Subs after Meridian Flip - Flats don't work properly

[kirkster501]

I am really baffled folks and could use your help.   I thought flats taken either side of the peer would work fine.  They normally do...  There is absolutely NO rotation of the camera taken place (TEC140 with Atik 460).  Flats taken at same CCD temperature, same focus position.  I tried different ADU for the flats and no difference.  Retaken lots of different flats and still same result.

This three hour data set in luminance is lost if I can't figure this out

A sub (east side) of the peer before the flip and my flats calibrate out dust donuts.  As it happens, flats ALSO taken East side of peer (facing my wall mounted flats panel).

 

But the subs AFTER the flip (with the same master flat as before the flip) and the flats are leaving strange residual donuts that are not calibrating out.  Vignetting removed but weird artefacts of the dust donuts are left.  To repeat no rotation of the camera has taken place at all.

 

Leaving me with a messed up result with 16 x 10 Minute subs.

[tooth_dr]

Do you manually rotate your lights prior to stacking?

[kirkster501]

Nope, I leave it all to Pixinsight to sort out.

So you mean rotate the lights before calibration?

[kirkster501]

If I play back the two subs quickly back and forwards, one before the flip  and the other one afterwards, the two bunnies on the upper right seem to move.  But they could not have done, the camera never moved other than flip with the telescope!

[tooth_dr]

44 minutes ago, kirkster501 said:

Nope, I leave it all to Pixinsight to sort out.

So you mean rotate the lights before calibration?

No you dont want to do anythin with them prior to that.

[alexbb]

Perhaps you have some floppy dust bunnies bending the other way under gravity pull

[Laurin Dave]

Hi Steve...  very strange, somethings moving,  I'd work out how close to the sensor they are using CCD Ware dust donut calculator (my rough guess is 7mm which puts them on the sensor protective glass so maybe that's come loose)

Dave

Edited April 25, 2019 by Laurin Dave

[ollypenrice]

DON'T SCRAP THE DATA!!! ? I'll come to that later...

As Dave says, the dust could have moved slightly during the flip. Not impossible. However, it could be that the software is rotating the images before applying the flat to them, in which case the flat will be out by 180. See what happens if you make a second master flat and rotate it 180, duly labelled. Then stack just the post flip images trying both the rotated and un-rotated flats. If one of the two stacks comes out OK you know it's a stacking software error.

If you cannot solve the problem this way, try the following bodge (which is the kind of thing I do regularly, though keep that to yourself! )

Make a full stack - with artefacts - and a stack from before the flip. Align these in whatever you use for that. Registar in my case.  Then, in Photoshop (AKA civilization) give both images a gentle log stretch as near as possible identical and bring in the black points to the same point. Check that the background sky values are the same in both images. Paste the full set with artefacts onto the-pre flip set without. Choose a soft edged eraser of the right size and run it over the artefacts. With a small eraser you could also erase that trail. In the unlikely event that the difference in noise shows up, the bottom layer being noisier and showing through where the top has been erased, don't flatten the layers yet. Make the bottom layer active, both layers visible, and run 'Reduce Noise' on the bottom layer. Then go to Edit, Fade Reduce Noise and fade it till the noise looks the same in the repair as elsewhere. All of this takes longer to explain than to do. 

It looks like nice data and will work perfectly well, I'm sure.

Olly

 

[kirkster501]

Thanks folks.  I am going to give the camera a check and get the dust blower on it.

Thanks for the tips Olly, I will have a look into that.

[tomato]

Steve, sorry but I cannot add anything to the tips given, other than to say they certainly are weird looking dust bunnies with that 3D effect.

Olly, your toolbox of Photoshop fixes for almost any AP issue never ceases to impress, you will post one on how to remove clouds one of these days.

[alan potts]

8 hours ago, tomato said:

Steve, sorry but I cannot add anything to the tips given, other than to say they certainly are weird looking dust bunnies with that 3D effect.

Olly, your toolbox of Photoshop fixes for almost any AP issue never ceases to impress, you will post one on how to remove clouds one of these days.

There already is a fix for that but to date it only works on a photo.

Alan

[Tomatobro]

two drops of condensation moving due to gravity on the flip? I don't know but just throwing out ideas

[Laurin Dave]

Hi Steve... did you figure out what was causing the embossing?

Dave

[ollypenrice]

On 24/04/2019 at 22:33, tomato said:

Steve, sorry but I cannot add anything to the tips given, other than to say they certainly are weird looking dust bunnies with that 3D effect.

Olly, your toolbox of Photoshop fixes for almost any AP issue never ceases to impress, you will post one on how to remove clouds one of these days.

Heh heh, nowt I can do about clouds, viz a night wasted last night, only discovered this morning! It looked OK at the time... (Having said that, my tiny mind is wondering if I could use haze-dimmed data to control  the starfield... We'll see.)

The great thing about Photoshop is that, once you know its generic tools, you can use them to make astro-specific tools just by thinking about what it is you want to do. I want to do this... I have these tools...  aha, yes!

Olly

[kirkster501]

Nope, I never got to the bottom of this.  I appreciate the thoughts but I don't think it can be dust bunnies in motion.  I could accept that theory if there were one such bunny but there are lots of these residual ring artefacts that you can see if you look at the last picture I posted.  I have not had enough clear sky time to be able to do further testing other than grab five Coma Cluster subs.  The only thing I can think of is there maybe could have been a minute rotation of the camera as the scope flipped.  I cannot see how, since everything seemed to be tight enough but no other logic can explain these strange artefacts.  I have taken everything off - all adapters etc - and tightened everything up.  I also cleaned up the sensor window.

I will have to use Olly's workaround to preserve this data set.

Here is a five minute  raw sub after the clean up:

[Laurin Dave]

Hi Steve...  came across this recent thread on Cloudy Nights..  uneven illumination of the OPs flats panel causing a very similar embossed dust bunny to those you had... 

Dave

  https://www.cloudynights.com/topic/661557-flats-not-correcting/