StarTools

Apologies for the length of this post!

I have been struggling for some time to get decent calibration and stacking of my lights. I started out using DSS for stacking but changed to ASTAP as it seems to give me cleaner, lower noise results.
I use an ASI294MC pro, which has a reputation for occasionally being difficult to prepare darks and flats for. It lives on an RC6, with the CCDT67 reducer. The scope has recently been well collimated using SkyWave. A CEM40 carries the RC6. I use APT to control shooting.

The symptoms I am getting include –
1. I have rapidly changing gradients across my stacked images. It looks as if the flats are not properly modelling the vignetting of the scope.
2. When I wipe the image after initial AutoDev and Crop, the gradients and vignetting are not well removed, and I often (but not always) end up with a green cast and light toned noise in the shadow areas of my wiped images. I suspect Wipe is removing some faint detail.
3. On some (but not all) images the image looks monochrome, and I need to push saturation up to 500% to get normal colour in galaxies and stars.
4. After going through all stages in the recommended order, when I take the processed image into the colour module, it has a strong green cast and needs about 1.8-1.9 units of green bias reduce, plus some red bias reduce, to produce a normal colour balance. I open the image using “option 2” and use the recommended ASTAP settings. The fits file metadata says that the file has the correct RGGB bayer pattern. Is it normal for an OSC image to have a green cast at this stage?

I use a Lacerta flat panel with the Lacerta dimmer to take my flats. Previously I’ve used 2 layers of T shirt material to get more dimming.
I have recently read that using the FlatsAid in APT (which I have used until recently) does not produce good flats with this camera, as the average ADU will be around 25k and the camera benefits from flats with lower ADU. Flats produced with APT Flats Aid, when stretched, show the lightest area displaced up and right from field centre.
So on my most recent set of flats I set the exposure manually, to produce an ADU in the 11-17k range, with 3sec exposures and 4 layers of white T shirt material. The flats produced with this method, when stretched, show the lightest area closely aligned with field centre. I am completely baffled why this should be, but using these flats to stack produces images where the wipe parameters can be less aggressive and more faint detail is retained. So with this change I may have resolved symptoms 1 and 2, but not 3 or 4.

I’ve included a sample ASTAP output file (of M33), and the master dark and master flat used to produce it, here - https://www.dropbox.com/sh/szotwrfiwu42 ... Z0DGa?dl=0
I’m rather at my wits end on this, so any advice would be gratefully received.

Clear skies

Paul E

Hi PaulE54

Had a brief go at your data in 1.9.536 and I can see the gradients you mention. (Should point out I am not one of the experts here.)

I found that the colours in the image were pretty good. This is a quick Optidev - Crop - Wipe - Optidev - Color (so no detail enhancements) :

I also had a look at your flat and dark. The flat looks fine, but the dark is a bit strange and looks, when stretched, as if there might be a light leak somewhere:


It might be worth re-shooting the darks to see if you can get rid of the shadows and lighter parts? Also, consider stacking without darks and seeing if that makes a difference?

Hi Paul,

Well as you may see from another ongoing thread, I sometimes have images that struggle with complex gradients (possibly exacerbated by Newtonian flats troubles) too. I do wonder if a dumber Wipe Jr (FilmWipe?) module, that isn't so hunting for LP undulations, might sometimes just model across a difficult image better? Or I just need to get better at Wipe in unique situations. As shown in that thread, turning on more features such as Uncalibrated may help, though with pulling back the sliders so as to not erase details if one can. I'm getting closer but still not dialing in the edge behavior the way I want even with the three options.

But enough about me. (You may want to fiddle with Uncal though)

A green cast is frequently expected, even after Wipe sets the channel floors. This is ultimately handled in Color. This being OSC, do you have all color manipulations turned off in ASTAP? Also what debayering did you use? I think bilinear is the suggested algorithm, though I actually found astro simple to be better myself (though just barely).

