Blotchy background on my 130 PDS

[Bibabutzemann]

Hi, 

I have a problem with my 130 PDS. After doing background extraction on my calibrated and stacked image, im still left with a complex background gradient:

The problem doesnt come from "complex" light pollution.

Here is comparison between a flat frame stack (additive, to visualize the problem) and around 70 lights frames without registration and no flatframes (for better comparison).

I applied background extraction to both image to remove vignetting so that im only left with the complex gradient. On both images its almost identiacal.

The light blue circle in the middle come from the MPCCIII. As you can see on the first image, the circle gets calibrated out just fine. But the flats cant fully diminish the blotchy background, even thougt its similar.

 

So i did some tests to find the source.

The problem doesnt come from the camera or CC, because when i rotate my camera 90 degree, the pattern also gets rotated by 90 degree on the image.

So next try was to look into the focuser with my flat panel attached. I saw some bright reflexions on the corner of the 2nd mirror and on the outside of the focuser tube.

So i flocked both areas and i also covered the back of the primary mirror to avoid light leak.

Again, i took some flats and stacked them additive to see if the problem got smaller.

Here is the result:

No differene lol. 

Im slowly running out of ideas now. 

Maybe someone could give me an hint. Thanks for reading.

 

[Budgie1]

Which make & model of camera do you have and are you using any filters with it? It may also be handy to know what the camera settings are, in case that has any relevance.  

[Bibabutzemann]

1 hour ago, Budgie1 said:

Which make & model of camera do you have and are you using any filters with it? It may also be handy to know what the camera settings are, in case that has any relevance.  

Hi, im using the Eos 1200Da with the Astronomik L2 Filter. I only work with raw files, in this case @ISO1600.

But as i mentioned, the pattern rotates, when i rotate the camera.

So i doubt that another camera would solve this issue.

 

 

Edited April 21, 2022 by Bibabutzemann

[Budgie1]

I asked about the camera and filter because the effect looked similar to what I've seen with my ASI294MC Pro when using the L-eXtreme filter.

As your camera is the Da version, I assume it's only had the front filter removed, so has retained the UV/IR cut filter? I'm just wondering if the addition of the L2 filter may be causing something, have you tried taking the flats with the L2 filter removed, just to rule that out? 

[Bibabutzemann]

i get the same problem without the L2 filter. My camera is full spectrum modded, so thats why im using the L2 filter. 

I tested without the L2 filter and still have the same gradient. It would have surprised me if the L2 filter would cause the issue, since its a clip in filter and therefore always rotates with the camera. 

Im sure the gradient is independent of the camera, because like i said, when i rotate the cam by 90 degree, the gradient appears rotated by 90 degree on the image.

 

There is only one way i could cange the form of the gradient: When i change the focus. But i dont know what to make out of it.

 

EDIT:

I made a gif which shows how the gradient changes when i move my focuser from all in to all out.. It contains 9 frames and each frame is a additive stack of 20 flats. i only removed the vignetting.

Edited April 23, 2022 by Bibabutzemann

[alacant]

Hi

My guess is that the focuser tube needs shortening and blackening to prevent it cutting the light cone.

Otherwise as stray light must be coming from somewhere, you have to be systematic. Here are a few other ideas:

Exposures...

• With a different camera
• 2" filter on the cc
• Without dew-shield
• Without mirror clips
• Without cc
• In daylight with both ends of the tube blocked with a black shower cap.
• From a session at a dark site
• With blackend secondary edges and spider. 

Or just accept it. Bottom of the range reflectors need a lot of attention to bring (them?) up to astrograph standard.

Cheers and good luck.

Edited April 24, 2022 by alacant

[ollypenrice]

What do you mean by, 'Doing background extraction?' This term usually refers to a post-processing technique rather than to the application of flats as part of calibration when stacking. Could you take us through each step of your calibration and stacking workflow?

Olly

[ONIKKINEN]

I too dont quite understand the part about background extraction and removing of vignetting. Hard to say whats going on if the background has been extracted in your images already. I will say that SiriL background extraction will result in a blotchy image like yours if the sampler placement was bad or the flats didn't quite do their job, often both.

