AutoDev and Wipe trouble

[weirdsong1]

Hello,

I am new to StarTools and excited by the possibilities of the image processing power. However, after several attempts I am running into the same problem with stacked image I open in StarTools. I've done my best to follow the tutorials on the website and some of the tips here in the forum, but am clearly missing something. I've done my stacking both in DSS and Siril (following the best practice guidelines as I understand them) and keep running to into problems from the start.

First, (using v1.6.4 currently) when opening the file with Linear from DLSR..., the file opens in grey scale (not sure why). In plain linear, AutoDev will produce the heavily greencast stretched image that I would expect in other processing programs, and then I move to optional binning (I've tried ever possible setting), crop, and then Wipe. At Wipe, things go haywire.

Example of Pinwheel Galaxy (6 hours total exposure, 120" subs, ISO 400, stacked with darks, flats and biases):

Star Tools wipe example.jpg (398.3 KiB) Viewed 4105 times

I've spent several hours messing around with settings and tried about a dozen different stacked files all with similar results. My instinct is that StarTools is picking up excessive noise (although I do not see this in other software) and so I've spent time messing with the "ignore fine detail" settings in AutoDev, but have not improved the results.

The image was shot on an unmodified Canon T5i, William Optics ZS71, and AVX.

I very much appreciate any help the community may be able to provide.

Thanks.

[almcl]

Ivo will likely be along in a while with good advice and insights, but in the meantime, could you post a link to your stacked but otherwise unprocessed image in say, Dropbox, Google drive or whatever?

I think your image needs the stacking artefacts cropping out and what are possibly dust bunnies masking out, before the wipe module.

There are other techniques for dealing with noise (my Canon T5i has given me quite a bit of practice ) in addition to 'ignore fine detail' but which works best does seem to depend on the image.

It may also be worth upgrading to the latest edition (http://download.startools.org/StarTools ... _alpha.zip) which is quite stable and has quite a few new features.

[weirdsong1]

Ivo will likely be along in a while with good advice and insights, but in the meantime, could you post a link to your stacked but otherwise unprocessed image in say, Dropbox, Google drive or whatever?

I think your image needs the stacking artefacts cropping out and what are possibly dust bunnies masking out, before the wipe module.

There are other techniques for dealing with noise (my Canon T5i has given me quite a bit of practice ) in addition to 'ignore fine detail' but which works best does seem to depend on the image.

It may also be worth upgrading to the latest edition (http://download.startools.org/StarTools ... _alpha.zip) which is quite stable and has quite a few new features.

I hope there truly is a simple fix. I've tried cropping, before Wipe, before AutoDev, and it doesn't seem to make any difference. I realize the importance of the so-called 'virgin' stacked images, which I've done my best to adhere to. I'm definitely befuddled, but hopeful as StarTools looks like a very powerful tool.

Here is the dropbox link with 2 files, one stacked with DSS, the other with Siril. Personally, I find the Siril stack better. https://www.dropbox.com/sh/20e7wmvuaccr ... OOZua?dl=0

Appreciate everyone taking the time.

Edit: This particular image does have a smudge on the center right of the image (from sap that fell on lens somehow). I realize that may cause problems, but that smudge is unique to this stack and all of the dozen or so images I've tried to process have the same wipe/autodev issues, smudge or not.

[admin]

Hi,

It appears your flats are not working at all (or are making things worse!). A well calibrated stack should not show such an uneven background, nor should it still be showing many dust donuts like this. Figuring out what is going wrong here should be you #1 priority.

This is really a pre-processing/calibration/acquisition issue, but if you really need to address this in post-processing, then you can usually get the Wipe module to mitigate some of the gradients and uneven lighting resulting form not using flats. Success will greatly depend on severity, how fast these gradients undulate and the presence of further issues (e.g. dust).

Your #2 priority should be dithering - if at all practical - as the dataset is showing some severe walking noise (streaks). Note that the latest 1.7 alpha introduces dedicated walking noise reduction if you cannot dither.

If you imagine that the artificially introduced uneven lighting and walking noise is detail, AutoDev is doing its job very well, preserving this faint nebulosity. Of course, in the case of this image (a smaller object-of-interest on an otherwise "empty" uninteresting background), the general advice is indeed to use a Region of Interest. For example like so;

StarTools_186.png (631.58 KiB) Viewed 4081 times

Hope this helps!

[almcl]

Following on, I had a go at the data in the DSS file. Rather than mask out the dust, I chose to crop as this also helped removed the coma affected stars. The red channel seems a bit low compared to blue and green but here's the result:

Pinwheel DSS.jpg (189.27 KiB) Viewed 4077 times

[weirdsong1]

Thank you Ivo and almcl for your feedback! I really appreciate you taking the time. Other issues aside, it's at least a little comforting to see that you are getting some processed result as opposed to the nonsense I was getting.

To some of your observations, obviously, I'm not using a guiding system right now which I'm guessing leads to the issues of the walking noise. I used to recenter the target every 30 minutes or so, to help avoid this drifting, but I really hadn't noticed any difference in my final images, so I abandoned it. Perhaps I will renew this practice until I add autoguiding. As to dithering, I know I made the rookie mistake of believing that the gradual creep of the tracking would be sufficient (I am aware that "in random directions" is a key to dithering), but I'm shocked to the extent it's creating a noise problem. I'm aware that APT can send dithering commands to my AVX mount, but am unclear on how this might work. There's a micro usb receptacle on the handcontrol which I believe I've read can be used to control the AVX from the computer. Would appreciated if anyone here has advice?

RE flats, I must admit that I've always been fuzzy on what exactly makes a good flat. I reacquire flats about every 2-4 weeks, mostly just to see the results (I've read people who only retake everything 6 months or so, and those who insist they are taken after every session). Mostly my practice has been to put the T5i on Av mode and shine a bright white screen from my laptop at the telescope (covered by a white tshirt). As long as the histogram is in the 50% range, I keep them.

Thanks again. I'll investigate the above issues, and look forward to putting together a stack that will truly test the power of StarTools.

[almcl]

I am no expert on flats (there's a thread where Ivo gently and kindly pointed out some of the problems with mine) but:

You can reuse flats to correct for vignetting and, if the dust doesn't move and the camera to scope connection doesn't get disturbed, you can also get away with reusing them for dust correction. But with a DSLR which gets taken off the scope every time this is unlikely to be successful for very long.

There's quite a good item on how to see DSLR sensor dust here https://photographylife.com/how-to-spot ... 20on%20it.
might be worth a go?

The other thing is the need for even illumination of the flat panel. If using a computer screen and a fast shutter speed, the screen refresh rate can actually create gradients in the flats that won't be there in the lights.

Another trap for Canon DSLR users is not covering the view finder with the little rubber cover from the neck strap. Although I use one, it does sometimes get displaced.

Something which may help with the lights is a good, long dew shield, especially in suburban, light polluted settings.