what is wrong here, central region blown

[bomberbaz]

Ok so yesterday I put up the top image showing the wavey gradient and so now I have also attached the accompanying master flat. This FLAT is 0.16 seconds, just the tablet screen dimmed and laid over the top of the scope.

As you can see quite uneven and suggested to be the reason for the poor image which seems feasible. The tablet screen not giving an even glow being my thought.

Next the new image and flat (also below) showing the central region blown.  The accompanying flat though seems decent. New flat 0.9 second, two pieces white paper and a brighter screen on top. Histogram at 50%

The base light data is the same for both and yes the flats are in the right order. Stacked in ASTAP.

What is going on here, I am mystified!

 

This is the second batch using the new flats. 

[Clarkey]

When was the first flat taken? I have had similar gradients which I could not fathom out with my Rising Cam IMX571. Although I never found out for definite, I think it was dew. Since getting a new heater strip for the camera and turning the internal dew heater to 'max' it has not been a problem. Your issue might be something completely different - but it is the best I can come up with.

[bomberbaz]

1 hour ago, Clarkey said:

When was the first flat taken? I have had similar gradients which I could not fathom out with my Rising Cam IMX571. Although I never found out for definite, I think it was dew. Since getting a new heater strip for the camera and turning the internal dew heater to 'max' it has not been a problem. Your issue might be something completely different - but it is the best I can come up with.

Thanks for the reply, these were taken Monday, the scope was dry when I brought it in, no evidence of dew.  I am completely baffled but one thing 9is the scope, an ED72 is new so maybe a visual session is needed to rule that out as a problem. That said, the problem seems to me at least to revolve around the calibration frames.

[alacant]

Hi

To help us diagnose, maybe post -links to- a light frame and one each of whatever calibration frames you used, along with a description of whatever hardware has produced such, optical train detail in particular.

Cheers

Edited April 27, 2023 by alacant

[bomberbaz]

I am having a lot of problems with the ASI Air and I am wondering if there is a link between that and this issue. I will post up links to light plus calibration that relate to the above shortly though to see if there is anything that can be gained from this.

[ollypenrice]

Flats are simply photos of the incoming beam as it hits the chip, which means they all have more or less the same form with only the dust bunnies varying. Neither of your posted flats take this form. The first one is closest because at least it has  a bright centre getting darker towards the edges. However, it doesn't look credible.  The bright inner circle dims with unusual suddenness and then there is very little further drop-off into the corners. There is also a bright band running upper left corner to lower right corner. The latter cannot be genuine.

The second flat is flat - too flat to be credible. Not even my TEC 140-with-flattener is as flat as that and it is corrected for medium format film.

What we then notice is that the illumination of the second galaxy image closely resembles the illumination of the first flat, with the same bright inner circle dimming quickly to a very even outer region. My conclusion would be that the second image has been largely unaffected by the second flat, which is to be expected since the second flat is so flat as not to change much when applied. This suggests that the first flat was on the right lines regarding the general illumination but had irregularities of its own in the form of the bright diagonal. This bright diagonal has influenced the second image.

What might be worth a try would be this: Set up the rig as it was for the first flats but, during capture sequence, rotate the screen continuously. It doesn't matter if its moving when the exposure is made because you're not trying to image the screen itself. When stacked, this ought to average out any screen irregularity.  Alternatively, just try a different light source. I'm pretty sure you present one isn't even.

Olly

Edited April 27, 2023 by ollypenrice
Clarification

[bomberbaz]

21 minutes ago, ollypenrice said:

Flats are simply photos of the incoming beam as it hits the chip, which means they all have more or less the same form with only the dust bunnies varying. Neither of your posted flats take this form. The first one is closest because at least it has  a bright centre getting darker towards the edges. However, it doesn't look credible.  The bright inner circle dims with unusual suddenness and then there is very little further drop-off into the corners. There is also a bright band running upper left corner to lower right corner. The latter cannot be genuine.

