Wipe - Not sure what it's supposed to "look" like when the field is even

[ionia23]

Hello all.

I've been having quite a journey learning the ins and outs of Startools. It's required unlearning everything I knew from the Photoshop world, but that's a good thing. It's forced me to go back to the beginning to really understand what raw data "means". That has helped quite a bit.

I'm not brave enough yet to post any of my works-in-progress, but I do recognize where one of my own stumbling blocks is, and it is the Wipe module. The various tutorials do a quite good job of explaining what it does, how it does it, and what it's for. What I'm struggling with is "what it is supposed to look like when I've achieved a decent flat field"? This is a little difficult to recognize when that powerful diagnostic stretch is turned on.

I supposed that's what I'd like to see: what an actual good, even background looks like while under the diagnostic stretch in Wipe. Maybe a before and after, but still in the stretch mode. Even better if it's an image that started with multiple problems. I may be seriously overdoing it with wipe rendering my images "chunky" for the rest of the session.

Thanks everyone.

[admin]

Hi,

I'm using this tutorial dataset;

https://download.startools.org/Tutorial ... tput1.tiff
(it was already color balanced, so I open it with the 1st option)

1st AutoDev, showing all the issues;

StarTools_2758.jpg (232.41 KiB) Viewed 3971 times

After Bin (50%), and cropping away stacking artifacts, I launch the Wipe module. Defaults come up as this;

StarTools_2759.jpg (339.02 KiB) Viewed 3971 times

While this may seem horrible, this is already close-to enough for further processing. For example this is 'keep'ing that result, and doing a subsequent AutoDev with RoI over M42;

StarTools_2760.jpg (196.84 KiB) Viewed 3971 times

But you can also start tweaking Wipe's result (including masking out the dust in the top left and the vertical line to the left), try to clean up more. For example like so;

StarTools_2761.jpg (499.34 KiB) Viewed 3971 times

You should notice the background is a lot more even in the diagnostics stretch. The most apparent background "defect" is starting to become just the noise, and that is a very good place to be. Th background is now so even that AutoDev yields this without any RoI set;

StarTools_2762.jpg (315.86 KiB) Viewed 3971 times

That is not to say you shouldn't use an RoI here, but it's just to show that it is now possible to leave a little bit more of the faint background detail in, without background unevenness becoming a problem.

At the end of the day, the two diagnostic stretch views were really different, showing different (very real) issues, however those issues all existed in the "margins" of the data. It depends on your stretch and how much you want to push your data, whether those differences really matter.

Rule of thumb is, of course, to 1. acquire/calibrate/stack the cleanest data possible, and 2. clean up the data as best as possible in Wipe. You should find, however that if #1 was done well, then #2 is much easier (to the point of not requiring much tweaks if at all).

I hope that helps!

[Mike in Rancho]

This is indeed a helpful reminder of what Wipe is actually for. I too probably tend to overuse it for squashing down sky noise, rather than appropriately doing more of that with the right AutoDev settings or even the dynamic range in Contrast, etc.

That video (or future AI? ) with real world flawed data would be helpful too! It would be good to know just what to be looking for as far as bumping up the aggressiveness setting.

I think one thing is, the manual/instructions say that after Wipe there should be a dark background. So, I've kind of tried to do that and it usually takes rather strong Wipe settings. But here it looks like it's actually an "even" background that is the goal.

[ionia23]

I've been getting some very interesting results by not "messing" too much with default settings. I figure that the incredibly complex algorithms in StarTools are designed to "detect" and make a reasonably good guess of what to do with the data.

Most of the practice I've been doing has been with very bad data, old images from days gone by. The kinds of things that would make your collective eyes start twitching. It's been a real surprise that even years ago with my old setup, I caught a lot more data than I thought I did.

For a workflow what I did was this:

1. Autodev. No changes, just to look at what's wrong. Stacking artifacts and vignetting. Object of interest (M106, I think...) totally off center. Cancel out of that and don't apply the stretch. Very bad chromatic aberration and star "smearing". Remember, this was shot with an old Sony cybershot camera shot through a 15mm eyepiece on a 203mm reflector, unguided, blah blah blah.

2. Turn off noise tracking and head over to Lens. I was floored what an amazing job it does "smooshing" all the aberrated color rings back into shape.

3. Crop. Lots of stacking artifacts were detected in step 1. Also using this as an opportunity to center the object of interest.

4. Save to a new file.

5. Back into Autodev, no ROI, save completely overstretched image to emphasize errors. Save tp a temp file called Stretched.

6. Go into Mask and clear it. Now comes the fun part: Identifying every single star in the image. This takes awhile. Save the mask for star repair later.

7. Back out of Startools, reopen, an open the pre-stretched file. Turn off tracking and head over into Repair to fix all the splayed out stars affected by poor tracking. Use the mask previously saved to layer all of them and give 'em a squeeze. Circular stars. Much better.

8. Over to Heal to repair large dark anomalies and anything really 'unnatural' that will throw off the various modules.

9. Turn on tracking

10. Wipe. Don't change anything. No masks, nothing. Whatever Wipe decides to do, leave it. Save those changes.

11. Back to Autodev. Now it's time for a better ROI stretch. Don't go crazy with how much you stretch the image. Save.

12. Back to Wipe. Now you can tweak the settings a bit to match the problems you're wiping out. I've found this to be very easy the second time through.

Now I have usable data for the other modules. This technique seems to work well, letting Startools do a lot of the work for me.

I'll post what I pulled off with an old shot of M31 when I get home. Definitely not a Picture Of The Day, but I'm pretty impressed.

[Mike in Rancho]

Interesting. And reminds me of the prior thread with the mask issue.

I guess I'd have to grab some data and try to follow along to see what is really happening. Just from a quick read and off the top of my head though, I'm a little concerned about the possible loss of the linear state, which much of ST is built around having in its vast time traveling memory. But maybe that's not happening?

A few times when I do some creative stuff and make changes to files in ST, I will take it back to AutoDev but not accept the stretch, basically (I think) reverting it back to linear for purposes of saving the file.

[admin]

9. Turn on tracking

Hmmm... how are the data still linear at this point?

[ionia23]

9. Turn on tracking

Hmmm... how are the data still linear at this point?

I probably just mis-typed the order of operations around that point in my description . That's my "oops" (one of many).