Can you attach a raw .FITS/.TIFF* file for people to have a look? One of each calibration frame (choose the flat randomly) and a light frame? Or just a calibrated light frame, that's better than nothing.

*Or .CR2 or whatever, since i just noticed its a Canon camera.

Edited April 24, 2022 by ONIKKINEN
Canon

[Bibabutzemann]

5 hours ago, ollypenrice said:

What do you mean by, 'Doing background extraction?' This term usually refers to a post-processing technique rather than to the application of flats as part of calibration when stacking. Could you take us through each step of your calibration and stacking workflow?

Olly

If i would only show the stacked flatframe, the vignetting would dominate and i couldnt show the problem. Thats why applied background extraction.

Normally my worklflow in Siril is like this:

1. median stacking of 25 bias frames

2. calibrate flat frames with master bias

3. multiplicative stack of 25 calibrated flat frames

4. calibrate lights with master bias and master flat

5. Registration and additive stacking of calibrated lights

 

5 hours ago, ONIKKINEN said:

I too dont quite understand the part about background extraction and removing of vignetting. Hard to say whats going on if the background has been extracted in your images already. I will say that SiriL background extraction will result in a blotchy image like yours if the sampler placement was bad or the flats didn't quite do their job, often both.

Can you attach a raw .FITS/.TIFF* file for people to have a look? One of each calibration frame (choose the flat randomly) and a light frame? Or just a calibrated light frame, that's better than nothing.

*Or .CR2 or whatever, since i just noticed its a Canon camera.

Here (google drive) is a zip fail containing:

A single flat frame, a single light frame, Master flat, a single Calibrated light frame, Master light (calibrated)

Edited April 24, 2022 by Bibabutzemann

[ONIKKINEN]

I notice a few things with the images and it does seem that flats dont work quite as they should here, but the blotchy result from siril background extraction always looks like that when my flats have not worked perfectly so that is more of a symptom that only is visible after running the tool rather than the actual issue. The tool really doesn't like mechanically failed flats and so the background remains flawed. I noticed that you have rotation between the single sub and the stacked image. Was this taken on 2 different nights so with different camera orientations? Or did you remove the camera during shooting? In any case, each time you remove the camera you need separate flats for that specific rotation and the previous ones will not work if you remove the camera and put it back.

First issue with the flats is that they are rather dark at 3115 ADU out of 16384 (14 bit camera). Would probably be better to either expose a bit longer or use a brighter light source for the flats. Not sure how much this effects all the other things here but i would look at bringing the flats exposure to around 50% of the histogram first.

To my eye it looks like you may have some kind of small motion in your optics between the flats and the lights, which is not a shocking surprise given you are imaging with a newtonian where many parts can move and so ruin flats. Its most apparent in the worm looking shadow on the top of your flats:

The same spot in your calibrated sub clearly shows that the shadow is still there and has not been removed by the flats:

It could be some hair or fiber that has moved on its own during imaging, but if not it tells of more issues.

To me it tells that there must have been movement somewhere in your imaging train between taking the lights and the flats. That could be the primary mirror moving in its cell, the cell itself being weak, the tube itself bending, the focuser drooping (quite likely will, if stock focuser), or the camera being rotated in the focuser. Dont know how that last thing could happen by accident, other than if you removed the camera between shooting the lights and flats but in that case not really by accident.

The resulting stacked image shows a dark center half-circle (purple:darkest, red:brightest) and a nonlinear gradient. I made the fine artwork below with paint and a screenshot in the rainbow false colour rendering mode and histogram stretch mode in siril:

This kind of error could be from light leaks, mechanically failed flats or both. Or inner reflections. Anything goes really. You do mention that things are flocked and light leaks proofed but i cant help but think that it still looks like light leaks are getting in somehow. Have you blocked the DSLR viewfinder? That would be light leak source number 1, leading stray light directly to the sensor.