The second flat is flat - too flat to be credible. Not even my TEC 140-with-flattener is as flat as that and it is corrected for medium format film.

What we then notice is that the illumination of the second galaxy image closely resembles the illumination of the first flat, with the same bright inner circle dimming quickly to a very even outer region. My conclusion would be that the second image has been largely unaffected by the second flat, which is to be expected since the second flat is so flat as not to change much when applied. This suggests that the first flat was on the right lines regarding the general illumination but had irregularities of its own in the form of the bright diagonal. This bright diagonal has influenced the second image.

What might be worth a try would be this: Set up the rig as it was for the first flats but, during capture sequence, rotate the screen continuously. It doesn't matter if its moving when the exposure is made because you're not trying to image the screen itself. When stacked, this ought to average out any screen irregularity.  Alternatively, just try a different light source. I'm pretty sure you present one isn't even.

Olly

Thanks Olly, that all makes good sense.

Just a quick question of you, would a massivly out of focus affect the image being received, as it was a very long way out of focus for the 2nd flat and I only realised this last night. Schoolboy error.

[alacant]

20 minutes ago, bomberbaz said:

long way out of focus for the 2nd flat

You must not change the focus -or anything else in the optical train- between taking light and flat frames.

 

[bomberbaz]

1 hour ago, alacant said:

You must not change the focus -or anything else in the optical train- between taking light and flat frames.

 

I think there is a very good chance the problem lies there and also I shall follow olly's advice re rotating the screen for a more even light grab

[ollypenrice]

43 minutes ago, bomberbaz said:

I think there is a very good chance the problem lies there and also I shall follow olly's advice re rotating the screen for a more even light grab

Ah yes, the illumination is changed enormously when the focus is moved. That would account for why your second flat is so flat.  Note that you are not focusing on the flat panel, you are replicating the illumination of the original focal position.  (The changes in focus between filters are not enough to need new flats, by the way.)

Olly

[jacko61]

On top of the advice given above, I would recommend stretching a white T-shirt over the scope before taking your flats.  This will give a much more even light when the tablet is switched on. 

Graeme

[bomberbaz]

Thanks Guys, especially @ollypenrice but all who chipped in. I have had to bin the M81 data, various changes to the camera chain means I am going to struggle with it and simply cannot be bothered with it now.

However the M3 effort below is a little over 30 minutes worth, which has only had a simple stretch is how I know it should look from my garden. Gradient right to left from the hotel car park lights across the road.

I re-did darks, dark flats and flats. Flats done as per advice from Olly and Jacko, results speak for themself. 

cheers

steve

Edited April 27, 2023 by bomberbaz

[AstroNebulee]

18 minutes ago, bomberbaz said:

Thanks Guys, especially @ollypenrice but all who chipped in. I have had to bin the M81 data, various changes to the camera chain means I am going to struggle with it and simply cannot be bothered with it now.

However the M3 effort below is a little over 30 minutes worth, which has only had a simple stretch is how I know it should look from my garden. Gradient right to left from the hotel car park lights across the road.

I re-did darks, dark flats and flats. Flats done as per advice from Olly and Jacko, results speak for themself. 

cheers

steve

Nice image Steve and hopefully all your issues behind you now. 👍

Lee

[ollypenrice]

I think a good gradient removal tool would flatten your M81 if you just stacked it without flats.

Olly

[bomberbaz]

2 hours ago, ollypenrice said:

I think a good gradient removal tool would flatten your M81 if you just stacked it without flats.

Olly

It's fine really olly, the data wasn't the best and losing it doesn't bother me. I am spinning enough plates as it is.

However thinking about it, nothing wrong with having a reserve project to learn more about. I shall recover the data to my back up drive and  have a go.

My processing skills need improving, so maybe use this an an opportunity to learn something new!

