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

Planetary Camera Rant - 1.1 micron pixels

astrophotography imaging planet ccd
This topic has been archived. This means that you cannot reply to this topic.
61 replies to this topic

#1 RickV

RickV

    Vanguard

  • -----
  • topic starter
  • Posts: 2,474
  • Joined: 17 May 2013

Posted 02 February 2016 - 05:59 PM

When I changed from a solar-system-imaging camera with 3.75 micron pixels to one with 2.2 micron pixels, my images of Jupiter and Saturn improved dramatically.  Why?  Simple - more pixels forming the image.

 

Today, sensitive, low-noise Back Side Illuminated (BSI)sensors available with 1.1 micron pixels and are used in many cell phones - as the secondary (selfie) camera.  When are we going to see these high-resolution BSI sensors used in planetary video cameras?  What are the planetary camera manufacturers such as Celestron, Orion, PointGrey, MallinCam, etc. waiting for?

 

With my Celstron C11 at native F/10, changing from 3.75 to 2.2 micron pixels caused my images of Jupiter's diameter to increase from 176 pixels to 300 pixels. That's an increase in magnification equivalent to adding a 1.7X Barlow... without a Barlow limitations.

 

3.75 micron pixels

Jupiter SSIc-1a.jpg

 

2.2 micron pixels

Jupiter C11 NI5.jpg

 

Why not just add a 1.7X Barlow?  Every time I've tried that my images went soft - no matter whose high-quality Barlow.  Why? The atmosphere, not the optics, limited the usable magnification.

 

e.g. The C11 native focal length is about 2700mm.  Most the image sensors I used had diagonals of about 9mm and act like 9mm focal length eyepieces.  Thus the magnification was about 2700/9 = 300X.  There are only a few times a year when I can use that high a magnification.  Adding a 1.7X Barlow pushed the magnification up to 1.7 x 300X = 510X... and my atmosphere never supported that.

 

Returning to the 1.1 micron BSI sensors.  Imagine, if there was a solar-system-imaging camera using a 1.1 micron pixel size BSI sensor.  At 300X magnification, my image with Jupiter's diameter of 300 pixels (with 2.2 micron pixels) would increase in size to 600 pixels!

 

Further, on those many nights when the atmosphere cannot support 300X magnification, one could use 150X magnification and still image Jupiter with a diameter of 300 pixels.

 

Consider the many amateur astronomers with telescopes smaller that a C11, they could image Jupiter at high resolution too.
e.g.  A 100mm ED telescope of say 900mm focal length and a 9mm diagonal sensor would have a magnification of 900/9 = 100X.  With a 1.1 micron pixel size BSI sensor, at 100X magnification, the image of Jupiter's diameter would be 200 pixels.  However, the optics of a 100mm ED scope are easily good for 200X magnification, so add in a 2X Barlow and obtain an image of Jupiter's diameter of 400 pixels.  That's a major and significant benefit.

 

Are there BSI 1.1 micron sensors for such a planetary-imaging cameras available?  Yes, a whole lot of them.

 

Sony Image Sensors
IMX219PQ 8  MP bsi 1.12 µm (H) × 1.12 µm (V) 3280 × 2464 @ 30fps - Nexus 9 rear camera
IMX214  13.5MP bsi 1.12 µm (H) × 1.12 µm (V) 4208 × 3120 @ 30fps
IMX219   8  MP bsi 1.12 µm (H) × 1.12µm (V)  3280 × 2464 @ 60fps @1080 with V-crop
IMX230  21  MP bsi 1.12 µm (H) × 1.12 µm (V) 5344 × 4016 @ 24fps
IMX258  13  MP bsi 1.12 µm (H) × 1.12 µm (V) 4208 × 3120 @ 30 fps
IMX278 13MP Pixel  1.12 µm (H) × 1.12 µm (V) 4208 × 3120 @ 30fps
IMX135 13MP Pixel  1.12 µm (H) × 1.12 µm (V) 4208 × 3120 @ 24 fps 10-bit A/D

 

ON Aptina Sensors
AR1335   13MP 1.1µm  x 1.1µm  @ 30fps

 

Samsung Sensors
S5K8B1   2Mp 1.12µm x 1.12µm 1920x1080 Bsi 30fps Secondary Camera sensor in Samsung Galaxy
S5K8B1Y 2Mp 1.12µm x 1.12µm 1920x1080 Bsi 30fps
S5K6D1   4Mp 1.12µm x 1.12µm 1920x1080 Bsi ISOCELL’s RWB

 

So, what are we waiting for?  The benefit is clear to see and the technology is available.
When are the solar-system-imaging-camera manufacturers going to give us 1.1 micron BSI cameras?

