How to create Flat files for Astrophography - Need help
Hello,
I have created Flat files for the first time and need some advice. How do I know they have been created correctly?
I have a Sony A7 connected to a Skywatcher ED100. I have placed a white table cloth over the end of the ED100 and placed a laptop with a white screen in front of it. I have taken 30 photos each at different exposures between 0.5secs and 2.5secs at ISO1000.
Using Nebulosity to measure the results.
0.5 Sec Exposure. Mean: 6245, Min: 2488, Max: 9652
1.0 Sec Exposure. Mean: 11933, Min: 4696, Max: 16372
1.6 Sec Exposure. Mean: 14678, Min: 7276, Max: 16372
2.0 Sec Exposure. Mean: 15368, Min: 9076, Max: 16372
2.5 Sec Exposure. Mean: 16114, Min: 11572, Max: 16372
I also did one with a LED torch in front, its results are
2.0 Sec Exposure. Mean: 16364, Min: 15860, Max: 16372
Can someone advise me on the results I should be aiming for. The LED torch images look saturated while the pictures taken with a laptop white screen gives me a variance of greys.
Kind Regards
Graham Galea.
Essentially the light should be as " white " as you can make it. That way it will have an even effect over all channels. This is obviously hard to get perfect as LED lights vs Torch light vs Computer screens all give off a different tone. A white screen on the laptop though is a pretty good white so good choice there.
The flat needs to be taken at the same ISO / gain as the image to which its being applied and also the same focus. The idea behind this is that a flat is designed to compensate for vignetting and imperfections in the image train. If the focus changes, so does the focus on the imperfections. If the ISO changes same deal.
Lastly and the bit that is not quite correct although you still may have captured by accident is you need to allow the camera to decide the exposure length to get the correct level of exposure. Too long and it I'll wash out some imperfections, too short and it won't have a strong enough effect. Set your camera to AV and take the shots. Don't set to full auto otherwise your camera will change the ISO also.
Thank Ryan,
I have learnt a few things today.
1. White LCD screen ok.
2. Choose an exposure so not to saturate. After further tests today, exposure at 0.5 sec and 1/3 sec are giving results within the expected Mean / Max boundaries for my setup. 1/4 sec is too low.
0.5 Sec Exposure. Mean: 6245, Min: 2488, Max: 9652
1/3 Sec Exposure. Mean: 4860, Min 2004, Max: 8660
3. Take Flats at same ISO. I have settled on 1000 ISO for my photos. I have used the same ISO setting for Lights, Darks, BIAS and Flats.
4. Use same focus. I haven't done this. Good lesson learnt. I'll apply this rule going forward.
5. Allow the camera to decide the exposure length. I haven't done this either, instead tethering my Sony A7 to my laptop, setting the camera to manual and choosing the shutter speed from the Sony Camera control software. I'll also explore this option going forward.
My setup and 1/3 exposure results attached.
Something not mentioned about the white clothe is it needs to be taut and uniform across the end of the telescope/lens, no sagging or wrinkles or overlaps as these introduce false artifacts that will get removed from your shots.
Second the screen needs to be flat/perpendicular to the end of the setup. In your shot above its angled to the end of the scope, this will introduce a false gradient.
Instead of white clothe you may get better results making a "filter" to slip over the end of your scope with say a piece of plumbing pipe with a piece of stiff tracing paper glued across the end. I've used tracing paper for years for making light tents for diffuse light photography.
Of course you really must take flats at the time you take subs as dust on the imaging sensor changes if you are putting a camera body on scope then using for regular photography too. Only if the body is 100% dedicated and permanently remains attached will flats taken at other times be of use for just removing "dust on sensor" artifacts only. "dust on the end of the lens" "front refractor element" or "mirrors in refractor" you can expect to change all the time as these are typically exposed to the air, even if you clean each time stuff ends up settling out of the air anyway. Which is why its important to take flats as part of the imaging session rather than another time as the dust pattern will differ.
Something not mentioned about the white clothe is it needs to be taut and uniform across the end of the telescope/lens, no sagging or wrinkles or overlaps as these introduce false artifacts that will get removed from your shots.