Edited April 27, 2023 by bomberbaz

[Elp]

Hopefully you've sorted the issue. Sometimes such things can happen usually if the flats are overcorrecting because their dark bias signal (also incuded within dark flats) has not been removed correctly, I still struggle with them sometimes. Your data however can be salvaged somewhat with synthetic flats, below is just a demonstration of your original RGB images 1 and 2 above, how they look in Siril with a Histogram stretch preview, and after synthetic flats have been applied. The process isn't perfect but it sometimes helps with problematic data in the absence of correct calibration frames, I get it a lot with Lextreme data.

[bomberbaz]

53 minutes ago, Elp said:

Hopefully you've sorted the issue. Sometimes such things can happen usually if the flats are overcorrecting because their dark bias signal (also incuded within dark flats) has not been removed correctly, I still struggle with them sometimes. Your data however can be salvaged somewhat with synthetic flats, below is just a demonstration of your original RGB images 1 and 2 above, how they look in Siril with a Histogram stretch preview, and after synthetic flats have been applied. The process isn't perfect but it sometimes helps with problematic data in the absence of correct calibration frames, I get it a lot with Lextreme data.

 

What did you do to achieve that, the third recovery was half decent, would be even better with more data.  Do you have any links to the process or methodolgy. I have never used synthetic flats (or even heard of them to be honest) 

[Elp]

Synthetic flats are exactly that, synthetic, artificial, made independent of your imaging session.

It is a similar process to that I detailed for you here:

You only need steps 1, 2 and 7 though (keep the stars) and modify like so.

Step 2: instead of painting the targets you want to keep black, paint over them with the clone stamp tool being careful to sample regions around the target similar to the background colour/intensity around your targets. So when clone stamped it kind of looks like there was nothing there where your target was (ie seamless). It doesn't have to be perfect but it helps if you take a little care.

Modified step 3: once your targets are clone stamped out goto filters > noise > median and choose value around 100 pixels, it will look like its kind of blurring out the image but you'll still see blotches where bright areas were.

Modified step 4: goto filters > blur > Gaussian blur and apply around 100 pixels again. You might need to adjust this so the image isn't posterised (hard steps between changes in colour levels leaving harsh edge steps between the changes like an ordinance survey map with its lines between changes in elevation).

Now apply the original step 7 "apply image" with the subtraction option.

Synthetic flat now applied.

Its not perfect as you can see from the stretched histogram image 2 above there's still a ring around the target, sometimes you can apply it more than once by doing the process a second time via your applied first synth flat image. Sometimes the results are a bit rough. Sometimes it doesn't work. But it's a useful technique to know. You normally have to do this after background extraction I find too when it hasn't quite worked.

[bomberbaz]

28 minutes ago, Elp said:

Synthetic flats are exactly that, synthetic, artificial, made independent of your imaging session.

It is a similar process to that I detailed for you here:

You only need steps 1, 2 and 7 though (keep the stars) and modify like so.

Step 2: instead of painting the targets you want to keep black, paint over them with the clone stamp tool being careful to sample regions around the target similar to the background colour/intensity around your targets. So when clone stamped it kind of looks like there was nothing there where your target was (ie seamless). It doesn't have to be perfect but it helps if you take a little care.

Modified step 3: once your targets are clone stamped out goto filters > noise > median and choose value around 100 pixels, it will look like its kind of blurring out the image but you'll still see blotches where bright areas were.

Modified step 4: goto filters > blur > Gaussian blur and apply around 100 pixels again. You might need to adjust this so the image isn't posterised (hard steps between changes in colour levels leaving harsh edge steps between the changes like an ordinance survey map with its lines between changes in elevation).

Now apply the original step 7 "apply image" with the subtraction option.

Synthetic flat now applied.

Its not perfect as you can see from the stretched histogram image 2 above there's still a ring around the target, sometimes you can apply it more than once by doing the process a second time via your applied first synth flat image. Sometimes the results are a bit rough. Sometimes it doesn't work. But it's a useful technique to know. You normally have to do this after background extraction I find too when it hasn't quite worked.

Thank you for this. I remember the walking noise post now, think I started but got lost and gave up.  Then forgot about it, the last bit is an age thing.

Anyway, now I have more experience I shall have another bash.