 

RickV



#2 MvZ

MvZ

    Surveyor 1

  • *****
  • Posts: 1,884
  • Joined: 03 Apr 2007

Posted 02 February 2016 - 07:01 PM

> Why not just add a 1.7X Barlow?  Every time I've tried that my images went soft - no matter whose high-quality Barlow.  Why? The atmosphere, not the optics, limited the usable magnification.

 

This makes no sense. If the atmosphere is the limiting factor, then a 2.2 micron camera + a good 2x barlow is the same as a 1.1 micron camera with no barlow, because the seeing (and Jupiter) have exactly the same 'magnification'...

 

The small pixel sensors typically do have low read noise (full well capacity is lower too, but this pretty much never a problem). For deepsky imaging I prefer using larger pixels, and I can use the same camera with a barlow for planetary imaging, but not really the other way around. Still, I would certainly like to give a 1.1 micron camera a go on my F/5 Dobson, just to simplify things further by getting rid of the barlow, and also to try something new. But not because I see major or significant benefits...


Edited by MvZ, 02 February 2016 - 07:16 PM.


#3 ccs_hello

ccs_hello

    Voyager 1

  • *****
  • Posts: 10,739
  • Joined: 03 Jul 2004

Posted 02 February 2016 - 07:58 PM

RickV

 

The constraint is the long focal length will reduce the light flux.  Tiny pixel pitch means small per-pixel area to gather incoming photons thus weak signal, especially high frame rate.

 

I'd suggest try IMX249 2.9um pitch BSI and 1920x1200 resolution, a reasonable compromise.

 

Clear Skies!

 

ccs_hello



#4 gregj888

gregj888

    Mercury-Atlas

  • -----
  • Posts: 2,855
  • Joined: 26 Mar 2006

Posted 02 February 2016 - 08:32 PM

Rick, unless you are using a system around f/2-f/3 the 1.1 um pixels are just severely oversampling and not gathering much additional data.  My real rough rule of thumb is the diameter of the Airy disk in microns is about the same as the f/#.**   Splitting the Airy disk across 2 pixels (linearly or 4 by area) is considered the best mix of resolution and intensity.

 

For double star speckle we routinely split the Airy disc across 5 pixels to improve the measurements (reduces aliasing), mostly for the angles.

 

There is no question you can get a "prettier picture" by oversampling, but there probably isn't any extra information beyond a factor of 3 or 4 but may be some there.

 

I agree with CCS, 2.9 should do well...  We are looking at a IMX290 based camera now, but will probably put a 1.5x Barlow  on it to get the scale we want to work at. 

 

Remember too, this is all about angles, so if the angle across the 1.9um pixel is the same as the 3.7 with a Barlow, seeing will be the same (tube currents  and Barlow quality have local effects) as well.

 

** works at 410 nm (UV) wavelength, errors on the over sampling side a bit but easy to remember.  Double in the NIR.



#5 Sunspot

Sunspot

    Voyager 1

  • *****
  • Posts: 11,024
  • Joined: 15 Mar 2005

Posted 02 February 2016 - 08:50 PM

Rick, unless you are using a system around f/2-f/3 the 1.1 um pixels are just severely oversampling and not gathering much additional data.  My real rough rule of thumb is the diameter of the Airy disk in microns is about the same as the f/#.**   Splitting the Airy disk across 2 pixels (linearly or 4 by area) is considered the best mix of resolution and intensity.

 

For double star speckle we routinely split the Airy disc across 5 pixels to improve the measurements (reduces aliasing), mostly for the angles.

 

There is no question you can get a "prettier picture" by oversampling, but there probably isn't any extra information beyond a factor of 3 or 4 but may be some there.

 

I agree with CCS, 2.9 should do well...  We are looking at a IMX290 based camera now, but will probably put a 1.5x Barlow  on it to get the scale we want to work at. 

 

Remember too, this is all about angles, so if the angle across the 1.9um pixel is the same as the 3.7 with a Barlow, seeing will be the same (tube currents  and Barlow quality have local effects) as well.

 

** works at 410 nm (UV) wavelength, errors on the over sampling side a bit but easy to remember.  Double in the NIR.

Am I understanding that the ICX chip is 2.9 microns?



