How close does light source have to be to end of telescope for flats?

[tooth_dr]

I was thinking of mounting a light panel on a swivel bracket for taking flats in my obsy.  This will mean that my panel will be about 30cm away from the end of the telescope.  Is this still ok?  The panel is much larger that the 80mm apeture of the scope.

[Adreneline]

Hi.

I use an iPad which I rest on the end of the scope which is covered with a white cloth. However, I tried taking flats on one occasion by just pointing the scope with white cloth at the white ceiling in the room where the gear is stored and I couldn't discern any difference in the resulting flats. I have also tried the 'point it at the dawn sky' approach but ended up with annoying gradients and a funny colour cast so I gave up on the method.

HTH

Adrian

P.S. I also borrowed a light panel but ended up with strange banding lines on the flats when using a dslr (but not a CCD) - I guess something to do with shutter sync problems.

[Skipper Billy]

My light panel is about 40cms away from the end of the scope - its an A3 light panel - it works just fine as there is no extraneous light.

One tip - leave the panel on a for at least a few mins before you start taking flats or you might get banding with some brands of light panel - I use a Huion and its not affected at all by banding.

[tooth_dr]

Great info guys thanks.  I am using an IPAD now, but with roof closed it's a bit of pain, plus it doesnt work with my larger scopes.  I was thinking of a 600x600 panel, a cheap one off Amazon.

[tooth_dr]

https://www.amazon.co.uk/dp/B01B286NMC/ref=twister_B07BHYW4WX?_encoding=UTF8&psc=1

[Oddsocks]

22 hours ago, tooth_dr said:

This will mean that my panel will be about 30cm away from the end of the telescope.  Is this still ok?  The panel is much larger that the 80mm apeture of the scope.

For a refractor, oversize and distance does not matter so much though it is a good idea to use an aperture mask over the panel to minimise the stray light bouncing around the observatory and entering the OTA via the gap in the focuser tube or un-filled screw holes in the OTA etc.

For a reflector, distance can be a problem if the panel is too close to the OTA aperture as this allows off-axis stray light from the panel to enter the focuser opening directly and can cause contamination of the flats, for a reflector, more distance and careful use of an aperture mask over the panel to suit the size of OTA will yield the best results, see attached gif below for an explanation.

For dimmable light panels it is best to use them on full brightness as this sets the PWM controller mark-space ratio to nearly 100% and minimisers flicker, to avoid shutter shading and improve consistency the flats exposure should really be longer than a second or so and to achieve this it is necessary to dim the over-bright panel with an optical gel, the one linked below is suitable and depending on the size of the panel one sheet of 1.22m x 0.53m can be cut in two and two layers over the panel will dim the panel sufficiently to give an appropriate exposure time.

https://www.sblite.co.uk/neutral-density/265-409-rosco-299-12-neutral-density-5060312563254.html#/23-sheet_roll-sheet

HTH

William.

 

 

[Oddsocks]

23 hours ago, tooth_dr said:

https://www.amazon.co.uk/dp/B01B286NMC/ref=twister_B07BHYW4WX?_encoding=UTF8&psc=1

Way too bright....thats equivalent to roughly a 100W tungsten filament lamp, you should be looking for a graphics tracing panel using natural daylight LEDs.

Something like this (linked below) in A4 or A3 size (though I have no experience of this make but this is the sort of thing, powered via a USB port on the computer or a separate, generic, mains AC to 5V USB charger-adaptor. The A3 HUION panel that I have is very good but expensive in the A3 size, the A4 size is more reasonable, but if A4 is big enough then search for a HUION panel as these work well).

https://www.amazon.co.uk/M-Way-Brightness-Adjustable-Architecture-Calligraphy/dp/B01N40QV9R/ref=sr_1_3?ie=UTF8&qid=1538041263&sr=8-3&keywords=tracing%2Blight%2Bbox&th=1

HTH.

William.

[tooth_dr]

3 hours ago, Oddsocks said:

Way too bright....thats equivalent to roughly a 100W tungsten filament lamp, you should be looking for a graphics tracing panel using natural daylight LEDs.

Something like this (linked below) in A4 or A3 size (though I have no experience of this make but this is the sort of thing, powered via a USB port on the computer or a separate, generic, mains AC to 5V USB charger-adaptor. The A3 HUION panel that I have is very good but expensive in the A3 size, the A4 size is more reasonable, but if A4 is big enough then search for a HUION panel as these work well).

https://www.amazon.co.uk/M-Way-Brightness-Adjustable-Architecture-Calligraphy/dp/B01N40QV9R/ref=sr_1_3?ie=UTF8&qid=1538041263&sr=8-3&keywords=tracing%2Blight%2Bbox&th=1

HTH.

William.

Thanks William. I did look at those tracing panels too, they will be more consistent too.

