M 81 and M 82

[decay]

First of all:
@Stefan B Hi Stefan, thank you very much for evaluating that ASTAP stack. I will later respond in detail. For now: my impression and results are pretty much the same like yours. I guess, that's a good sign

Yes you can put your Bias into both Darks and Dark Flats ...

Hi Mike, Ok, thanks - that means that I got it right.

... if you want to set up your calibration that way. Typically I think that is more of a Nikon (D5300 and up) technique, as Canons are better able to utilize real Darks, but the basic calibration formula should work.

Background is only, that I took flats and bias frames for use with DSS. All attempts using darks failed for me, the result was always worse than without darks. Somewhere I read, that the 2000D would have a low dark current and I my impression is that this is true. Maybe I will take darks in future, but for now this seems not to be the most urgent point. But I will most likely take flat darks from now on ...

Though I am a bit confused by your description of a "second run" with the Bias reloaded into Dark Flats. AFAIK if the master flat is already created those ought to be ignored? But if not, the calibration would be incorrect as the flats would be double bias-subtracted.

Yes, I guess, I was a bit confused writing this description, but you understood, what I wanted to say I am quite sure, that I reloaded the bias frames into Dark flats, but perhaps - and I hope - I only messed it up or it had no effect. I cannot believe that double subtracting the bias frames would work. I will take a closer look, when I do this stack again without "Auto levels".

If you want, take your time with ASTAP and just step through things manually, one-by-one. That'll help to learn it, and every bit of software has a learning curve. Even ST, which you know, is made for beginners.

If you load your dark flats and hit analyze you can see the statistics for each frame. Then, same with the flats. Then there's a button that says replace flats and dark flats with master flat. This happens automatically if you just load everything and hit stack (and that's why your files "disappeared.") Next, same thing with your darks (or whatever you loaded into the darks tab). Finally, lights can be so analyzed, and the stats reviewed to see if you want to cull any bad shots.

Thanks for your detailed description. I'm quite sure, I did things that way. But as said, maybe I messed something up. I will do that again with a little more calm, as you suggested and then report back and upload the new stack as well. So please don't waste time with the old stack.

Do not worry about 2x G making your images green. They are averaged, not added. This improves the green SNR a bit, however, and that can be taken advantage of in Synth L creation. Ivo takes care of that math, of course, because he's Ivo! Just open or a proper OSC compose will do it. Anyway, the green cast typically found in OSC has more to do with the channel response of the sensor and bayer matrix, where green tends to be the strongest.

Ok, I think I have to say goodbye to my theory it would be green because of of 2x G pixels on bayer pattern. Thank you for clarification! I will dive a little bit deeper into the things you explained. That sounds quite interesting ...

If you want, go ahead and link a handful of subs of each type (light, flat, bias - whichever were used) and maybe we can check things out for you.

That would be very kind of you! But I will first check things in ASTAP again on my own before wasting your time. Maybe tomorrow and then I will report back here ...

Best regards, Dietmar.

[decay]

Hi Mike,

If you want, take your time with ASTAP and just step through things manually, one-by-one.

I did exactly that and now everthing works fine! Don't know, what went wrong; I reused the folders for multiple runs, maybe I loaded the transformed FITS-Files in addition - at the end of the day I simply messed it up, no excuse .

Don't worry about the green -- it is entirely normal (unless something else went wrong - like camera settings that could be altering the RAW files?) and will be properly handled by Wipe (mandatory) and Color.

Here is the new ASTAP-Stack without "Auto levels" checked on:
https://c.web.de/@334960167135216273/s4 ... GeDFawxamw
I did a complete rendition, and as you expected the green colour cast is no problem at all. The result is quite fine.

Here is a single light frame:
https://c.web.de/@334960167135216273/qt ... HqsQaFeNfw

Maybe you could take a short (!) look at both, the stack and the light frame, and give an assessment, whether this is OK that (very green ) way? That would be very kind!

Thanks a lot, Dietmar.

[decay]

Hi Stefan,

I always use flat darks but when a software does not have a tab or something similar for them I use the software's corresponding Bias tab. The other way round should also work. Both flat darks and bias frames are for calibration of the flats (if you do not use any of them you'll probably get a master flat which leads to overcorrection, i.e. inverse vignetting/brighter instead of darker corners). And both do the job by subtracting the read noise/offset from the flats if I am not mistaken. Which one does a better job is a matter of debate and probably dependent on different factors like CMOS vs CCD but in my experience they do what they should do with my DSLR, no matter if bias or dark flat.