#6 ccs_hello

ccs_hello

    Voyager 1

  • *****
  • Posts: 10,739
  • Joined: 03 Jul 2004

Posted 02 February 2016 - 10:05 PM

IMX290

 

See  http://www.cloudynig...-2#entry6974962



#7 PiotrM

PiotrM

    Skylab

  • *****
  • Posts: 4,195
  • Joined: 03 Jan 2010

Posted 03 February 2016 - 02:05 AM

Note that BSI for CMOS isn't used like in high end CCD (or some CMOS) to get even better performance but just to make such small pixels to work. So BSI should not be the key factor for planetary camera sensor.

#8 chonum

chonum

    Airylab

  • *****
  • Vendors
  • Posts: 262
  • Joined: 02 Sep 2008

Posted 03 February 2016 - 03:44 AM

At some stage, the rule is to sample the PSF to its third.

The PSF size is 2.44.Wavelength.FL/Diameter.

If we use the 550nm wavelength as a reference, then 1.34.F/D

 

If you use for example a common F/10 SCT, the PSF if 13.4µm, that should be sampled to 4.4µm. Then with 2.2µm pixel you are at a x6 sampling, meaning you just waste 75% of the light for nothing, just increasing your exposure time.

 

For a newton that could make sense, but fow do you do when the seeing is not great and you want to decreasing the sampling factor?

*

Another thing. When using a short F/D with small pixel sensors, how do you insert your ADC that will bring huge SA+Astigmatism at low F/D ?

ADC is almost mandatory for any serious HR planetary imaging, especially on Saturn.


Edited by chonum, 03 February 2016 - 03:47 AM.


#9 ccs_hello

ccs_hello

    Voyager 1

  • *****
  • Posts: 10,739
  • Joined: 03 Jul 2004

Posted 03 February 2016 - 08:23 AM

Would like to add, if the image sensor has too many pixels (tiny pitch, megapixel race),

make sure the camera supports true ROI (Region of Interest) readout mode to get high enough frame rate.

Or too many pixels will be counter-productive from Solar System Imaging point of view.

 

P.S. CMOS-based image sensors are weak in this area

 

Clear Skies!

 

ccs_hello



#10 Cotts

Cotts

    Just Wondering

  • *****
  • Posts: 12,001
  • Joined: 10 Oct 2005

Posted 03 February 2016 - 08:37 AM

Would like to add, if the image sensor has too many pixels (tiny pitch, megapixel race),

make sure the camera supports true ROI (Region of Interest) readout mode to get high enough frame rate.

Or too many pixels will be counter-productive from Solar System Imaging point of view.

 

P.S. CMOS-based image sensors are weak in this area

 

Clear Skies!

 

ccs_hello

FYI, my Canon 60Da does video at 640x480 pixels (a very tiny ROI compared to the full sensor)  which is very suitable for planetary and (my favourite) double star imaging.  Sure, there are CCD's that are better, but not all DSLR's can be eliminated from this sort of imaging.

 

Dave

 

Dave



#11 RickV

RickV

    Vanguard

  • -----
  • topic starter
  • Posts: 2,474
  • Joined: 17 May 2013

Posted 21 February 2016 - 10:11 PM

Thanks for all the comments and suggestions.

Allow me to respond...

 

RE Surveyor 1:  If the atmosphere is the limiting factor, then a 2.2 micron camera + a good 2x barlow is the same as a 1.1 micron camera with no barlow, because the seeing (and Jupiter) have exactly the same 'magnification'

 

I don't quite see the validity of your argument.  I obtain magnification by an improved "eye" and you by a barlow; in my case I am not adding optical magnification while you are.

 

RE Airylab: At some stage, the rule is to sample the PSF to its third.
The PSF size is 2.44.Wavelength.FL/Diameter.
If we use the 550nm wavelength as a reference, then 1.34.F/D
If you use for example a common F/10 SCT, the PSF if 13.4µm, that should be sampled
to 4.4µm.

Then with 2.2µm pixel you are at a x6 sampling, meaning you just waste 75% of the light for nothing,

just increasing your exposure time.

You're really saying: Minimum Pixel Size = F#/3... remarkable close to what I found by experiment (read on).

 

Here's what I found by reviewing the literature and then conducting some experiments.

From Wikipedia...

If one considers diffraction through a circular aperture, we have:
Sin θ = 1.22 λ/D where
    θ is the angular resolution (radians),
    λ is the wavelength of light,
    and D is the diameter of the lens' aperture.
    