If flats had worked there would not be this kind of nonlinear circle-y look to the center and it would be edge to edge linear (like the line i drew diagonally). Could be some overcorrection at play as well since the center is slightly darker than maybe it should, but not sure how to go about fixing that.

On 23/04/2022 at 20:03, Bibabutzemann said:

There is only one way i could cange the form of the gradient: When i change the focus. But i dont know what to make out of it.

Your focuser probably has some side to side travel in it when you rack the focuser in and out, which is why you should collimate in the focus position. Racking the focuser in and out will change the collimation in the case if this happens and that is why you see the effect change in different positions. Try putting a laser into your focuser and rack it in and out, i wouldn't be surprised if you see the dot move a bit on the primary mirror.

So my thoughts: Mechanical issues somewhere preventing accurate flats from being taken, if they were taken for each session and before touching the camera. I would look at the focuser first. I have mostly removed issues like this with my newtonian with an upgraded focuser and looking at reinforcing my tube so that there is less flexure.

To try and band-aid fix this since you cant really retake flats now, you can try to run background exctraction per sub in siril. It might leave a cleaner background than running the tool in the stacked image. You can also then see which subs had functioning flats and which did not, and could remove the ones where the background is messy after the extraction. If you are unsure how to do this, calibrate your images like you normally would and open the sequence of the calibrated subs. Then open the background extraction tool and set it to 1st degree order, 20 samples per line and 1.0 tolerance and generate the samplers. Then just click the Apply to sequence and siril will remove the background from all of your subs this way. Then you can browse the file list and see if you spot some of them being out of line (probably wise to look at first and last sub and compare) and remove those if you want a cleaner background. I think the image you have here is workable though and it wouldn't be that difficult to clean up later in Photoshop or gimp or whatever you use.

I wrote quite a ramble above, i hope there is at least half a sentence that makes sense somewhere in there 😃.

Edited April 24, 2022 by ONIKKINEN

[ollypenrice]

10 hours ago, Bibabutzemann said:

If i would only show the stacked flatframe, the vignetting would dominate and i couldnt show the problem. Thats why applied background extraction.

Normally my worklflow in Siril is like this:

1. median stacking of 25 bias frames

2. calibrate flat frames with master bias

3. multiplicative stack of 25 calibrated flat frames

4. calibrate lights with master bias and master flat

5. Registration and additive stacking of calibrated lights

 

Here (google drive) is a zip fail containing:

A single flat frame, a single light frame, Master flat, a single Calibrated light frame, Master light (calibrated)

Several aspects of this process strike me as odd.

1) With a CMOS camera a master bias is not necessarily suitable as a dark for flats. With CCD a bias does work fine. I would try making dedicated flat darks. (Same settings as the flats themselves.) Note: I would not, on a Newt, attempt to make any kind of bias or dark on the scope. Light leakage is hard to prevent.

2) I make a master bias or master flat dark and then apply that, at the stacking stage, to the flats - which means each flat is calibrated before it is stacked. I don't know how important this is but it works for me.

3) Why multiplicative stacking of flats? I, and everyone else I know, uses either average or median stacking for flats. The authors of AstroArt, in which I do my stacking, recommend average so that's what I've always done. I'm highly suspicious that this may be your problem.

Olly

Edited April 25, 2022 by ollypenrice

[Bibabutzemann]

@ONIKKINEN

I dont know if this has to do with sirils conversion into fit file, but in APT the histogram is at around 50% when i took these flats. But i will try out different exposure times next time

I took the flats when the scope was already indoor, but i didnt touch the focuser or camera. But next time i will take some flats mid session.

I will check for further light leakage. The view finder is already blocked.

The rotation in the master light comes imdeed from the fact that it was taken over two nights. I dont remember if i touched the camera or if this rotation comes from different polar allignment. Now i made another stack with only one night, but the problem remains the same.

