A problem with flats

[Roy@Aldermaston]

I've mostly used artificial flats with the subs from my 1100D, but its about time that I got to grips with the real thing. So far my efforts have worked well at removing dust bunnies, but have been less successful with vignetting.

The problems is this: by lights show the usual pattern of vignetting, being bright at the centre of the frame and fading in a circular fashion towards the edges. Not so with the flats: the centre of the vignetting is displace towards the bottom of the frame, such that when stacked the final image shows a horizontal gradient along the top of the picture. I've tried the "tee shirt" method and using a laptop screen to provide illumination, but either way I end up with the same problem - any ideas. I've attached a stretched flat to show the effect.

Thanks,

Roy

[rfdesigner]

If you're using darks, then flats need their own dark, the bias frame, or they don't quite work.

Also flats need to be about 50% of full scale, and you need quite alot of them (I'm skimping with 10, ideally I'd go for 20)

If you're shining a point source light through the shirt then that point source must not be in line with the scope or the scope might still resolve it.

Finally, light pollution tends to produce as a gradient in the image. is the gradient more obvious in the red channel than in the blue?

Derek

[IanL]

You need to check whether your flats are, well, flat What to do depends on your source for generating the flats.

- First make a set of flats.

- Now if you are using a laptop screen, light box, EL panel or similar self-illuminating source, keep everything the same and rotate the light source 90 degrees and make a second set of flats.

- If you are using a fixed light source, like a t-shirt and natural light, then you need to rotate the OTA and camera 90 degrees (without moving the the camera relative to the OTA or re-focussing), and again make a second set of flats.

- Now stack one set of flats together to make a master flat, and stack the other set of flats to make a second master flat. (For a quick test you could just use one flat from each orientation but it won't be as accurate or clear if there is a problem). Now subtract the second flat from the first using your processing software and then stretch it as you did with the example.

Either method should produce two sets of identical flats if your light source is providing even, flat illumination. By subtracting one flat from the other and stretching, if you see a gradient or other pattern, that tells you that your flat light source is not flat at all. A perfect flat subtracted from another perfect flat should end up with a flat result image (it might be black, grey or have a colour cast depending on your light source and whether you got identical exposures in both orientations, but that isn't the issue, you are looking for it to have the same average brightness all across the frame). We have't calibrated the flat (yet) so it will be noisier than usual, but if you use a tool that averages a pixels in a small area and take measurements across the frame you should get a good idea of the % variation from place to place, ideally as close to 0% as you can get.

With the daylight/t-shirt method it is very hard to avoid directional light entering the scope and causing a gradient, even on a cloudy day.

With artificial light sources, again they may not be uniformly illuminated causing similar problems. As a DLSR user you may next want to splt the RGB channels and examine them separately. If the blue and green channels look okay but the red is not, you may need to work on your light source to get more red in to the exposure. I always find the red channel tends to be under-exposed in my flats (working on a revised lightbox to try to fix this).

Once you have tested all that and ironed out any problems, you can then work on calibrating your flats properly with bias and flat darks. These can introduce other unwanted effects which is why I think it is best to leave them out of the equation at first until you know your technique works.

[ArmyAirForce]

I made this light box which improved my flats. Just a car tail light bulb about 6 inches behind a translucent sheet of white plastic, in a plywood drum. It runs from one of my dew heater outputs. It fits over the end of the scope, so I can take my flats at a dark site without blinding everyone!