Author Topic: Is this a gradient?  (Read 3977 times)

0 Members and 1 Guest are viewing this topic.

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Is this a gradient?
« on: Nov 13, 2013, 08:51:14 »
Morning,

I was advised to add a spacer to the optical path to address elongated stars and I now get a whiter image than I'd normally get, is this is a gradient?



This a stack of 7 x 360secs Blue with the darks frames removed.

The Thing

  • Mr October
  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 2826
  • The OAS International Branch
Re: Is this a gradient?
« Reply #1 on: Nov 13, 2013, 09:46:19 »
Looks like vignetting to me. Have you used flats as well as darks in this stack?
¯\_(ツ)_/¯ (Gallic shrug :))
My Website AstroManche

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #2 on: Nov 13, 2013, 09:59:43 »
I haven't produced the flats yet, that's tonight's job.

JohnP

  • Non-OAS
  • Galaxy Class
  • *
  • Posts: 1633
    • JP Astro Images
Re: Is this a gradient?
« Reply #3 on: Nov 13, 2013, 10:55:29 »
Like Duncan said I am sure flats will get rid of the gradient. Image looks very promising by the way - John

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #4 on: Nov 13, 2013, 12:26:34 »
Thanks, I've not had this problem before it's all a bit new to me.

So on the this basis has the extension tube caused the vignetting?
Is it too long?
How do I calculate the correct length?
Is elongated stars in the corners the lesser of two evils?
« Last Edit: Nov 13, 2013, 12:43:06 by Ivor »

JohnP

  • Non-OAS
  • Galaxy Class
  • *
  • Posts: 1633
    • JP Astro Images
Re: Is this a gradient?
« Reply #5 on: Nov 13, 2013, 12:33:31 »
most astronomical processing progs have scripts/ procedures for removing gradients even if the flat doesn't. I would rather deal with gradients than 'dodgy stars'... John

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #6 on: Nov 13, 2013, 15:06:44 »
You definitely need the correct CCD spacing to the reducer/flattener else you will get dodgy stars.  I'll be surprised if that causes a noticeable change in vignetting.

RobertM

  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 4298
Re: Is this a gradient?
« Reply #7 on: Nov 13, 2013, 18:43:27 »
It's difficult to tell.  Can you post a single stretched but otherwise unprocessed sub?  That should give a good indication of  the source i.e. Whether it just needs flats and/or LP gradient removal.

Robert

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #8 on: Nov 13, 2013, 22:27:55 »
Quote
You definitely need the correct CCD spacing to the reducer/flattener else you will get dodgy stars.  I'll be surprised if that causes a noticeable change in vignetting.
e

The spacer was 254mm which I thought was quite large but I'm not sure how to calculate the correct distance and I never got a response from anyone on the SBIG or WO forums.

I've produced some flats tonight I think they're OK, it's first time with the reconfigured light box.

Below is the flat, histogram, file stats do they look about right?

 



I've also attached a single stretched sub



Thoughts?

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #9 on: Nov 13, 2013, 23:02:53 »

Spacing of 254mm?  10 inches?

Remind us of exactly what configuration you have and we might be able to help.

Mark

Carole

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 8342
    • Carole's images
Re: Is this a gradient?
« Reply #10 on: Nov 14, 2013, 09:31:49 »
Is that single stretched sub with or without flat applied?

Carole

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #11 on: Nov 14, 2013, 09:40:09 »
Quote
Spacing of 254mm?  10 inches?

A good spot a missing decimal point 25.4mm (I'll double check the length as well)

My kit is as follows:

SBIG ST8300M
CCD Kodak KAF-8300
Pixel Array 3326 x 2504
CCD Size 17.96 x 13.52 mm
Total Pixels 8.3 Million
Full Well Capacity ~25,500 e-
Dark Current 0.002e-/p/s @ -10C
Antiblooming 1000X
Shutter Even-Illumination,
Mechanical Exposure 0.09 to 3600 seconds
A/D Converter 16 bit
Gain 0.37e-/ADU
Read Noise ~9.3e-
Backfocus 0.69 inches (17.5 mm)

SBIG FW8-8300
Holds eight 36mm drop in filters for un-vignetted imaging at any f/ratio with the ST-8300 or ST-4000XM cameras.
Small footprint: 6.3 x 8.1 inches.
Thin design - adds only 20mm backfocus (Total backfocus for ST-8300 with CFW8-8300 is 38mm).