Unfortunately I need to cover a 305mm apeture so will need an A2 version.

Would I be better placing the ND board you linked above over the end of the scope or on the tracing panel. 

[Oddsocks]

5 hours ago, tooth_dr said:

Unfortunately I need to cover a 305mm apeture so will need an A2 version.

Would I be better placing the ND board you linked above over the end of the scope or on the tracing panel. 

That can be the issue with large reflectors, the HUION panel I have is the A3 and I think I paid around £70 for it a few years ago, that was needed for a 10" Dall Kirkham as the A4 didn't cover it. Still, when I consider the €300 I paid for an electroluminescent flat panel a few years before the HUION and that always produced interference strobe lines in my flats, I think the £70 was worth it.

The gel ND filters I linked to are polyester thin-film and though tough they can scratch or crumple so it's best to place them permanently against the panel. In my observatory I added a piece clear perspex over the panel and three layers of the ND film are sandwiched between the panel and the perspex to keep them flat and clean, the perspex it just taped to the panel with gaffer tape and can be wiped over easily when needed.

If you did try one of the domestic lighting panels that you linked to you will need around four thicknesses of the gel filter to reduce it's output and since you don't know how evenly the panel is lit you may need to add extra diffusion, maybe a piece of white opaque or obscured  perspex instead of clear to hold the gel filters against the light panel would be good idea.

William.

[Oddsocks]

On 26/09/2018 at 11:46, Adreneline said:

I have also tried the 'point it at the dawn sky' approach but ended up with annoying gradients and a funny colour cast so I gave up on the method.

That would never work, the correct method for sky flats is to point the telescope to the opposite quadrant that the sun occupies below the horizon, taking dawn flats, the telescope should be pointing low to the west as close to 180 deg away from the sun as possible and the flats should all have been taken at least 30 minutes before sunrise. Sunset flats just the reverse, telescope pointing low to the east and no flats to be taken until half an hour after sunset.

The problem with colour casts is that many of the free, and even some of the paid-for, image processing packages don't handle OSC and DSLR flats properly. The software should normalise the flat by calculating the chrominance value for each pixel in the four pixel group, RGGB, average the chrominance values for the group and then rewrite the average back to the four pixels so that the flat essentially becomes monochrome and therefore does not leave any colour casts after it is used for calibrating the light.

Unfortunately this does take processing power, memory and extra CPU time so many packages, as you have found, do not bother with normalisation and just assume you have a perfectly colour balanced flat to begin with. Bit lazy of the software programmers really not to at least offer the user the option of normalising colour flats....?

William.

[Jkulin]

Can I make a very quick observation here, your flats should not be taken at full power if your camera cannot adjust the ADU to between 20,000 to 30,000, in the case of my Kodak chip I have to use double ND Gels which look dark grey when illuminated in order to achieve my preferred ADU of 25,000 and flats for my NB Filters can take anything from 10 to 100secs, whereas my LRGB are around the 2-4 sec mark.

[tooth_dr]

22 minutes ago, Jkulin said:

Can I make a very quick observation here, your flats should not be taken at full power if your camera cannot adjust the ADU to between 20,000 to 30,000, in the case of my Kodak chip I have to use double ND Gels which look dark grey when illuminated in order to achieve my preferred ADU of 25,000 and flats for my NB Filters can take anything from 10 to 100secs, whereas my LRGB are around the 2-4 sec mark.

Hi John.  Currently I am using my laptop on a white screen and adjusting the number of layers of a white T-shirt stretched over the end of the scope to achieve an ADU of around 25000, at around 7-10 seconds.

 

I'm reading lots of conflicting information regarding calibrating these sensors, so I'm not sure if I'm doing it right!

 

Flats = 100% you need these, and I use APT to help me get them right.

Bias = do I need these?

Darks = do I need these?

[ollypenrice]

50 minutes ago, tooth_dr said:

Hi John.  Currently I am using my laptop on a white screen and adjusting the number of layers of a white T-shirt stretched over the end of the scope to achieve an ADU of around 25000, at around 7-10 seconds.

 

I'm reading lots of conflicting information regarding calibrating these sensors, so I'm not sure if I'm doing it right!

 

Flats = 100% you need these, and I use APT to help me get them right.

Bias = do I need these?

Darks = do I need these?

The answer for CMOS and CCD are not the same.

For mono CCD this is what works for me, calibrating in AstroArt:

- Master bias has two applications, a) as an alternative to darks and b) as a flat-dark or dark-for-flats.

- Defect Map made by clipping 2000 ADU from a long exposure master dark helps fix hot pixels.

Because I wouldn't use a master flat I hadn't looked at first, I begin by making that and lying to the software ...

I then do the main stacking and calibration like this...

For me this works better than the textbook recipe of regular darks but I suspect there are hidden variables which mean that you might find otherwise. I'm certain that a master bias is the only 'dark for flats' you need with CCD though. It is important to calibrate flats.