Maybe you are right, and i chased a symtom here. As you showed so nicely with the colourful image, there is a dark area around the center, which could indicate over correction. i had the same dark area the center in my m101 image. But i cant help thinking , that some internal stray light leads to the complex gradient. Maybe its like alacante suggested, that the mirror clips doing some work here. But perfect flats should handle these gradients as well. Maybe my screen is too dependant on viewing angle, and therefore leads to over correction in the center area.

I will certainly try out different flat techniques next time.

Thanks for giving so much thought to my problem, you gave me lot of things to consider 😃 I will report back when i find something out!

Edited April 25, 2022 by Bibabutzemann

[alacant]

53 minutes ago, ollypenrice said:

I make a master bias or master flat dark and then apply that, at the stacking stage, to the flats

So does the OP. We simplify that even further, without using camera-produced bias at all. The argument against dark-flat frames and for fixed bias value was introduced by @lock042 -who knows about this stuff-  here. here

Certainly with our dslrs it works well and produces clean images.

11 hours ago, Bibabutzemann said:

my worklflow in Siril is like this:

Correct, and works for us. Have a look at removing a constant value as linked to above.

@Bibabutzemann if you haven't already done so, cut 10mm from the focuser tube.

Cheers

Edited April 25, 2022 by alacant
vínculo

[Bibabutzemann]

29 minutes ago, ollypenrice said:

Several aspects of this process strike me as odd.

1) With a CMOS camera a master bias is not necessarily suitable as a dark for flats. With CCD a bias does work fine. I would try making dedicated flat darks. (Same settings as the flats themselves.) Note: I would not, on a Newt, attempt to make any kind of bias or dark on the scope. Light leakage is hard to prevent.

2) I make a master bias or master flat dark and then apply that, at the stacking stage, to the flats - which means each flat is calibrated before it is stacked. I don't know how important this is but it works for me.

3) Why multiplicative stacking of flats? I, and everyone else I know, uses either average or median stacking for flats. The authors of AstroArt, in which I do my stacking, recommend average so that's what I've always done. I'm highly suspicious that this may be your problem.

Olly

Hi Olly, thanks for your input

1) the problem exists even if i dont do bias frames at all. I hardly notice any difference. To prevent light leaks i always take bias and dark frames with the camera not attached to my Newt. But i rarely take dark frames. I only find them helpful when i use the l extreme filter to avoid banding.

2) In my processing steps maybe i wasnt clear enough. Point two was calibrating every single flat frames with the master bias.

3) Here i should also clearify that im doing a median stack with the flats. But in Siril you can also choose a normalisation method. I choose multiplicative because i learned that in a tutorial. But even without normalisation or when i stack my images in DSS the problem remains. So i doubt that this is a processing specific issue.

5 minutes ago, alacant said:

So does the OP. We simplify that even further, without using camera-produced bias at all. The argument against dark-flat frames and for fixed bias value was introduced by @lock042 -who knows about this stuff-  here.

Certainly with our dslrs it works well and produces clean images.

Correct, and works for us. Have a look at removing a constant value as linked to above.

@Bibabutzemann if you haven't already done so, cut 10mm from the focuser tube.

Cheers

Yes, i should certainly should cut 10mm off! I noticed that even when the focus is all out, the tube still protrudes. I dont understand that design choice. Until now i didnt wanted to cut because i thought when i will buy the GPU CC i will need the whole tube. I assumed wrong there.

[ollypenrice]

11 minutes ago, alacant said:

So does the OP. We simplify that even further, without using camera-produced bias at all. The argument against dark-flat frames and for fixed bias value was introduced by @lock042 -who knows about this stuff-  here.

 

 

The link just takes me back to the top of this thread...

Olly

[ollypenrice]

4 minutes ago, Bibabutzemann said:

Hi Olly, thanks for your input

1) the problem exists even if i dont do bias frames at all. I hardly notice any difference. To prevent light leaks i always take bias and dark frames with the camera not attached to my Newt. But i rarely take dark frames. I only find them helpful when i use the l extreme filter to avoid banding.

2) In my processing steps maybe i wasnt clear enough. Point two was calibrating every single flat frames with the master bias.

