How to get the correct exposure length for flat frames?

[Hawklord]

Hi All,

When I recently first started RGB imaging on my VC200L with flat field calibration I guessed at the exposure length for the flats. I thought that as long as they looked a bit like the ones that I had seen others using then they would be OK! However when I switched to using the refractor and calibrating with new flats they were not up to the job with vignetting in the resulting stack remaining.

Camera is the Sbig ST8300M

So, my questions are these, and apologies if the answers are obvious:

I’m told I need to aim for an ADU value of ~20,000 for my flats, is this the peak intensity (of the central portion say) or the average across the whole frame?

On loading my flats onto CCDSoft and Nebulosity it gives one an intensity (or I) value for the pixel the cursor is on (and in the case of Nebulosity, the local area and average also). Is this the same as ADU?

So can I use these programmes to tell me whether I've got the exposure about right?

All advice welcome.

Cheers

[brianb]

I’m told I need to aim for an ADU value of ~20,000 for my flats, is this the peak intensity (of the central portion say) or the average across the whole frame?

There shouldn't be an immense amount of variation (apart from a few dead pixels which may be very bright or very dark more or less independent of exposure). Take your flat frame, apply a median filter with a radius of about 5 pixels and pick a value.

There's no need to be accurate about exposure for flats. If you underexpose them badly they'll be noisier than they need to be, and the noise will transfer to the finished image, which is why you want to keep the exposure up. But you certainly don't want to be pushing the brightness into the zone where the response is non-linear, since that makes flat subtraction not work properly. About 1/3 to 1/2 full well is usual for CCDs with anti-blooming gates; without the anti-blooming gate, you can go much higher, response should be linear almost up to 100% full. And if you camera has a switchable blooming gate, the flats should be made with the same switch setting as the "light" (image) frames.

[martin_h]

I use Nebulosity, I use View-Pixel Stats, look at Entire image: Mean, and aim for 20/23000

Works for me

[Hawklord]

Brilliant. Thanks

[msinclairinork]

Sorry to hijack your thread Wayne.

Probably a silly question but I assume that you would have to find these values each night to account for focus and tempreture? or not?

Thanks,

Michael

[brianb]

you would have to find these values each night to account for focus and tempreture?

Exposure for flat frames is "fixed" unless you change something major like the light source but flats need to be redone if you change anything ... even if you could guarantee you'd mount your camera in exactly the same orientation and that focus would be exactly the same, the dust donuts would be different. A change in temperature is probably OK to ignore unless you need to tweak the focusung.

[msinclairinork]

OK brian I think I understand.

So you could find the best exposure time for your flats inside the house one night.

Then every imaging night just take a set before packing up but at that exposure time you set whilest in the house?

Thanks

Michael

Edited February 10, 2011 by msinclairinork

[Hawklord]

OK brian I think I understand.

So you could find the best exposure time for your flats inside the house one night.

Then every imaging night just take a set before packing up but at that exposure time you set whilest in the house?

Thanks

Michael

That's pretty much what I did for the larger scope. Just had to remember not to alter the focus or orientation at all. For the refractor it's a bit easier since I can use the laptop screen outside before packing up.

I'm assuming that as long as the image train is the same, that each new flat can be the same exposure as used before, with just a cursory check at the intensity level.

Cheers

[msinclairinork]

thanks for that.

Michael

[cardiofuse]

On my FLI camera; I can have an average pixel value between 25000 and 40000 and they both seem to work very well.  I'm assuming this must be in the  linear range of the CCD....