By the way, I use the flat darks as darks also. Of course they aren't exposed as long as the lights but I hope that thus I get rid of part of the fixed pattern noise.

Thanks for your extensive remarks. Using bias frames both as darks and as flat darks works quite fine in ASTAP (I just messed it up at first) and the theory fits. I will probably take flat darks in future as well or at least try it.

Thank you very much for the comparison between the DSS and ASTAP stacks. Your rendition of the ASTAP stack looks pretty much like mine! And I came to the same conclusion, that the outcome of M81 shows stronger and more pronounced details in the outer galaxy regions. And I played a bit with the rendition of the background and I have the impression, that this works a bit better with ASTAP. Hard to judge, anyway. (Sometimes I believe, I have some kind of background mania.)
If you compare the differences of DSS and ASTAP stacks between your data and my data, what would you say? Does that behave in a similar way? Or is the difference in your stacks way greater?
At the end of the day the resulting image counts, but I wondered whether there is way to check the quality of the stacks at an earlier stage ? When I open up the ASTAP stack in ST, after the first Autodev, the grain of the background noise looks "smoother" in some way and the objects are more pronounced compared to the DSS stack. But it's hard to find a reasonable measure for this ... just a thought, nevermind

Now I am very curious about your opinion about my Siril stack:
https://c.web.de/@334960167135216273/1l ... WKuzKvJ1cQ

It was calibrated with bias frames for lights and flats, just like the DSS and ASTAP stacks (and it it green ). Thanks again for your help, I hope that this is a bit interesting for you as well ...

Best regards, Dietmar.

[Mike in Rancho]

Hi Dietmar,

Nice stuff you got there. Glad you got ASTAP more figured out. And when I use it for DSLR files I actually also use the AstroSimple interpolation algorithm. It seems to me to be an improvement over DSS bilinear.

So, after looking at the single sub and processing the new stack, some random thoughts... (and just being nitpicky in case you want to chase after anything in the future, but again it's really nice)

The simple header in the CR2 indeed says you have a 2000D. After a little digging around, it seems that is the European name for the T7, also known as the 1500D in some parts of Asia. There is no other 1500D that I can find. And in fact, if you dig into the EXIF file, the camera model is even listed as "EOS Rebel T7 / 2000D / 1500D / Kiss X90."

So, if something is recognizing your camera as a 1500D, I think that is exactly correct. And also, it should be readily interpreted by dcraw/libraw as a fully-supported camera. Now, why you may have had a little Bayer trouble at first, I don't exactly know.

For your greens (which again isn't really a problem I don't think), are you using any kind of White Balance setting? The EXIF says "manual," but I don't know what that means as I only have Nikons. For my D5300, I have it set to the default "Auto," whatever that means. In any event, I know that DSS has options for (1) no white balance applied, (2) use the camera's white balance, or (3) it's own white balance (if no box is checked). Ivo's recommendation is of course (1). I do not see any such options in ASTAP, so I don't know what Han does regarding this issue, actually.

The single sub struck me as a little bit bright, but not terribly so, and the histogram was at 1/4 or less. Perhaps it's just the big 200/1000 Newt really gathering starlight and so quite a few cores seem saturated. But that's a balance one must make, and perhaps worth sacrificing if you don't want shorter subs.

I did notice some little strings of colored hot pixels in the stack. It is possible that darks could correct those (although I think ST got all of them anyway). Or dithering, which I believe is what most non-dark-taking DSLR imagers do.

I also couldn't quite tell if there may have been a slight focus issue, or perhaps some trailing with the 90s subs, or maybe the Baader MPCC needs a little dialing in with some spacers, or all three? Unless you already have the MPCC dialed in as best it can be. But the finished stack isn't bad and those are really just left with some star warpage on the left side. Still...better than what I was able to do with the Baader!

For ST I started off with just a slider click cropping on each one, binned to 35%, and then a Wipe with pretty high aggressiveness, 98 I think. In Color I used star sampling with the SVD mask retained, and then went back to full mask. That still left some green galactic cores and I need two slider clicks (meaning the mouse wheel, not the single increment end-arrows) of green reduce while looking at MaxRGB, and that was that. Adjusted saturation where I wanted and put highlight repair to 5.

I really like the web-like red inner detail you captured in M82.