The factor 1.22 is derived from a calculation of the position of the first dark

circular ring surrounding the central Airy disc of the diffraction pattern.

Two objects are said to be just resolved when the maximum of the first Airy pattern

falls on top of the first minimum of the second Airy pattern (the Rayleigh

criterion).

The angular resolution may be converted into a spatial resolution, Δℓ, by

multiplication of the angle (in radians) with the distance to the object.

For a microscope, that distance is close to the focal length f of the objective. For

this case, the Rayleigh criterion reads:
Δℓ = 1.22 λf/D

A similar result holds for a small sensor imaging a subject at infinity: The angular

resolution can be converted to a spatial resolution on the sensor by using f as the

distance to the image sensor (f=focal length in a telescope); this relates the

spatial resolution of the image to the f-number, f#, where f# = f/D (focal length /

diameter of aperture)
Δℓ = 1.22 λ(f/D) = 1.22λ(f#)

For blue visible light, the wavelength λ is about 420 nanometers.
Δℓ = 1.22 x 420nm f#, Δℓ = 512nm x f#.
For an f/8 telescope, Δℓ=4,000nm or 4µm.  e.g. We need pixel size of 4µm or less.

Rick's Comment:  A general way of stating this is that pixel size = f#/2.
e.g.
f/20 needs pixels of 10 micron or less,
f/10 needs pixels of 5 micron or less,
f/6 needs pixels of 3 micron or less, etc.

In a digital camera, making the pixels of the image sensor smaller than this would

not actually increase optical image resolution. However, it may improve the final

image by over-sampling, allowing noise reduction.

 

Experiments:
I used a tripod mounted Williams Optics 66mm ED, Focal length = 388mm, F/5.9 to

examine a paper resolution target at 10 meters (30 feet).  I was able to reduce the

aperture of the telescope to F/8.35, F/11 and F/18.75 by inserting black cardboard

disks with a center hole in front of the objective lens.

 

Cameras:
AmScope MU1000: pixel size 1.667µm, 2x2 binned for 3.34µm and 4x4 biined for 6.7µm.
Celestron NexImage 5: pixel size 2.2µm
MallinCan SSIc: pixel size 3.75µm

 

Figure 1 shows the effect of pixel size on images at F/5.9.

FIGURE 1 F5-9.jpg
To my eye, resolution is good with both 1.66µm and 2.2µm.
This would be a pixel size of about 2µm at F/6 or min pixel size = F#/3.

 

Figure 2 shows the effect of pixel size on images at F/8.35.

FIGURE 2 F8-35.jpg

To my eye, resolution is good with both 1.66µm and 2.2µm.
This would be a pixel size of about 2µm at F/8 or min pixel size = F#/4.

 

Figure 3 shows the effect of pixel size on images at F/11.

FIGURE 3 F11.jpg

To my eye, resolution is beginning to fail somewhat with both 1.66µm and 2.2µm.
Why?
F11 with a 388mm focal length means an aperture of 35.3mm.
Maximum magnification of a telescope is usally stated as about 2x aperture in mm.
Max. Magnification = 2 x 35.3 = 70X.
The image sensors here have diagonals of about 9mm, so actual magnification =

388mm/9mm = 43X... well below the maximum of 70X.
Anyone have an explanaton?

 

Figure 4 shows the effect of pixel size on images at F/18.75.

FIGURE 4 F18-75.jpg

To my eye, resolution is failing quickly with both 1.66µm and 2.2µm pixels.
F18.75 with a 388mm focal length means an aperture of 20.7mm.
Maximum magnification of a telescope is usually stated as about 2x aperture in mm.
Max. Magnification = 2 x 20.7 = 40.4.
The image sensors here have diagonals of about 9mm, so actual magnification =

388mm/9mm = 43X... slightly beyond the maximum of 40X.

 

Overall, it is clear to me that smaller pixels produce greater image magnification (a good thing) and produce superior images (Viking 1: "prettier picture" - a good thing) at all F#s.

 

I'm open to comments.



#12 Cotts

Cotts

    Just Wondering

  • *****
  • Posts: 12,001
  • Joined: 10 Oct 2005

Posted 22 February 2016 - 10:10 AM

A while ago I posted a thread here to survey the plate scale that the very good planetary imagers on CN use.   There was a very strong tendency for the best imagers to be using 0.10" to 0.15" per pixel with 12" to 20" scopes. So, too, Christopher Go and Damien Peach.   This correlates well with the idea that the ideal sampling is to put your scope's resolving power on two pixels.  Spreading the resolution over more pixels wastes light and not having the resolution on enough pixels (or a single pixel being a multiple of the resolving power) wastes resolution.   Assuming you want to image at a certain specific plate scale - say 1.2"/px - then you have to weigh a few factors.