3) Here i should also clearify that im doing a median stack with the flats. But in Siril you can also choose a normalisation method. I choose multiplicative because i learned that in a tutorial. But even without normalisation or when i stack my images in DSS the problem remains. So i doubt that this is a processing specific issue.

Yes, i should certainly should cut 10mm off! I noticed that even when the focus is all out, the tube still protrudes. I dont understand that design choice. Until now i didnt wanted to cut because i thought when i will buy the GPU CC i will need the whole tube. I assumed wrong there.

OK. I don't know how the background extraction works in Siril but, in Pixinsight, I find it far better to apply a minimum number of sampling points than to cover the background with them. The idea is to extract the large-scale gradients. I don't know if this would be relevant in this case.

Olly

[alacant]

23 minutes ago, alacant said:

The argument against dark-flat frames and for fixed bias value was introduced by @lock042 -who knows about this stuff-  here.

here

Apologies to those who attempted to follow the link.

[Bibabutzemann]

1 minute ago, ollypenrice said:

OK. I don't know how the background extraction works in Siril but, in Pixinsight, I find it far better to apply a minimum number of sampling points than to cover the background with them. The idea is to extract the large-scale gradients. I don't know if this would be relevant in this case.

Olly

Yes i also learned that in Siril it should work better with only a few points. But that only works if you assume a more or less linear light pollution gradient. But in my image there is a more complex gradient that combines with the light pollution, causing this mess

@ONIKKINEN oh i forgot to mention, i tried doing background extraction to every single frame. Problem areas still remained the same sadly...

[ollypenrice]

44 minutes ago, Bibabutzemann said:

Yes i also learned that in Siril it should work better with only a few points. But that only works if you assume a more or less linear light pollution gradient. But in my image there is a more complex gradient that combines with the light pollution, causing this mess

@ONIKKINEN oh i forgot to mention, i tried doing background extraction to every single frame. Problem areas still remained the same sadly...

Is this complex gradient present in a single sub exposure straight from the camera?

Olly

[alacant]

2 hours ago, Bibabutzemann said:

i should certainly should cut 10mm off

This is why the flat frame changes as you rack the focuser in and out. I'd do that first before diagnosing  further. 

The GPU works fine with the shortened tube. In fact the focus barrel coincides with the end of the cc.

Cheers

 

Edited April 25, 2022 by alacant

[alacant]

2 hours ago, Bibabutzemann said:

i tried doing background extraction

1200d? So before applying background extraction, you should first apply de-banding. BOTH whilst the image is still linear. Unless you have bad light pollution, you'll most likely find that you'll not need the extraction.

HTH

[ollypenrice]

7 hours ago, alacant said:

here

Apologies to those who attempted to follow the link.

The problem, for me, with threads like that is that the theorists - who understand the situation better than I do - do not agree with each other. My solution is to experiment and find out what works to my satisfaction. Sometimes my findings are at odds with what the theorists tell me I should find.  Since it isn't the theory I'll be hanging on the wall, I go with my findings. That doesn't mean I'm contemptuous of theory. That would be foolish in a scientific discipline. It just means that, in the absence of sufficient competence on my own part, I go with what works.

Olly

Edited April 25, 2022 by ollypenrice

[alacant]

24 minutes ago, ollypenrice said:

go with what works.

+1

We largely ignore the theory and post only that which hands on experience provides.

Interesting though it may be for some, we thrive not on theory, nor on so called 'experts', but pragmatism.

Experience also helps. For example, to date, the number of 130pds telescopes we've fixed is reaching double figures.

 

Edited April 25, 2022 by alacant

[Bibabutzemann]

maybe someone could provide me with 20 flat frames they have taken with the 130 pds and a similar size sensor. Then i could rule out if this gradient is only specific to my gear.

[alacant]

3 hours ago, Bibabutzemann said:

20 flat frames

Hi

Ask on the 130 thread?

We keep only stacked flat frames taken with modified pds'.

 

 

Edited April 25, 2022 by alacant