Anyway just my two cents of little tips, hopefully helpful. Stefan will probably have better analysis.

Decay m81_m82_astap_green ST8 1A.jpg (414.97 KiB) Viewed 5128 times

[Stefan B]

Hi Dietmar,

Now I am very curious about your opinion about my Siril stack:
https://c.web.de/@334960167135216273/1l ... WKuzKvJ1cQ

I will have a look but can't tell when I'll be able to do so. Maybe not before next week, sorry.

Regards
Stefan

[decay]

Hi Stefan,

I will have a look but can't tell when I'll be able to do so. Maybe not before next week, sorry.

Yes, of course! Please feel free to reply next week or next month or to stop this discussion entirely. You have already helped me so much. There may be more important things in life ...

Best regards, Dietmar.

[decay]

Hi Mike!

Thank you very much - once again. So much information at once, much appreciated! I still have to work up all the things from your last posts. I will respond one by one in multiple posts - that's too much all at once

and just being nitpicky in case you want to chase after anything in the future

Of course, being nitpicky is sometimes necessary to get better. And good AP is obviously the sum of many small things that all have to be good. And yes, I want to chase these things you mentioned!

Till then, Dietmar.

[Stefan B]

Hi Dietmar,

just sharing the results of my quick processing and comparisons ... couldn't wait to have a look

m81_m82_siril.jpg (170.99 KiB) Viewed 5092 times

M82 comparison astap siril.jpg (59.16 KiB) Viewed 5092 times

M81 comparison siril astap.jpg (71.72 KiB) Viewed 5092 times

Had the impression that the noise grain in the siril stack is smaller than in the ASTAP stack (rather single pixel noise), which is a good thing, right? Thus it was easier for me to reveal the faint stuff in M 81. The details in M 82 look better in the Siril rendition for me, too. On the other hand, brighter details in M 81 look a bit better in the ASTAP version. I am not sure if that's due to stack or due to the post-processing.

These are my immediate impressions. What were your results!?

Regards
Stefan

[decay]

Hi Stefan!

Had the impression that the noise grain in the siril stack is smaller than in the ASTAP stack (rather single pixel noise), which is a good thing, right?

Yes, that was my impression as well, right after first Autodev (and later on as well). Your rendition of M82 / Siril looks nearly like a copy of my rendition, which is a good thing as well, right?

On the other hand, brighter details in M 81 look a bit better in the ASTAP version. I am not sure if that's due to stack or due to the post-processing.

Yes, hard to say. I think there are two approaches to compare:
1. post-processing of both stacks on their own with the goal to get the best possible version of each.
2. post-processing of both stacks side by side, trying to do mostly the same on both sides.

For the first approach, Siril made the race for me, but not that striking.

I will try the second approach next days with at little calm and refer here with the result. At lot of homework to do

Nevertheless, results may depend on this concrete data set. All in all both, ASTAP and Siril, seem to do a better job than DSS. This will be a difficult decision for me

Best regards, Dietmar.

[Stefan B]

Hi Dietmar,

Your rendition of M82 / Siril looks nearly like a copy of my rendition, which is a good thing as well, right?

That depends if you like the renditions

Yes, hard to say. I think there are two approaches to compare:
1. post-processing of both stacks on their own with the goal to get the best possible version of each.
2. post-processing of both stacks side by side, trying to do mostly the same on both sides.

I went for option 1. Since you want to know which stack allows you to get the best possible final image. I don't want to do scientific experiments which one is the better stacker. And still, option 2 might not be suitable to get to know which one is the better stacker since both stacks might be different enough to need different processing to do them justice.

Nevertheless, results may depend on this concrete data set. All in all both, ASTAP and Siril, seem to do a better job than DSS. This will be a difficult decision for me

But it's not a difficult to decision to abandon DSS, is it? In terms of Siril vs ASTAP I prefer ASTAP at the moment since it gave me consistent high quality results. But once using ASTAP resulted in strange hot pixels in the stack with one of my data sets. In contrast, Siril did a great job. So I like having Siril as a great alternative.

I have no experience with non-free stackers like APP and PI. I heared a lot of good things about both. Especially PI is said to have a great stacker. But as long as I get good results with free stackers I am not willing to invest in APP or PI since I wouldn't use the features besides stacking. Probably not worth for me.

I will try the second approach next days with at little calm and refer here with the result.

I am curious about the results.

Regards
Stefan