 

Tiny pixels 1-2 micron are less sensitive so longer frame exposures needed - freezing the seeing becomes more difficult.  But you can shoot at shorter focal lengths,  f/10 or so, which lessens stress on the mount/scope/aiming combination.  

 

Larger pixels 5-6 micron are more sensitive so shorter exposures can be used to freeze the seeing.   But you will need to ramp up the f/ratio of the imaging train to f/30 or even f/50 which brings real challenges to the mount/scope/aiming combo.

 

Is there a 'Goldilocks solution'?

 

Dave



#13 Bart Declercq

Bart Declercq

    Apollo

  • *****
  • Posts: 1,039
  • Joined: 21 Jan 2007

Posted 23 February 2016 - 07:48 AM

A while ago I posted a thread here to survey the plate scale that the very good planetary imagers on CN use.   There was a very strong tendency for the best imagers to be using 0.10" to 0.15" per pixel with 12" to 20" scopes. So, too, Christopher Go and Damien Peach.   This correlates well with the idea that the ideal sampling is to put your scope's resolving power on two pixels.  Spreading the resolution over more pixels wastes light and not having the resolution on enough pixels (or a single pixel being a multiple of the resolving power) wastes resolution.   Assuming you want to image at a certain specific plate scale - say 1.2"/px - then you have to weigh a few factors.

 

Tiny pixels 1-2 micron are less sensitive so longer frame exposures needed - freezing the seeing becomes more difficult.  But you can shoot at shorter focal lengths,  f/10 or so, which lessens stress on the mount/scope/aiming combination.  

 

Larger pixels 5-6 micron are more sensitive so shorter exposures can be used to freeze the seeing.   But you will need to ramp up the f/ratio of the imaging train to f/30 or even f/50 which brings real challenges to the mount/scope/aiming combo.

 

Is there a 'Goldilocks solution'?

 

Dave

 

This doesn't really make any sense working with large pixels at F/30 is no more difficult than working with small pixels at F/10 - it in no way creates more stress on the mount/scope/aiming combination since the image scale is the same in both cases.

 

What does make a difference is how many pixels you've got - for example if you're using a 640x480 camera and you're working at 0.12"/pixel Jupiter is approx 400 pixels in diameter, so vertically you have only about 40 pixels (or 5 arc seconds) of margin around the subject, which makes aiming and tracking tough and you need a very stable and accurate mount.

 

If you're using a 1280x960 camera, you've got several Jupiter diameters of breathing room around the planet - with Firecapture you can even use ROI to increase the framerate, while using auto-align & autoguide combined to keep the planet centered. If your mount is so unstable that this is not enough room, you're not going to get good planetary images regardless of the camera used.

 

This is entirely independent of the pixel size of the camera itself, the only effect that has is on how much barlowing you may need, which is just a slight change in the optical train, but places no particular extra constraints on mount or scope.

 

Of course, if you're going to use a 1.1µm pixel size camera, you will reach the "optimal" pixel scale at F/6 - so SCT users won't be able to effectively use that camera unless they're using a focal reducer, while Newtonian scope users will be annoyed because they're usually working with F/4.5-F/5 telescopes and would need a 1.2x barlow to reach optimal pixel scale, I don't know many barlow lenses that do a mere 1.2x so they would be stuck undersampling (without a barlow) or oversampling (by using a 1.5x barlow, which is the "least strong" barlow typically available).

 

I've used cameras ranging from 7.4µm to 2.4µm and all it changes is how much barlowing you need, and larger pixels are nice because then you've got more flexibility, being able to choose a somewhat smaller image scale in worse seeing conditions to get a brighter image to increase framerates to slightly compensate for the worse seeing.



#14 PiotrM

PiotrM

    Skylab

  • *****
  • Posts: 4,195
  • Joined: 03 Jan 2010

Posted 23 February 2016 - 05:23 PM

Yes, there is no sensitivity or seeing effect on big/small pixels if in both cases you are imaging on optimal-max telescope resolution.

#15 RickV

RickV

    Vanguard

  • -----
  • topic starter
  • Posts: 2,474
  • Joined: 17 May 2013

Posted 24 February 2016 - 09:06 PM

For planetary imaging, what I want is the largest high-resolution image of the planet.

