Lights Darks and Bias frames

[rcheshire (Rowland)]

I've been learning how to process images using AstroStack - its simple to use and cross-platform - it works with Linux.

Last night I took 10 lights, 10 darks, 10 flats, all 64 seconds and 10 bias frames 1/4000th - noise reduction turned off.

I average the darks, flats and bias frames to produce a single image for each set. I'm not completely sure whether averaging the bias frames is necessary.

This is what I found, applying various combinations of flats, darks and bias frames, in AstroStack.

The darks removed virtually all the signal, whether I used a single dark or an average of all the darks - maybe more images and more signal would improve the SNR?

The averaged bias frame obliterated the image - using 1 was OK.

The best results were obtained using a single or averaged flat, but there was significant improvement with the averaged flat. The final image was noise free and there was adequate signal - some missing but not significant in this case.

Interested to know if I'm on the right track?. I'm inclined to think that it's a bit of mix and match to get the best result - there is no fixed formula?

An issue I have with AstroStack is that the saved jpg images are unreadable - cant open them in any image program. Fortunately it has a preview so I'm able to see the results in real time, but down to taking desktop snapshots of the previews. Maybe I'm missing a plugin?

[Octane (Humayun)]

Bias frames are unnecessary. The bias is contained within the darks itself. Bias frames are only required when you're matching different exposure length darks with differing light frame lengths. I personally feel that this should be avoided unless using a good quality CCD camera where you can manipulate data with pixel math.

Flat frames should facilitate a histogram on your LCD preview which peaks 1/3rd to half-way across the histogram. They should be taken at the lowest ISO your camera can. If your flat lights are 1/100th of a second, then so should your flat darks be.

Try using some other software under a Windows virtual machine if you really must run Linux.

H

[Bassnut (Fred)]

You probably arnt, but just in case, are you taking lights and darks as JPegs?. That wont work, the compression wont allow subtracting jpeg darks from jpeg lights, use RAW only.

Are you using the file calibrate feature?, rather than just combining darks/flats/bias and lights?, just wondering, as you say you are tring "different combinations", and the darks are killing signal, dosent sound right at all.

[leon]

I reckon your best bet is to turn ICNR on 2, forget about Bias and then take your lights through the same optical train you are using.

If nothing changes in your optical train than those lights will suit until you do change something.

Always go with ICNR on, it might take twice as long to do your image run, but you will be pleased with the results.

Leon

[multiweb (Marc)]

If you have access to a program that does pixel maths you can test the following. Subtract the bias from your flat frame then divide your sub by the result and you should have flat fielding working properly.

The bias are contained in any darks you make but you will need to subtract then for flat fielding.

[rcheshire (Rowland)]

Thank you. I really appreciate the feedback - this has elevated my understanding enormously.

I'm shooting RAW and will have a look for a pixel maths program, and calibration of RAW images - something I've missed, I think.

I'm not familiar with ICNR and will look that up.