Jump to content

  •  

CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.

Photo

Pleiades...bright, and bringing out the microlens diffraction?

  • Please log in to reply
52 replies to this topic

#26 Gary Imm

Gary Imm

    Mariner 2

  • -----
  • Posts: 222
  • Joined: 11 Apr 2016
  • Loc: Onalaska, TX

Posted 08 November 2016 - 05:37 PM

Jon,

I agree that this shouldn't be a big issue. Pleiades and Gamma Cas are probably a few of the main offenders. Still, I am going to play with using lower gain settings and shorter sub times to see if I can minimize the problem. Also, for objects near bright stars like Gamma Cas, I am going to frame them to minimize the number of bright stars in the frame.

I really would like to understand the nature of the problem in more detail. Is it the nature of the CMOS chip, and is there a solution that ZWO could incorporate in the next version of the camera? I love the camera and would like to see it continue to improve.

#27 Jared

Jared

    Fly Me to the Moon

  • *****
  • Posts: 6073
  • Joined: 11 Oct 2005
  • Loc: Piedmont, California, U.S.

Posted 08 November 2016 - 06:33 PM

Well, the good thing is that stars this bright just aren't all that common.  You can get rid of the worst of the effects in Photoshop pretty easily.  This was roughly three minutes of work on one of your posted JPG's just by selecting the worst offenders, feathering the selection, applying a color noise reduction filter (twice at 100%), and then adding in a bit of saturation.  Most of the three minutes was actually using a history brush to re-paint the few dim stars that got messed up by the adjustment.  It's not perfect, but it gets rid of most of the rainbow affects.

 

KlJufFT.jpg


Edited by Jared, 08 November 2016 - 06:36 PM.

  • Jon Rista likes this

#28 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 16 November 2016 - 03:04 PM

There's another example here: http://www.cloudynig...round-big-stars

 

This time it's narrowband H-alpha taken with an ASI1600MM camera.  A very good example actually - the artefact is pretty clear.

 

Mark



#29 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 16 November 2016 - 07:35 PM

I finally found my Ghost of Cass data. This is what my Gamma Cass looks like:

 

Microlens Reflection.jpg

 

Don't exactly know what to make of it. Diffraction? Reflection? Combination of both? It's rather annoying, though, as there is really no way to fix that...



#30 Thirteen

Thirteen

    Gemini

  • *****
  • Posts: 3157
  • Joined: 12 Jul 2013
  • Loc: Milford, Michigan

Posted 16 November 2016 - 08:51 PM

That certainly looks like the reflection of the sensor bouncing between the sensor and either coverglass, or filter.   If you know the angle of the light cone and the distances to those surfaces you could actually calculate what its reflecting off of.   I don't think it looks like a diffraction effect in that image.   


  • sharkmelley likes this

#31 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 17 November 2016 - 01:18 AM

I finally found my Ghost of Cass data. This is what my Gamma Cass looks like:

 

 

 

Don't exactly know what to make of it. Diffraction? Reflection? Combination of both? It's rather annoying, though, as there is really no way to fix that...

That's the best example so far!

 

What we'll need later is the same image taken with the same equipment but using a different narrowband filter.  If the spacing of pattern changes with frequency then that would be an indicator of a diffraction effect.

 

Which scope did you use for this image?  Is that a full size crop or have you downscaled it?  This will enable us to start doing some calculations on the geometry.

 

Mark


  • jhayes_tucson likes this

#32 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 17 November 2016 - 01:27 AM

This is with my 600mm f/4 Canon lens. The crop is 100% native scale. Filter was Ha. Sensor to AR window distance 6.5mm. Sensor to filter distance ~14mm. 

 

This was a stack of 93 frames.


Edited by Jon Rista, 17 November 2016 - 01:28 AM.


#33 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 17 November 2016 - 01:34 AM

Would it be better if I centered the star and acquired Ha, SII and OIII data? This is with the star in the fairly extreme corner of the frame...and the quality there is not great. There are a multitude of additional diffraction effects that seem to show up in the corner that don't in the center, although the background grid pattern does seem to. Plus, this sequence of subs was totally undithered, and properly dithered subs might give us a clearer picture of the pattern(s). 



#34 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 17 November 2016 - 01:39 AM

Would it be better if I centered the star and acquired Ha, SII and OIII data?

Yes, dithered images of a centred star using those filters would be perfect!

 

Mark



#35 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 17 November 2016 - 01:47 AM

Alright. Next clear night I'll grab some data. 