 

Largest Image:  Means most pixels.
a) Longest focal length
b) Smallest pixels
c) Enough aperture and pixel sensitivity to permit high enough frame rate (~30 frames/sec) for ‘lucky imaging’

 

High Resolution Image: Means that all pixels carry some new information
a) Optimum pixel size = ~F#/4

 

Don’t Exceed Maximum Magnification: 2 x Aperture (mm)

 

Base for Calculation: Diameter of Jupiter is 300 pixels using a 2.2 micron pixel camera (Celestron NexImage 5) with a Celestron C11 at F/10 (focal length = 2794mm).

 

Given the above criteria, I calculate the pixel diameter of Jupiter with various telescopes/cameras.
Image Diameter (pixels) = 300 pixels x [2.2um / pixel size (um)] x [focal length(mm) / 2794mm]

 

Telescope-Pixels-1.jpg

 

My C11 at F/10 with a NexImage 5 can produce a 300 pixel image of Jupiter’s diameter.

Sadly, it appears that obtaining a larger image of Jupiter requires substantially greater apertures.

It may require a C14 to reach 400 pixels and a 20 inch scope to reach 500 pixels.

 

Aaaaah... what can you do?

 

 



#16 Bart Declercq

Bart Declercq

    Apollo

  • *****
  • Posts: 1,039
  • Joined: 21 Jan 2007

Posted 25 February 2016 - 03:57 AM

Telescope-Pixels-1.jpg

 

My C11 at F/10 with a NexImage 5 can produce a 300 pixel image of Jupiter’s diameter.

Sadly, it appears that obtaining a larger image of Jupiter requires substantially greater apertures.

It may require a C14 to reach 400 pixels and a 20 inch scope to reach 500 pixels.

 

Aaaaah... what can you do?

For planetary imaging, what I want is the largest high-resolution image of the planet.

 

What can you do? Just use a barlow lens, I really don't get why some people seem to be so hesitant about that...


Edited by Bart Declercq, 25 February 2016 - 03:58 AM.


#17 happylimpet

happylimpet

    Fly Me to the Moon

  • *****
  • Posts: 6,697
  • Joined: 29 Sep 2013

Posted 25 February 2016 - 06:41 AM

 

 

What can you do? Just use a barlow lens, I really don't get why some people seem to be so hesitant about that...

 

 

Its funny...when I was a nipper, Barlow lenses were the spawn of Satan, and no-one had a good word to say about them. Now they're our best friends!



#18 RickV

RickV

    Vanguard

  • -----
  • topic starter
  • Posts: 2,474
  • Joined: 17 May 2013

Posted 25 February 2016 - 10:16 AM

A Barlow doesn't solve the problem of getting a larger, higher-resolution image.

A Barlow can provide a larger image but often no more resolution; that was the point of making the table... to show the limits on image size imposed by the resolution criterion that the optimum pixel size = ~F#/4.

 

 

 



#19 Bart Declercq

Bart Declercq

    Apollo

  • *****
  • Posts: 1,039
  • Joined: 21 Jan 2007

Posted 25 February 2016 - 10:57 AM

A Barlow doesn't solve the problem of getting a larger, higher-resolution image.

A Barlow can provide a larger image but often no more resolution; that was the point of making the table... to show the limits on image size imposed by the resolution criterion that the optimum pixel size = ~F#/4.

 

Hmm, I think you're confusing ideas and that's leading to misunderstandings. It starts with your very first statement

 

 

When I changed from a solar-system-imaging camera with 3.75 micron pixels to one with 2.2 micron pixels, my images of Jupiter and Saturn improved dramatically.  Why?  Simple - more pixels forming the image.

 

You could have gotten *exactly* the same result by using a 1.7x barlow rather than switching cameras, and your subject line is about using a 1.1µm camera for planetary observations - the barlow doesn't magically make the telescope sharper, but it does allow you to reach the resolution limit without needing a specific camera pixel pitch.

 

Let me link back to my (by now quite ancient) post concerning resolving power and optimum pixel size:

http://www.cloudynig...d/#entry3621781

 

On an 11" telescope you'll want to use at least an image scale of roughly 0.15"/pixel to achieve maximum resolution, which does correspond to roughly a 300 pixel Jupiter - so indeed, to get a larger *useful* image of Jupiter (where useful = additional detail) will indeed require a larger telescope. However, this has *nothing* to do with how big your camera's pixels are, which is why this whole thread is so confusing - yes, you need a bigger scope to get moar detailz, but I would have though that's pretty obvious to most astronomers.