William Optics Focal Reducer & Flattener IV
Lens-to-chip distance : 66-86mm (currently set at 74mm ish)
Power : 0.75x ~ 0.8x
Rotating T-ring adapter
Clear Aperture : 50mm
L 84mm : W 77mm : H 123mm

William Optics FLT110
Aperture  110mm 
Focal Ratio  F/7
Focal Length  770mm
Resolving Power  1.05"
Limiting Magnitude  12.0
Tube Diameter  115mm

I’ve realised after posting this last night that the flat I’ve posted is without the extension tube so I’ll reproduce them tonight and it will make a useful comparison. Do the stats associated with the flat look about right?

Quote
Is that single stretched sub with or without flat applied?

Without the flat.

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #12 on: Nov 14, 2013, 11:10:47 »
Now I understand.  I hadn't come across an adjustable flattener before, where you "dial-in" a different number depending on scope.  So your extension tube is effectively to make your camera CCD to flattener distance the same as if you were using a DSLR with that flattener.

A useful discussion here:
http://stargazerslounge.com/topic/135422-a-question-on-wo-field-flattener-iv-spacing/

Mark

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #13 on: Nov 14, 2013, 22:48:26 »
Right the results of testing tonight speak for themselves, the new flat and the reprocessed blue channel are below.





So hopefully I'll be able to save the data from this now.

William Optics designed the whole setup based around using a canon so I didn't experience any vignette problems (pre ST8300M) before I just thought the flats addressed the dust mites. 

Quote
A useful discussion here:
http://stargazerslounge.com/topic/135422-a-question-on-wo-field-flattener-iv-spacing/

Thanks Mark great link, I've been searching for the camera and telescope I should have searched for the flattener as well.
 
Chris the author pf the thread on SGL, has done quite a lot of research on his site which was very helpful, based on his calculation I need a spacer of 23.9mm, my spacer is 250mm edge to edge



but this doesn't seem right surely I should be measuring the narrowest gap between the two threads?

To get the measurement completely correct on his kit he measured the curvature of the image
Quote
http://www.toinfinityandbeyond.me.uk/astronomy/Blog/Entries/2012/2/24_Distance_setting.html
does anyone recognise the tool he used to do this? I've sent him a message on SGL but but in case he doesn't reply I be interested in knowing how to work this out.

 

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #14 on: Nov 15, 2013, 06:57:05 »
The tool is CCDInspector.  It measures the FWHM of every star in the image and plots a chart.  The curvature is the max FWHM over min FWHM as a percentage or something similar.  It is not a very stable or meaningful quantity since if you defocus an image slightly then the reported curvature reduces.  However the chart is useful for comparing one frame to another after making an adjustment (at least when it is working well, which often it doesn't).

When measuring extension tubes you need to measure the increase in the light path which is not the overall length of the device since the threads screw into something - giving overlap.

mickw

  • Exile Under Darker Skies
  • R.I.P.
  • Galaxy Cluster
  • *
  • Posts: 5683
Re: Is this a gradient?
« Reply #15 on: Nov 15, 2013, 07:02:13 »
Quote
but this doesn't seem right surely I should be measuring the narrowest gap between the two threads?

You should be measuring between the mating surfaces, so do not include the male thread beyond the knurled ring.
That would make the measurement about 4mm lower - 246mm (or 21mm if you fancy using a decimal point  ;) )

I believe flatteners do have a +/- 5mm leeway in their use - I have been known to be wrong.

Don't forget you need to include any other adapters/rings between the CCD and the flattener when calculating (T ring adapter, filter wheel etc.)
Growing Old is mandatory - Growing Up is optional

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #16 on: Nov 15, 2013, 07:20:05 »
Quote from: mickw
I believe flatteners do have a +/- 5mm leeway in their use - I have been known to be wrong.