  • sharkmelley likes this

#36 jhayes_tucson

jhayes_tucson

    Fly Me to the Moon

  • *****
  • Posts: 7180
  • Joined: 26 Aug 2012
  • Loc: Bend, OR

Posted 17 November 2016 - 11:36 AM

Jon,

I think that Mark is spot on.  The Gamma Cass image sure looks to me like a stray secondary reflection from the micro-lens array but the way to tell is to look at it in different wavelengths.  The scale will change if it's due to diffraction rather than reflections.  You may have a couple of things at work here.  The colors that you see around the bright stars may also be due to the fact that the sensor is at (or near saturation) and you my be running into some non-linearity in the really bright regions.  That could produce variations in color around the star--due to slight variations in signal between the different filters.  In that case, the problem is easily handled with some minor masked processing (as I'm sure you know.)

John


Edited by jhayes_tucson, 17 November 2016 - 11:38 AM.

  • Jon Rista likes this

#37 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 17 November 2016 - 11:42 AM

I do wonder, if it is primarily reflections, if that could be corrected somehow. I really do want to image the whole region around Ghost of Cass, as I am aware there is quite a bit of faint hydrogen around there. However this artifact is pretty egregious! :p Very much non-aesthetic for a "pretty picture".



#38 jhayes_tucson

jhayes_tucson

    Fly Me to the Moon

  • *****
  • Posts: 7180
  • Joined: 26 Aug 2012
  • Loc: Bend, OR

Posted 17 November 2016 - 12:40 PM

Unfortunately, the best way to handle this kind of problem is with high quality AR coatings.  There are some other ways to handle strays with polarization tricks (e.g. optical isolators) but I none of them will work in this kind of situation.  I think that you may be stuck with this as an artifact inherent in the camera.  There may be a way to model it and process it out; but, I suspect that you may be better off giving it up and working on faint features further away from really bright stars.

 

John



#39 freestar8n

freestar8n

    Vendor - MetaGuide

  • *****
  • Vendors
  • Posts: 8893
  • Joined: 12 Oct 2007

Posted 17 November 2016 - 03:48 PM

Don't exactly know what to make of it. Diffraction? Reflection? Combination of both? It's rather annoying, though, as there is really no way to fix that...


Diffraction off the microlens array moving up away from the sensor and hitting the sensor window and returning back to form a diffraction pattern on the sensor.

It should show a wavelength dependence in the pattern, and some wavelengths may create stronger reflections than others.

It would be very hard to imagine some way that a magnified image of the actual sensor pattern could be created and imaged back onto the sensor - so I don't see any way it could be an actual image. But a diffraction pattern is easily explained and is a good match with simulations for other similar patterns as shown in the thread I mentioned above.

For a square pattern sensor array the diffraction pattern has the same layout that an image would - but they are very different since if it were an image, each spot would have a 1-1 correspondence with one of the microlenses - and if one had a defect on it you would see it in that one spot. But if it is diffraction each spot is due to all the microlenses being illuminated by the star summing together in phase - which is very different.

Frank

#40 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 18 November 2016 - 01:26 AM

I think what we may be seeing in Jon's example is the following (excuse my crude attempt at drawing with the mouse):

 

JonDiffractionPattern.jpg

 

Essentially it is a set of overlapping discs.  Each disc is an image of the lens aperture created by diffraction off the microlens array - just like this example with the KAF-8300 sensor:

https://stargazerslo...xt-fw-disaster/

 

The geometry of the rectangular grid of disc centres is determined by the microlens spacing (3.8 microns), the distance to the sensor cover glass (which reflects the pattern back onto the sensor) and the wavelength of H-alpha light.  Given the distance to the cover glass, the diameter of each disc is determined by the F-ratio of the lens - in this case F/4.  Depending on the F-ratio the discs may or may not overlap - they will overlap for fast lenses but not for slow ones.

 

There may also be a larger disc, marked in blue.  This could be generated by diffraction off the microlens array followed by reflection off the camera cover glass - hence the larger size.  There may also be effects caused by potential overlapping large blue discs (not shown).

 

A bit more work is needed to confirm this (or refute it) with a precise mathematical analysis of the geometry. I should get time over the weekend but feel free to jump in and beat me to it!  This particular example may or may not be slightly more complicated because the star is offset from the centre of the image.

 

Mark


Edited by sharkmelley, 18 November 2016 - 01:39 AM.