#20 Kokatha man

Kokatha man

    James Webb Space Telescope

  • *****
  • Posts: 17,099
  • Joined: 13 Sep 2009

Posted 25 February 2016 - 06:57 PM

Rick, I think a bit of deconvolution is necessary in your "logic" so to speak... :flowerred:

 

Nick, I have one of those Spawns of Satan :mrevil:   safely under lock & key :locked: ...well, I lie, I think it's in the old hallstand - I must remember to genuflect next time I walk past it! :shocked:  :scared:

 

...sorry, but at least I've used a couple of emoticons I haven't previously! :lol:

 



#21 Odairpm

Odairpm

    Sputnik

  • *****
  • Posts: 40
  • Joined: 14 Apr 2013

Posted 25 February 2016 - 08:06 PM

Hi Rick,

I agree about your clains about the advantage of smaller pixels in magnifications power and best resolution. All lens add to optical system have a harfull result in image with lost light and distinctness (independently as so good they are).

I had imaging with a Flea 3-U3 8.8mp ( Sony IMX 121) with pixels of 1.55 microns  at F/10,  and  have had good images with impressive detais, despite of SCT9.25 and a average seeing. My comparisons with another sensors, in the same conditions, show me the quality of this sensor.

However, I tinhk that there are another variables out of the equation. The architecture of sensor, the Image Astronomical System, seeing, relative humidity etc.

But I belive that is necessary another assays to change some concepts.

My best regards

JUPITER
Album: planetary
3 images
0 comments


Edited by Odairpm, 25 February 2016 - 09:10 PM.


#22 Bart Declercq

Bart Declercq

    Apollo

  • *****
  • Posts: 1,039
  • Joined: 21 Jan 2007

Posted 26 February 2016 - 05:19 AM

All I can do is show results - this is a Jupiter image taken with a 12" Newtonian scope and a Televue 5x Barlow (extended to reach an effective 6x magnifcation) using a camera with 5.6µm pixels - it's true that a barlow will add another optical element to the optical path, but the effect of a good barlow on the image quality is virtually negligeable and unless you're living on the moon, the impact of atmospheric turbulence will be orders of magnitude more significant regarding image quality than the barlow.

 

Besides, what happens if you buy a new telescope with a different F-ratio, do you really want/need to buy another camera when you do?

 

15119316130_9b4ccfa920_o.jpg



#23 DesertRat

DesertRat

    Fly Me to the Moon

  • *****
  • Posts: 6,266
  • Joined: 18 Jun 2006

Posted 26 February 2016 - 08:33 AM

I think Bart has answered the sampling criteria well.

 

I will add that there is some evidence, at least for brighter objects like the planet Mars, some oversampling can have a beneficial result.  This may be the result of placing the pixel to pixel noise variance at a smaller scale than the smallest details one is after.  In any noise control done in processing it is easier to tame the noise without blurring out the hard fought for low contrast details of the planet.  Of course seeing has to be good for this to succeed..

 

Small pixels do have some advantages however.  For example they are more capable of resolving interference fringes in a digital holographic microscope arrangement.  Smaller pixels allow larger reference angles.

 

But for astronomical imaging we want good signal to noise characteristics.  For that larger pixels have an advantage.  It should be appreciated that as pixels get smaller than 1u you are approaching the wavelength of the light itself, and the sensor, in addition to getting noisier as a result, is also getting less and less efficient at photon capture.

 

This is physics, and no reason to rant.

 

Glenn



#24 RickV

RickV

    Vanguard

  • -----
  • topic starter
  • Posts: 2,474
  • Joined: 17 May 2013

Posted 26 February 2016 - 07:52 PM

gregj888: We are looking at a IMX290 based camera now, but will probably put a 1.5x

Barlow  on it to get the scale we want to work at.

The IMX290 is a BSI sensor with 2.9 micron pixels (1920x1080).  Sounds not too bad.

 

Bart Declercq: Let me link back to my (by now quite ancient) post concerning resolving power and optimum pixel size: http://www.cloudynig...d/#entry3621781
Thanks for the link Bart, an excellent read!

 

Odairpm: I had imaging with a Flea 3-U3 8.8mp (Sony IMX 121) with pixels of 1.55 microns at F/10, and  have had good images with impressive detais, despite of SCT9.25 and a average seeing. My comparisons with another sensors, in the same conditions, show me the quality of this sensor.
Yes, I too had been looking at that particular camera... fast frame rates too... but a mite pricey at US$1,000 and production of that sensor is due to end in 2016.