More like 0.5mm!  I guess that was a typo ;-)

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #17 on: Nov 15, 2013, 07:23:30 »
One other shortcoming with CCDInspector is that it can report a totally flat field even when the stars towards the corners are totally squiffy in shape.  That's the problem with relying on a single statistic - in this case FWHM.

mickw

  • Exile Under Darker Skies
  • R.I.P.
  • Galaxy Cluster
  • *
  • Posts: 5683
Re: Is this a gradient?
« Reply #18 on: Nov 15, 2013, 07:33:09 »
Quote from: mickw
I believe flatteners do have a +/- 5mm leeway in their use - I have been known to be wrong.

More like 0.5mm!  I guess that was a typo ;-)

Oops - stupid decimal points  :oops:
Growing Old is mandatory - Growing Up is optional

RobertM

  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 4298
Re: Is this a gradient?
« Reply #19 on: Nov 15, 2013, 09:34:43 »
The stars look ok give or take a knats so I don't think it's spacing.  I'd be more concerned that the centre of the flat is offset from the centre of the camera.  This usually means that something is not square to the optical axis.  You can also see that there is some strong vignetting appearing on the left  top/bottom corners of that flat.  There could be a number of reasons for this such as scope collimation, focuser sag or something impinging in the light path.

Do the flats look ok without the reducer ?

Robert

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #20 on: Nov 15, 2013, 10:47:25 »
This is feeling like a eureka moment, I've been really struggling with my kit since getting the ST8300 at least I'm understanding why now.

The CCD to FF gap for the FLT 110 is 73.5mm and the FF adjustment gives me a range between 66 - 86mm.

The back focus of the various bits of kit is as follows:

Camera + Filter wheel 38mm
Extension tube 20mm
Gap between FF lens and housing 10.6mm

This equals 68.6mm, which means I need to add 4.8mm to the FF gauge (78.4mm) to make the true distance 73.5mm.

The adjustable FF is great idea but the lack of documentation makes it very confusing.

I’m not hearing a lot of love for CCDINSPECTOR is there an alternative?

Quote
The stars look ok give or take a knats so I don't think it's spacing.  I'd be more concerned that the centre of the flat is offset from the centre of the camera.  This usually means that something is not square to the optical axis.  You can also see that there is some strong vignetting appearing on the left  top/bottom corners of that flat.  There could be a number of reasons for this such as scope collimation, focuser sag or something impinging in the light path.

Do the flats look ok without the reducer ?

I have refractor so there’s no collimation. I’d put the issues in the corners down to the subs being taken on two nights but now you comment on it on the flat, I’ll check everything is square tonight as part of the test run.

I can’t use the the ST8300 without the FF as there isn’t enough back focus on the focuser, and I don’t know how to calculate the extension tube required.


MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #21 on: Nov 15, 2013, 11:31:16 »
I read the SGL thread completely differently.
 
I understood it like this:
1) 73.5mm is the number you "dial in" as the FF adjustment - it differs from scope to scope.
2) The FF to CCD distance should then be the standard DSLR distance of 55mm - this is the same for every scope. So your 68.6 is already too large.

Mark

RobertM

  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 4298
Re: Is this a gradient?
« Reply #22 on: Nov 15, 2013, 12:27:45 »
I have refractor so there’s no collimation.

That's incorrect although WO might be one of the manufacturers who think they don't need to be collimatable so leave out the adjusters to reduce cost.  Most decent manufacturers include collimatable lens cells.

I still think something is off somewhere as the flat centre is not in the middle however the extra spacing that Mark has identified would give you much more reduction.  That will also have the effect of amplifying any vignetting and would explain the two darker left corners.  Once you have the correct reduced distance then those corner issues will go away and you might not notice the slight off centreing.

An alternative way of getting the FR distance correct - If you know exactly what the new focal length is i.e. 0.8 * 770 = 616mm then you can use plate solving to work out the focal length from a captured image then adjust the focal reducer distance to get a solved image f/l of 616mm (or the equivalent plate scale of 1.81 arcsec/pixel).

HTH
Robert


MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #23 on: Nov 15, 2013, 12:46:25 »
This diagram confirms how I understood things:
http://www.williamoptics.com/accessories/images/FLAT4_Layout.jpg

If your CCD is 55mm from the rear shoulder of the FF then you will obtain the necessary 73.5mm CCD spacing from the rear lens element when you have dialled in the 73.5mm distance.

