Potential methods of obtaining flat field data

Hey everyone,

I was thinking about how flats could be obtained by each system if they are to be operated remotely. With the cameras used in these systems there is the element of non-flatness that needs to be considered if we were to obtain flats using the traditional method of shooting the dusk sky. There is also the method of wrapping some kind of fabric over the opening of the unit so as to disperse the daylight, but this would require someone to be present each time.

So, I was thinking maybe a light source could be mounted to each unit. This light source would be either very diffuse or have some kind of diffusing element placed in front of it. I am imagining this could be controlled via some kind of addition to POCS. A flat surface or diffusing material could be permanently placed wherever the unit is faced while parked, with this acting as the surface of which the flat field is taken. I hope that makes some kind of sense. Please see the diagram below for potentially more clarity.

In the diagrams I show the potential orientation of the unit as well as four potential arrangements of the light sources, three being on the front of the unit itself and one being placed behind a diffusing material that is placed in front of the unit.

Please let me know what you think. Thanks!


Interesting idea Jacob. In my experience doing proper flats is pretty tricky. Projecting a light source off a diffuse screen doesn’t necessarily give you a flat field. In the past Ive had to use Halogen lamps in the form of flood lights and bounce them off two walls to get a nice uniform illumination. Id be surprised if we could achieve this by pointing an LED at a screen. But its worth testing.

I like the idea of a diffuser that the camera stares at. The issue is the large FOV of the lenses means the diffuser needs to be either very close, or large. At some sites there may be an issue with extending the footprint of the PANOPTES unit. But I guess if the diffuser could be attached to the pier then it might be fine.

My secondary concern about the diffuser is that it is mostly a screen which will catch the wind, and how do we protect it from UV degradation? Maybe the latter is fine as it makes the surface even more matt (diffuse) with time.

I might try and do some tests in the lab to see what makes a good diffuser for PANOPTES. Feel free to do some tests of your own as well.

Yes, there are definitely some points there I did not consider. I thought of this and just chucked it down but have done no tests so far in terms of what lights and materials would be needed. I was thinking maybe a fairly flooded beam and a semi-opaque screen of some kind will be where I am going to start. I have not yet considered reflection or transmission angles either, but that will come together in the tests anyway.

I will look into this a bit more over the weekend when I do some tests of my own and keep the forum up to date with any new stuff. I look forward to seeing what you find.

Here is a quick test I did on flat fielding with a shirt stretched over a whiteboard (which could act as the previously mentioned diffusive material) using a Canon 60D w/ settings 50mm f/1.8, ISO100, 1/4s. Light used was a R300 LED Ring Light in a configuration similar to the one seen here: https://www.google.com/search?q=r300+ring+light&source=lnms&tbm=isch&sa=X&ved=0ahUKEwjAppicz6PfAhWQaCsKHRtQBAMQ_AUIDigB&biw=1440&bih=821#imgrc=fm-XQ2fTxuVJGM:.

The three figures below are one single flat separated into green, blue, and red. They show the gradient between the maximum and minimum values of the normalised flat.

These images came out reasonably flat. I think with better materials than my crinkled shirt, a ring light mounted on each opening of the PANOPTES units could work as a remote flat fielding method. Ring lights can be picked up relatively cheap and filters or even baking paper can be placed on the front of them to soften that LED harshness.

Please, share any thoughts you have on how this can be improved or implemented.


If you’re interested here are the stdevs ~
Green Flat-field standard deviation = 0.021
Blue Flat-field standard deviation = 0.057
Red Flat-field standard deviation = 0.143

1 Like

Not bad. They are smooth, but not perfectly flat. Are the Std devs fractions out of 1? So for the green channel 0.021 is 2.1% std dev? Seems low given that the counts vary by 40% peak to valley from the center to the edges of the frame.

Interesting idea to use ring LEDs and baking paper. Could work.

I set the threshold for each flat differently so as to accentuate the gradient across the field most effectively. That was purely for a visual representation. For the green flat you can see I set the threshold so that the colour represents values from 0.70 and 1.125. For the blue and red flats the thresholds are different as well. Again, that was for visual purposes. Sorry, the standard deviation here is the deviation from 1. I have not put enough information here. The images have been normalised to 1 and the standard deviations here are about the point 1. I will take these out as they are not so relevant.

I think the ring LED is viable given how efficiently it can be mounted to each unit, along with their low power usage.


I’ve had similar thoughts regarding both darks and flats. My scope (PAN006) was located in a dome (no more), and since that is a protected environment, I figured it would be reasonable to place targets for both in the dome, below the scope.

For taking darks, my idea was essentially a trash can below the parked position (i.e. cameras pointed straight down), where the can would be lined with as black a material as we can get at a suitable price. This could be matte black paint, black flocking (glued to the inside), or a black fabric such as a cheap velvet. For flats, the idea would be more like yours, with multiple layers of diffusers above a light source, maybe even lights of multiple colors. In both cases, taking the images in the late evening, with the dome closed, would provide a pretty dark environment in the dome, so there wouldn’t be much light to interfere… though the professor who loaned me the space in the dome was skeptical that it would be dark enough.

These ideas seem to fall apart when we consider placing them where rain and snow can get on them, or wind can damage them. For that I wondered about building a motorized mechanism into the camera box, which would move a hinged flap over the lens hoods. These would actually help protect the lenses during the day. I’ve seen these for sale for telescope covers (e.g. for a remotely operated telescope, to keep the dust off the objective or out of the tube of a Newtonian between observing sessions). Some are designed to provide a source of flats.

Thanks very much for going beyond my thought experiment phase to conducting actual tests, and for sharing your results. I’m intrigued by the idea of putting the light source on the head unit itself, though I wonder if that will make it hard to get sufficient diffusion. I was thinking of multiple layers of translucent plastic, each separated by some inches, with a light source behind that.

@nem I’m wondering how we can determine whether the a “flat” field is “really” flat without expensive equipment, especially since we know that the lens will introduce vignetting, so we expect that the corners are darker than the center. Just move the camera side to side between images? What can we achieve “in the lab” by taking photos with the lens off or severely out of focus?


I see you are speaking of your dome location in the past. Let’s chat regarding moving forward with the enclosure design we spoke of. Could we not design it with the inside painted black and make the unit light tight for bias and dark frames? Could we also attach a transilluminator inside for flat fielding?

I think if there is some sort of motorised doors or flaps to block light entering the enclosure, then the darks can be taken within the unit itself, if it were painted black and light tight like Tom suggested. Also, something like this could be considered - https://www.teleskop-express.de/shop/product_info.php/info/p9404_Lacerta-LED-Flatfield-Box---Diameter-185-mm---with-dimmer-0-100-.html. These could probably be made quite simply. This could be integrated to the motorised doors as well which would save there having to be any fully external parts to the system, keeping it compact.
Let me know what you think.