  • happylimpet and Jon Rista like this

#41 happylimpet

happylimpet

    Skylab

  • *****
  • Posts: 4037
  • Joined: 29 Sep 2013
  • Loc: Southampton, UK

Posted 18 November 2016 - 05:24 AM

Should be working but....using

 

d.sin(theta)=n.lambda

 

for a diffraction grating, and using the first order so n=1, lambda=656e-9,and d=pixel size=3.8e-6,gives

 

sin(theta)=lambda/3.8E-6 =0.172 or theta~10degrees.

 

then trigonometry using sin theta =opposite/hypotenuse assuming theta small so hypotenuse=adjacent, and about 60 pixels separation between orders (as marked above)

 

gives a total distance for the diffracted ray (ie twice the distance from the chip to the reflective surface) of

 

x = 3.8e-6 x 60 / 0.172 = 1.3mm

 

so the reflective surface would be 0.65mm above the chip!

 

Hmmm maybe Ive made a mistake - or is there a protective glass surface 0.65mm above the chip?

 

Looking at it perhaps the 60 pixels is an underestimate, or has it been resampled for cloudynights? If an underestimate, this would make the reflective surface further from the chip.


Edited by happylimpet, 18 November 2016 - 05:31 AM.


#42 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 18 November 2016 - 06:54 AM

0.65mm sounds quite plausible for the sensor cover glass.

Mark

#43 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 18 November 2016 - 11:33 AM

Yeah, that sounds like the sensor cover glass. Very interesting. When I had my sensor compartment open the last time, it did not appear as though the cover glass had any AR coatings of any kind. If we can confirm that it is the cover glass, I can send Sam some feedback and I'm sure ZWO will find a way to fix the issue in future cameras. 



#44 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 18 November 2016 - 08:01 PM

Well, I was just starting to take some subs for this...and my transparency took a dive. Seems I have a whole new weather system forming directly over my house now. O_o



#45 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 18 November 2016 - 09:45 PM

What a pain - the weather :(

 

I've played around with diffraction calculations on your image, assuming the following:

H-alpha wavelength 0.65 microns

Microlens spacing 3.8 microns for the ASI1600

Aperture F/4

If I then assume the cover glass distance to be 0.45mm above the sensor (found by experiment) the discs line up as follows:

 

JonDiffraction1.gif

 

I've only included the discs that line up with the main features.

 

If I include the next set of discs, they don't line up so well:

 

JonDiffraction2.gif

 

So I think I'm on the right track but the fact that your star is not central in the image is possibly leading to some asymmetry in the outer regions of the pattern.  If I've correctly identified the mechanism here, then at a slower aperture F/5.6 the discs stop overlapping at the H-alpha wavelength.  However with OIII at F/4 the overlap would be greater.

 

Mark


Edited by sharkmelley, 18 November 2016 - 09:49 PM.


#46 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 18 November 2016 - 10:25 PM

I did manage to get 16 RGB subs, and 8 Ha subs. I'll integrate them and get crops of each channel for you as soon as I can.



#47 DesertRat

DesertRat

    Fly Me to the Moon

  • *****
  • Posts: 6266
  • Joined: 18 Jun 2006
  • Loc: Valley of the Sun

Posted 18 November 2016 - 11:21 PM

Interesting info Jon and impressive analysis by Mark!

 

Jon,

I can only guess you would rather image than conduct another physics experiment. :lol:

 

But it would be useful to see what effect stopping down to f/8 or 11 would have on this phenomena.  If a microlens diffraction effect it should become more concentrated.  Possibly a little defocus might provide additional information.

 

Glenn



#48 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 18 November 2016 - 11:56 PM

I'll see what I can do next time it is clear. I want to stick with the same star so we can have consistent data to work with, and it is not in a position I can image it for very long at the beginning of the night. But i'll get as much data as I can over the next few clear nights.



#49 sharkmelley

sharkmelley

    Vanguard

  • *****
  • Posts: 2403
  • Joined: 19 Feb 2013

Posted 29 December 2016 - 04:40 AM

The first post in this thread contains an excellent example of the effect.

 

http://www.cloudynig...-or-tilt-issues

 

Mark



#50 Jon Rista

Jon Rista

    ISS

  • *****
  • topic starter
  • Posts: 23803
  • Joined: 10 Jan 2014
  • Loc: Colorado

Posted 29 December 2016 - 01:24 PM

Mark, I actually gathered a bunch of LRGB subs, and I think Ha subs, for this. I then promptly forgot about them. :p I'll get it all uploaded soon here for you.




CNers have asked about a donation box for Cloudy Nights over the years, so here you go. Donation is not required by any means, so please enjoy your stay.


Recent Topics






Cloudy Nights LLC
Cloudy Nights Sponsor: Astronomics