Alternatively you could "dial in" 66mm which gives you an extra 7.5mm (73.5-66) which you could add onto the 55mm giving 62.5mm.  This would preserve the correct 73.5mm distance from CCD to rear lens element.  But your 68.6mm is still too great even so.

So reduce your 20mm extension and then you'll have sufficent adjustment to make it all work.  I'll let you do the maths!

Mark

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #24 on: Nov 15, 2013, 15:47:42 »
Mark, I think we are saying the same thing but I've not articulated it well enough. Your post also made me realise I’d made a mistake in my calculation by excluding a double negative.
The 68.6mm I stated includes the gap between the lens and the front shoulder of the FF and my camera, the camera plus extension tube in my case equals 58mm.

I made this more complicated than it needed to be by working out the inner distances in the FF rather than basing my calculations off the FF rear shoulder (which WO strangely don’t do in their diagram)

I’ve also realised the 10.6mm lens to FF gap stated in the SGL post is wrong it should be 11mm. If I correct the double negative I included in my first calculation I need to adjust the FF to 70.5mm.I anticipate some minor adjustment from this to cater for the filter. I added the measurements to the WO diagram.



It will be interesting to see if this adjustment removes the problems on the left side on the flat as well.

Cheers

Ivor

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #25 on: Nov 15, 2013, 21:07:53 »
Looks like you've got it right now ;-)

Carole

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 8342
    • Carole's images
Re: Is this a gradient?
« Reply #26 on: Nov 16, 2013, 20:45:41 »
I find this a real PITA especially with 3 cameras, and 4 telescopes, a flattener or reducer and sometimes a filter wheel so every time I change my set up I have to re-calculate.  I have made a chart.  Normally I put all the pieces together and then measure as this takes into account the thread that gets "screwed in". 

Even when I have the exact measurement recommended I still get elongated stars in the corners sometimes.  I guess swapping cameras and scopes around doesn't help much and one day I'll get it all right hopefully. 

Also I find my images are very slightly offset and this probably
Quote
means that something is not square to the optical axis
which might explain the elongated stars in one corner only sometimes, but I can't think how I can get anything more orthogonal, I can't put the camera in crooked......I feel it's to do with only having 2 screws in the draw tube to tighten the camera on, 3 I am sure would rectify this, so not sure what I can do about it.   Maybe if those 2 screws were at 120 degrees to the draw tube LOCK then this could compensate, but I don't think they are.

Am reading this thread with interest.

Carole


Rocket Pooch

  • Non-OAS
  • Galaxy Class
  • *
  • Posts: 5027
Re: Is this a gradient?
« Reply #27 on: Nov 16, 2013, 21:13:46 »
What you doing Pm?

RobertM

  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 4298
Re: Is this a gradient?
« Reply #28 on: Nov 17, 2013, 12:30:24 »
What you doing Pm?


She's obviously on thread jacking duty again :-?

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #29 on: Nov 17, 2013, 22:39:33 »
Quote from: Ivor
I’m not hearing a lot of love for CCDINSPECTOR is there an alternative?

Not really, so I ended up writing my own.  A prototype exists. The principle is to measure the elongations of stars by assuming they have an elliptical gaussian shape (a big assumption because some optical aberrations give star shapes much more "squiffy" than an ellipse), determining the direction of the major and minor axes of that star's ellipse and then calculating the FHWM in the major and minor axis directions (not the x and y directions as other software often does) and hence determining the elongation.  It is difficult to do this reliably because of various random effects so it only becomes statistically significant when averaged over many stars in an area.
So I do this in the extreme corners of the image, the middles of the edges and the image centre.  It gave a plot like this:



On each detected non-saturated star a "windrose" is plotted to show the direction of elongation and then a larger windrose is plotted to show the average over all stars in that neighbourhood.  Statistics are generated for each rectangle showing FHWM of the minor axis (in pixels) elongation (in pixels) direction of elongation and the number of stars used in the calculation (the more stars, the more reliable the measurement).