Because the bias is a different animal in CMOS cameras I gather you need to do 'proper' darks for flats, ie at the same settings as used for the lights. (Bit of a pest!)

On the original question, my experience agrees with William's diagrams. I sit my panel on top of the extended dewshields with refractors but his Newt diagram makes perfect sense as well.

I use typing paper as my dimming medium so I can keep my flats exposure times to around a couple of seconds by varying it as needed, adding or removing sheets.

Olly

 

 

 

[Jkulin]

1 hour ago, tooth_dr said:

Flats = 100% you need these, and I use APT to help me get them right.

Bias = do I need these?

Darks = do I need these?

Hi Adam,

I bow to Olly's experience skills and abilities, but he has a dark art that I would one day love to follow and learn about, but until I can wave the magic wand that isn't going to happen.

For me when I had your Atik 383, they are a noisy camera and I always made Superbias's up in PI and Master Darks up and used the preprocessing script to do its thing.

I'm sure Olly's method is excellent, but he is starting with good data from excellent skies so he can concentrate on stuff like Bias's, Flats and Darks because if there is something wrong with the image it is highly unlikely to be down to the environment, whereas where I live it could be any multitude of issues, and I presume with you as well.

Just my thoughts of my experiences with your Atik, my Moravian however has proved to be a revelation and considering it is the same chipset it is like chalk and cheese on the noise front.

[RayD]

On 26/09/2018 at 11:31, tooth_dr said:

I was thinking of mounting a light panel on a swivel bracket for taking flats in my obsy.  This will mean that my panel will be about 30cm away from the end of the telescope.  Is this still ok?  The panel is much larger that the 80mm apeture of the scope.

I guess if you consider traditional T-Shirt flats, this uses the sky as a light source, and that's quite far away.

300mm isn't very far, and the light shouldn't be very bright to get a recommended 2 or 3 second or so exposure, so I would say it should be fine, but take a couple of test shots and make sure the light source provides a sufficiently illuminated and even field before fixing in place.  I use Alnitak Flip-Flat covers on both my refractors and these work perfectly.  I don't have a Newtonian reflector, but the comments regarding reflected light make perfect sense.

Edit:  For the pedantic, I know the light source is actually the Sun, but you don't point at the Sun, just the Sun lit sky ?

[Stub Mandrel]

For the original question, I use a distance of 2-4 feet. I also rotate the OTA and take three shots at each of twelve orientations, which is probably OTT but takes care of any gradients. This is clearly unrealistic for huge OTAs or fixed installations where I suppose you need to take more care over having an evenly illuminated target.

I bought a tracing pad but it was very unevenly illuminated.

I use Av mode to calculate exposure, then switch to manual so all flats have the same exposure.

The only problems I have ever noticed are:

A blob of dust moving between imaging and flat making if I have left a long gap.

I once used a green sun lounger as a target to compensate for the astro modding of the camera and it gave poor results. A pale blue sheet of card worked well. A plain magnolia wall works well, but I now use an unpainted canvas which works well in  doorway backlit by natural light.

DSS gives me nice accurate monochrome flats and the only colour that has caused issues was bright green.

 

[ollypenrice]

5 hours ago, Jkulin said:

Hi Adam,

I bow to Olly's experience skills and abilities, but he has a dark art that I would one day love to follow and learn about, but until I can wave the magic wand that isn't going to happen.

For me when I had your Atik 383, they are a noisy camera and I always made Superbias's up in PI and Master Darks up and used the preprocessing script to do its thing.

I'm sure Olly's method is excellent, but he is starting with good data from excellent skies so he can concentrate on stuff like Bias's, Flats and Darks because if there is something wrong with the image it is highly unlikely to be down to the environment, whereas where I live it could be any multitude of issues, and I presume with you as well.

Just my thoughts of my experiences with your Atik, my Moravian however has proved to be a revelation and considering it is the same chipset it is like chalk and cheese on the noise front.

You're very kind but I don't think that the quality of the sky impacts in any way on the effectiveness of the calibration process. The calibration files and processes have no knowledge of the sky outside. They have never seen it. They simply record anomilies of the system itself. Some of our images are certainly not made entirely of good data but, partially, of very bad data since one of the cameras we use is dodderingly old and plagued by column defects. We still manage to make it play, though. (I have to say that its days are numbered, for all that!!!)

On several occasions friends and guests have tried to lure me away from calibration and stacking in AstroArt, extolling the virtues of PI. As yet none of these evangelists has ever beaten the AstroArt result but who knows, one day it may happen!