 

Bart Declercq: All I can do is show results - this is a Jupiter image taken with a 12" Newtonian scope and a Televue 5x Barlow (extended to reach an effective 6x magnifcation) using a camera with 5.6µm pixels
Bart, Nice photo!
That's a pixel size of about F#/5 ... F/30 / 5 = 6um pixels.
However, that is a focal length of about 9,000mm... long!

 

DesertRat: But for astronomical imaging we want good signal to noise characteristics.  For that larger pixels have an advantage.  It should be appreciated that as pixels get smaller than 1u you are approaching the wavelength of the light itself, andthe sensor, in addition to getting noisier as a result, is also getting less and less efficient at photon capture.  This is physics, and no reason to rant.

Small BSI pixels can have signal to noise characteristics as good as larger non-BSI pixels.

 

As for the rant, well OK, I understand now while I am not likely to see a 1.1 micron pixel in a planetary camera; they seems to offer little gain over pixels in the 2 to 3 micron size.

 

Can we look at the effects on the image of frames per second?
As I recall, with my C11 at native F/10 using a NexImage 5 (2.2 micron pixels), on Jupiter with a max histogram entry of say 170 out of 255, the highest framerate that I was able to attain was about 40 frames per second.

 

Jupiter rotates in about 10 hours, so a pixel on one side of the image will move to the other side of the image in 5 hours.  With 2.2 micron pixels, Jupiter's diameter appeared to be 300 pixels.  Thus, the image would shift 1 pixel in = 5h x 60min x 60 sec / 300 pixels = 60 seconds.

 

If I limit my exposure to 60 seconds to avoid image blur due to planetary rotation, at 40 frames per second, I have: 60 sec x 40 fps = 2400 frames.  How many frames do I need to produce 1 decent image?

 

Faster frame rate cameras are becoming available such as the new QHY5-III Series.

QHY5-III Series USB3.jpg

 

In particular, the QHY5III178 with an IMX178(BSI) sensor with 2.4 micron pixels is capable of 300 fps at 640x480.  The QHY5III290 with an IMX290(BSI) sensor with 2.9 micron pixels is capable of 350 fps at 960x540 pixels.

 

What sort of image improvement might one expect raising the fps by 7.5X; i.e. moving from 40 fps to 300 fps?

 



#25 gregj888

gregj888

    Mercury-Atlas

  • -----
  • Posts: 2,855
  • Joined: 26 Mar 2006

Posted 26 February 2016 - 08:44 PM

Rick,

 

Generally FPS effects noise level, faster readout, more noise.  CMOS is changing that, but has a bunch of new issues to throw into the soup. CMOS cameras pre-process the data before you see it.  Darks may or may not be darks and may or may not be applicable to the next frame...  Photometry has to be done very carefully, even relative photometry because each row or column may have it's own and different calibration, sometimes per frame. We are still learning to use these tools...  and they change with each sensor model or seem to at the moment.

 

I don't use FPS as a measure since it doesn't tell me anything, exposure does.  To answer your question, 40 fps to 300 fps, no improvement at all unless seeing is very very very bad and then you shouldn't be imaging.

 

I play with double stars so FPS only matters in how long a set of data takes to acquire. 10fps is enough.  Exposure is what freezes the seeing and that depends on the size of the scope and the seeing itself.  My exposures are in the 30ms-250ms (NIR really good night) range so rolling shutter doesn't seem to matter even though I would intuitively think a global shutter is better.

 

Really large scopes or a need for precision timing, a global shutter is preferred and maybe required.  At present, you have to trade some sensitivity to get the global shutter, I can use the sensitivity more...

 

To freeze seeing up to a 0.5m/20" scope figure 10ms on the short end, at least I wouldn't be imaging if that didn't freeze things. With a 680nm or so low pass I've gotten diffraction rings at 250ms (one night, but haven't tried often, 8" SCT, 3.? FPS).  Nominal is 50ms so with ROI in the 10-15 FPS range.  Even a camera that only does 10 FPS full frame will let me spend most of my time exposing target in my small 256 x 265 ROI.

 

350 FPS gives you an exposure of about 2ms ...  need a pretty bright object to be worried about that and I only know of one bright enough without a strobe.

 

<dang spell check...>


Edited by gregj888, 27 February 2016 - 12:07 AM.



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