could have done, but this was a big improvement.

Wayne

[admin]

Great to hear!
The more challenging/compromised the data, the more tweaks are necessary. It's such an unfortunate fact of life, as it's not a problem for veterans like you and me, but it's the one thing that trips up newbies - even if they do everything "right" (in terms of processing sequence, calibration, etc.).

Great to hear you're getting along with 1.6! As a side note, if you have just started using 1.6 and do narrowband composites as well (I seem to remember you do/did?), you may be interested to know that the Color module now has a ton of special presets and matrix options. E.g. in the Compose module, just load your S, H, O datasets as R, G and B. Set the exposure times and use "L + Synthetic L From RGB, RGB" for "Luminance, Color". Now process as normal (luminance and color are separated, so you'll be processing in mono mostly until you hit the Color module).

Then, once you hit the Color module towards the end of your processing, you can choose a great number of popular color renderings and blends which StarTools then combines with your luminance on the fly. E.g. this is all done without impacting the detail/luminance portion. So you can decide at the last moment, far into your processing flow, if you want SHO mapped to OHS, HOS or "20% SII + 80%Ha for R, 100% OIII for G, 85% OIII+ 15%Ha for B", etc. You don't have to re-import and re-process or re-mix. Tracking will know how to get back to the original coloring, remap the channels, and forward propagate the result again for just the coloring. It's all seamless.

[whixson]

That’s great to know! I was re-combining and starting from scratch. Yes I do quite a bit of narrowband, so will explore that more. I also just got a new OSC camera that I will be using with a dual or Tri-Band filter and either extract the narrowband components for combine or just process as a color image.

PS is there a way to follow a parallel path and end up with different versions?

[admin]

That’s great to know! I was re-combining and starting from scratch. Yes I do quite a bit of narrowband, so will explore that more. I also just got a new OSC camera that I will be using with a dual or Tri-Band filter and either extract the narrowband components for combine or just process as a color image.

PS is there a way to follow a parallel path and end up with different versions?

Excellent. My memory doesn't suck then

To do different versions very quickly/easily, it's as simple as launching the Color module at the end of your processing, choosing your first rendition/color scheme, 'Keep'ing that, switching Tracking off, doing noise reduction, saving this rendition, then pressing Undo, launch the Color module, etc.

As has always been the case, re-launching the Color module - as long as Tracking is on - will completely restart the Color calibration/tuning process.