What's a noisy camera? My own cameras are an Atik 11000 and an Atik 460. If I look at the darks as an indicator of noise then the bad old Kodak 11000's darks look like a snowstorm in Siberia. The darks from the 460 look virtuously dark. But when I process the images, stacked as I describe, I just get on with it with the 11000 data. The 460 data has to have its background sky massaged to get the pixel values anywhere near consistent. I think that what concerns us is not noise but residual noise. The residual noise in my inherently noisy Atik 11000 is remarkably low. Really.

Olly

[Jkulin]

Sorry Olly, I probably didn't explain very well, what I should have said is that you could make a Silk purse out of a Sows Ear.....so with good data, excellent processing skills then one would not be able to determine if your images were ever noisy or not, this was with regards to the final image, whereas my very humble processing skills showed up every single defect of the seeing, pollution and camera noise, making it much harder to produce anything in the same league as you.

[tooth_dr]

On ‎28‎/‎09‎/‎2018 at 16:13, ollypenrice said:

Defect Map made by clipping 2000 ADU from a long exposure master dark helps fix hot pixels.

Olly - I tried stacking my Pacman using the method outlined above, but I couldn't work out what the above quote means, so I omitted that step.  If you have time can you explain what that means?

Yours is the bottom, image stacked and processed using bias + flat to make master flat, and bias as master dark.  Top is using Darks as darks, flats as flats, bias as bias, no dark flats.

 

Thanks

Adam.

 

 

 

[ollypenrice]

14 minutes ago, tooth_dr said:

Olly - I tried stacking my Pacman using the method outlined above, but I couldn't work out what the above quote means, so I omitted that step.  If you have time can you explain what that means?

This is the image stacked and processed using bias + flat to make master flat, and bias as master dark

 

 

Sure. I make a master dark of long exposures first. Use the longest exposure you're likely to use for your lights for this master dark. Then, in AstroArt's Arithmetic, go to Clip and Minimum, type in 2000 and thus black clip the master dark by 2000. This will just leave the persistent hot pixels. You put this 'Defect Map' in the box I indicated above and check the bow for 'Aplly to images.'. I simply took this process from the AstroArt instructions, by the way.

If you're using a Sony chip this Defect Map might not be necessary. Dither guiding might also make it superfluous.  Your Pacman stack looks pretty clean from here. How do you feel this process compares with your previous one?

Olly

[tooth_dr]

3 minutes ago, ollypenrice said:

Sure. I make a master dark of long exposures first. Use the longest exposure you're likely to use for your lights for this master dark. Then, in AstroArt's Arithmetic, go to Clip and Minimum, type in 2000 and thus black clip the master dark by 2000. This will just leave the persistent hot pixels. You put this 'Defect Map' in the box I indicated above and check the bow for 'Aplly to images.'. If you're using a Sony chip this Defect Map might not be necessary. Your Pacman stack looks pretty clean from here. Dither guiding might also make it superfluous.

Ollyu

Thanks Olly

For the sake of completeness - this is a stack of completely uncalibrated lights in DSS (same lights as above 23 x 600)

 

 

[ollypenrice]

28 minutes ago, tooth_dr said:

Olly - I tried stacking my Pacman using the method outlined above, but I couldn't work out what the above quote means, so I omitted that step.  If you have time can you explain what that means?

Yours is the bottom, image stacked and processed using bias + flat to make master flat, and bias as master dark.  Top is using Darks as darks, flats as flats, bias as bias, no dark flats.

 

Thanks

Adam.

 

 

 

There isn't a lot to choose between them, it seems, but the image using calibrated flats and bias-as-dark (the lower one) has a slightly flatter background sky, notably along the upper part of the right hand vertical edge and, to a lesser extent, on the left hand vertical edge (maybe?) I think the active ingredient will be the calibration of the flats rather than the choice of darks versus bias-as-darks.

Olly

Edit: in the absence of any very startling difference it's worth noting the time saved by using bias as dark. If you don't have lots of data this won't matter much but with three cameras running I can end up with a mountain of stuff to calibrate each morning and not having to swap darks around is a big hassle-reducer. Tom and I once collected 24 hours' worth of data on Orion in a single night with four scopes running. We could hardly keep up with it. Nice problem to have!

 

[tooth_dr]

Thanks @ollypenrice I've downloaded AA, so will give that a whirl.

[Xiga]

Just wanted to throw another option out there - Astro Pixel Processor. I've found it to be nothing short of amazing when it comes to calibration. Creating a Bad Pixel Map is a doddle too. It's my go-to program now for calibration, stacking and gradient reduction.  

[martin_h]

On 26/09/2018 at 11:31, tooth_dr said:

I was thinking of mounting a light panel on a swivel bracket for taking flats in my obsy.  This will mean that my panel will be about 30cm away from the end of the telescope.  Is this still ok?  The panel is much larger that the 80mm apeture of the scope.

I park my scope on its side facing the obs wall. On that wall I have mounted an el panel, the scope is about 18" from the panel.