Do you have samples of individual subs of each type (I'm assuming L, D, F, and DF here), one each should be good, that you can link? The masters can only tell so much - although that may also be my not yet fully understanding file analytics in PI (can't figure out how to turn on 32-bit stats yet! ).

I might not be able to get to them until tonight, but will scope them out if you can provide a set.

Just roughly looking at the files though in preview, with an STF applied to the dark and the final stack, I do see a few potential pitfalls --


Notice the master dark has a gradient to it, as well as a corner (top right) that looks like hard (meaning physical blocking of light) vignetting. I suspect a number of things. Primarily a light leak, but also perhaps collimation and/or an image circle too small for the sensor (or something else in the optical train cutting off that corner). Are you taking darks with the camera off the rig, cover installed, and perhaps in a closet or box? Same would go for your dark flats. If you can't remove it, then some other method of fully blocking off all light from...everywhere.

I'm curious why I don't see the amp glow in the master dark, this being a 294 and all. I will have to muse on that, perhaps I am misunderstanding something. I don't see uncorrected amp glow in the stack though.

On the final stack/master light you can where subtraction of that dark seems to have caused an overcorrection in that one corner and a little in the opposite side. There are some stacking artifacts but not a big deal, crop those off before using Wipe of course. But also as you can see there's what again looks like unusual vignetting with that super dark lower left corner.

I could keep speculating but maybe let's try this - take care of the potential light leak in the darks first and then see where things stand on a new stack. With that issue removed we might be able to then start homing in on other issues. You should be able to just take new darks and dark flats though your acquisition program, with the camera covered and in a box/closet, at all the appropriate settings and cooled temp.

Ultimately we might also have to look for any light leaks in the lights/flats, but maybe we'll see after trying new D+DF?

Hope this helps!

EDIT: I was beaten to it!

Though stacking without darks (unless replaced with say, clean bias on both sides) usually leads to flats miscalibration.

Thanks both....really useful input. I'll add some single frames into the dropbox folder (now done, see below).
That gradient on the darks is definitely worth a look. It runs from lightest at top left to darkest at bottom R. Values from 2100-ish to 2700-ish.
And now you mention it, where has the amp glow gone?

Lots to think about now. I have had the 294 apart recently to dry out the dessicant tabs, but it went back together fine. All the connections are screwed, so can't think where light can be leaking. I'll cover the entire thing up and take some new darks tomorrow. If there is light leakage it will also affect the flats and especially the dark flats to some degree.

edit -
the only think I can think might be "hard vignetting" is a 1.25 inch UV/IR filter very close to the camera. I'll have a close look at this.
I have loaded some single frames to the dropbox folder above - a light, a dark, a flat and a dark flat.

My ASTAP settings (v2022.03.22) -
“stack method” settings are –
• stack method is sigma clip average. Sigma factor is 2.
Raw OSC image settings are -
• Convert OSC images to colour is ticked.
• Auto levels is OFF
• Normalise OSC flat is ON
• Colour smooth is greyed out
• Preserve_R_nebula is ticked – should it be?
• Bayer pattern is correctly recognised as RGGB
• Demosaic is set to bilinear interpolation.
• RAW conversion is blank – think this is OK for CMOS?


Paul

Hi Paul,

Some great analysis and potential leads here already - my money is also on the strong possibility we are dealing with a calibration frame issue here.
Here too (as in Mike's thread), Wipe is asked to remove an artificially introduced (e.g. non-celestial) gradient in the face of multiple other oddities of non-celestial nature; just like in Mike's thread Wipe will always err on the side of caution.

My favorite sanity check, is always to start with a stack that is not calibrated at all (e.g. light frames only), and then to progressively add the other calibration frames;

• Adding the flat frames should reduce the vignetting, gradients and dust donuts. If not, they're bad.
• Adding the flat darks should never introduce gradients. If not, they're bad.

This process of elimination will hopefully turn up areas for immediate improvement, and possibly even areas for improvements down the line.