Second the screen needs to be flat/perpendicular to the end of the setup. In your shot above its angled to the end of the scope, this will introduce a false gradient.
Instead of white clothe you may get better results making a "filter" to slip over the end of your scope with say a piece of plumbing pipe with a piece of stiff tracing paper glued across the end. I've used tracing paper for years for making light tents for diffuse light photography.
Of course you really must take flats at the time you take subs as dust on the imaging sensor changes if you are putting a camera body on scope then using for regular photography too. Only if the body is 100% dedicated and permanently remains attached will flats taken at other times be of use for just removing "dust on sensor" artifacts only. "dust on the end of the lens" "front refractor element" or "mirrors in refractor" you can expect to change all the time as these are typically exposed to the air, even if you clean each time stuff ends up settling out of the air anyway. Which is why its important to take flats as part of the imaging session rather than another time as the dust pattern will differ.
I generally peak flat illumination around 70% on the live view histogram and take bias and flats at the camera's lowest ISO - which minimises ISO noise in older cameras. Very basic approach that works well without too much fuss.
This is something I have got to get right. I have so far been processing without flats as I can see they are having the wrong effect.
What is happening to me is that the correction of vignetting is too strong, the 9.25 SCT with 0.63 reducer shows significant vignetting at the edge of my sensor frame, but when I have tried using flats I have ended up with it over-corrected so the image brightens at the corners instead of fading out. I presume any dust bunnies would show the same effect but the sensor was pretty clean at the time I last tried so it was not obvious.
I made one out of an Electroluminescent panel and the cardboard box that it arrived in.
One thing to be aware of; It's amazing how quickly light changes during the day. It may look stable, but I was doing some tests and came to realise my results were skewed because the natural light was changing. No clouds or anything, just the sun moving across the sky. I don't know how much effect this has on flats in practice because I wasn't testing for that. However, it did give me pause to construct my own flat box that works a treat and is repeatable - no figuring out exposures each time.
Oh, and another thing, if using an artificial light source (especially one that relies on pulses of light like most LED panels, EL panels etc) you'll need to make sure your exposures are long enough to smooth them out; no shorter than 0.3s.
Paul. I guess you have tried different flat illumination.
I more or less fiddled around with exposure time until dust bunnies and the expected vignetting was visible on an unstretched image but no part of the image was saturated.
I guess the next thing to try would be to fiddle around with different exposure times and try a single light, single flat image with each of the flat times to see if any work better. If one seems to be working well I should hopefully be able to work out by examining the flat what sort of peak and average/mean ADU works.
I can agree with the post above, using a single setting the results seem to change rapidly image to image when using the sky. I guess I should build a light box, but for a 9.25 inch SCT that won’t be particularly cheap. If I do I should size it for a possible future 10 inch newt.
This is mine. It's not pretty, but needs must and all that. At that size, EL panels are not cheap, but not too expensive either - about $100 with 12v Inverter Here's one on ebay
It's incredibly light, too, being mostly cardboard. So far so good, I just need to be careful to keep it out of the dew though.
I did put some lee diffusion filter #216 in there too just in case.
With some gaffa tape and a stanley knife I was able to make it sit on the RC10 tube nicely and even made an adapter that allows me to repurpose it for use on my skywatcher Epsrit 120mm. The light leaks are so minimal I can use it at a Dark Sky site without complaints (I can barely tell if it's on myself).
As mentioned, you need to be careful to keep exposure times long enough to smooth the pulsed nature of the illumination.
Another point, I don't think the exact ADU levels of exposure for the flats is that critical. The important thing is that neither the top or bottom of the histogram is being clipped, so you want to expose somewhere near the middle of your range. Also, since the times will be constant, it's not a bad idea to shoot the flat darks per filter as exposures won't be changing night to night.
One caveat, mine doesn't do well in the red end of the spectrum (maybe get a warm white panel, if you do). Eg, in S2 on my rig, I have to bump the gain to 313 on my rig and I still get exposures of 11.5 sec! Bit of a pain having to wait 5 minutes for flats for that filter, but Ha is half that, and O3 exposures are half a second (being green).