Conclusion.  I found this gave pretty reliable results on the whole for the particular aberrations I was seeing but to be honest, no better than simply "eyeballing" the image!  I suppose it helped by confirming with statistics what was already obvious from looking at  the image.

A lot more development work would be required to turn this into a genuinely useful and simple to use tool - I just don't have time to progress it further.

One thing that would make it more useful would be to "subtract" the elongation found in the centre of the image from the elongations found elsewhere (this would be done in a vector arithmetic manner), since the centre elongations are generally caused by guiding issues rather than optical issues and it would be useful to separate out these two effects.

Mark
« Last Edit: Nov 18, 2013, 05:04:39 by MarkS »

RobertM

  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 4298
Re: Is this a gradient?
« Reply #30 on: Nov 18, 2013, 14:25:39 »
It's already a very useful tool !  The only issue I have is that my copy doesn't release memory so dies after about 40 or so images.  Yes the are things that can be done to make it simpler but it already does a really good job.

Robert

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #31 on: Nov 18, 2013, 22:38:00 »
A weekend away delayed the new flat capture so here it is with the resulted RGB channels merged in PS and slightly darken. A slight gradient remains bottom left to upper right, nothing significant though.






The new flat has clearly has made a big improvement in the resulting data, so I'll do the CCDinspector test tomorrow as the weather is promising.

Is CCDinspector straightforward to use? the free version only has 30 days so I want to know if I need to research its use.

Secondary issue why does the image have no colour, but that's should really be a different thread.


Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #32 on: Nov 20, 2013, 12:30:12 »
I ran CCDInspector three times last night setting the FF with a .5mm adjustment either side of the calculated 70.5mm. I think I might of made a mistake in my workflow as I didn't refocus after each adjustment but I'm not sure it's particularly relevant based on the results.

FF set to 70mm



FF set to 70.5mm




FF set to 71mm




So 70.5mm appears to be a good position, but are the stats good and is there anything I can do to improve them?

Rocket Pooch

  • Non-OAS
  • Galaxy Class
  • *
  • Posts: 5027
Re: Is this a gradient?
« Reply #33 on: Nov 20, 2013, 15:30:19 »
Is the focuser not square to the focal plane?

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #34 on: Nov 20, 2013, 16:51:31 »
It appears to be, there's no flex in the system either. What makes you think there is?

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #35 on: Nov 20, 2013, 17:28:43 »
There's a diagonal gradient in CCDInspector.
If the CCD were orthogonal it should be a circularly symmetric gradient.

RobertM

  • O. A. S.
  • Galaxy Class
  • *
  • Posts: 4298
Re: Is this a gradient?
« Reply #36 on: Nov 20, 2013, 18:43:57 »
As I mentioned before.  The stars didn't look bad so it was never going to be the reducer spacing causing the vignetting.

If the focuser isn't square then taking flats with it a various positions should show vignetting getting better or worse.  If that isn't the cause then is the corrector/camera fixed in squarely ?  The last thing is lens cell collimation but I hope it doesn't come to that.

Robert

Ivor

  • Past Chairman
  • O. A. S.
  • Star Class
  • *
  • Posts: 595
Re: Is this a gradient?
« Reply #37 on: Nov 20, 2013, 21:05:54 »
It's first time I've used CCDInspector so I'm not sure what to expect or deduce. Looking at the 70.5mm curvature map I see a red circle at NNE being bisected by the inner circle, looking at 71mm the red circle is SES and touching the edge of the frame. Based on the help with CCDInspector the colour mapping is as follows:

Black:    lowest FWHM
Blue:     slightly defocused
Green:   more defocused
Red:      highest defocus

As I'd expect the centre of the image to be the most focused I'm now at a loss at what to read into this.

What are acceptable values? Assuming my flats and subs will be better with FF at the correct setting how much will this matter?

MarkS

  • O. A. S.
  • Galaxy Cluster
  • *
  • Posts: 7303
Re: Is this a gradient?
« Reply #38 on: Nov 20, 2013, 23:56:18 »
Quote from: Ivor
What are acceptable values?

If you find out then plese let me know, since I've never been able to work what they mean!

Mark