PaulE54 wrote: ↑Wed Jan 11, 2023 6:13 pm
My ASTAP settings (v2022.03.22) -
“stack method” settings are –
• stack method is sigma clip average. Sigma factor is 2.
Raw OSC image settings are -
• Convert OSC images to colour is ticked.
• Auto levels is OFF
• Normalise OSC flat is ON
• Colour smooth is greyed out
• Preserve_R_nebula is ticked – should it be?
• Bayer pattern is correctly recognised as RGGB
• Demosaic is set to bilinear interpolation.
• RAW conversion is blank – think this is OK for CMOS?


Paul

Hey Paul, I'll take a look at those new subs and see if anything else jumps out at me. Will post what I find.

Normalize OSC Flat -- I don't know if I would go there.
Han's manual even says don't use unless you're having issues due to taking flats using using very red or blue light.

Preserve R Nebula I'd probably avoid also, though it says it is only affecting regions around bright stars. But even so, it's still manipulation that I don't know is necessary. Maybe a cosmetic thing? And here, with a galactic non-red-nebula target, I'd worry that it might go artifacting your stars.

Just a quick holding reply to express my thanks to you all. What a helpful forum and community! Loads to investigate here, I'll report back.

Paul

Hi again Paul,

All your file settings seemed fine to me, everything matched. F + DF were 3 seconds, appropriate for a 294. I didn't see evidence of light leaking in the DF. Stretching that out showed a pretty even field of common extreme deep patterning (not a problem).

The flat and light I could only eyeball but didn't see anything terribly amiss. Light could have LP gradient as well.

What did throw me off were the dates. It seems the dark came from several months ago, the lights were taken last month, but the flats and dark flats were new? No change in the optical train since you captured M33?

But as suspected the leak in the darks is the major snafu at this stage.

I subtracted the dark from your light and inverted the image for easier viewing. All the white pixels are black clipped, meaning 0.0, where your dark was actually brighter than your light.


So, much data being erased. And even where you aren't fully clipped, I think it's greatly mangled across the image anyway by such a heavy subtraction. And, with the left side of the equation thrown out of whack, flats correction is going to go off the rails as well.

Note too that since calibration is done in the bayered state, I wouldn't trust the colors here at all after these pixels are debayered.

So, hopefully nothing more needs done here than taking new darks and stacking all over again.

I would delete those darks and any dark masters that were made from them though.

Mike, that analysis is so useful - many thanks. I'm running the various stacking trials that Ivo recommended as we speak, so I expect the "lights and master flats only" test (no darks) to provide an improved result - though clearly not as good as with correct darks. I'll also take a new set of darks as you recommend with the camera off the scope and capped. And of course I do need to track down that light leak....

Reason the flats and DF are so new is as per original thread I'd seen a recommendation to shoot flats at lower ADUs - so I did that the other day.

I'll report back in a day or two.

Paul

Ok that went much quicker than I expected!

Key finding - I have numerous sets of darks dating from when I first got the camera in late 2018. While waiting for numerous sets of ASTAP stacks to run, I've opened sample dark frames from all my sets of darks, including some 30sec and 20sec darks that I took AFTER the problematic set from 13th Aug 2022. And NONE of them show the gradient.

I had read that it was good practice to re-shoot darks every 6 months or so, so I'd been doing that

So for some reason I had a light leak when I took the Aug 22 set. Probably sloppy technique - maybe I didn't cover the scope carefully enough? It doesn't exist on the most recent shorter exposure sets, so not a scope / camera fault.

To show how badly gradients on your darks can screw things up - with faulty darks this was the best I could do - With correct darks I was easily able to get this - could probably do better with more care and time And to answer the question about "can't see the expected amp glow for a 294", when you stretch a single dark frame from a set without the light leak - So many thanks for all the help guys, incredibly useful. Now I'm off to reprocess all the other images I've taken between Aug last year and now....

Paul E