recovering overexposed flats

[seeker372011 (Narayan)]

I have two and a half hours of lights from some time ago that I must throw away because I overexposed the flats (operator error-blame it on tiredness at 2 am after a long day at work or alternatively on stupidity) and now can't get rid of the (horrendous) amount of dust and also vignetting

is there any way that i can save the situation?

thanks
Narayan

[Dennis]

Ouch – that is unwelcome news. I have read of a technique of generating synthetic flats by using the dust and scratches filter in Photoshop to “get rid” of the stars and then applying a Gaussian blur. If I remember correctly, this works best on images without extensive nebulosity, so maybe it’s only useful for galaxy type images?

Cheers

Dennis

[seeker372011 (Narayan)]

Originally Posted by Dennis

Ouch – that is unwelcome news. I have read of a technique of generating synthetic flats by using the dust and scratches filter in Photoshop to “get rid” of the stars and then applying a Gaussian blur. If I remember correctly, this works best on images without extensive nebulosity, so maybe it’s only useful for galaxy type images?

Cheers

Dennis

yes I tried that but it doesn't work unfortunately

[iceman (Mike)]

Can you reduce the exposure time using raw processing tools and then convert them to TIF?

or are they literally overexposed histogram white clipped?

[seeker372011 (Narayan)]

Originally Posted by iceman

Can you reduce the exposure time using raw processing tools and then convert them to TIF?

or are they literally overexposed histogram white clipped?

actually the histogram is not clipped and the ADU seems to be only of the order of 11,000-my target is around 20,000..maybe I have got it wrong and its under exposed and not over

totally confused here

the only thing I did different from my usual workflow is I forgot to bring the gain down when I was taking flats as I usually do..surely this should overexpose not underexpose..and visually I can see the dust etc on the flat -it just wont subtract out

[Dennis]

Flat Fields are DIVIDED.
Dark Frames are SUBTRACTED

[seeker372011 (Narayan)]

Originally Posted by Dennis

Flat Fields are DIVIDED.
Dark Frames are SUBTRACTED

oops

i meant divided of course

if it were simple subtraction I could do it in Photoshop .....

[monoxide]

nebulosity scales flats doesn't it?

[Dennis]

Did you either re-focus the 'scope, rotate the camera in the focuser or remove and replace the camera in the focuser between the light frames and the flats?

Doing any of these will cause the flats not to register with the lights and thus they wont work.

What software are you using to apply the flats?

Cheers

Dennis

[jase (Jason)]

No, don't throw any data away. Most is salvageable if you know how.

There are a couple of ways to resolve this.

• Sum combine: If the flat is under exposed and you haven't hit your ADU (which would appear to be the case from the information you've provided - 11,000 with a target ADU of 20,000). If you've got a few subs with 11,000, after subtracting darks, sum combine two of them. In this instance your target ADU will be 22,000. Once you've sum combined the individuals as pairs, you can then median combine the summed pairs to improve the flat accuracy.

• If your flat ADU is not too far off the target, you can use pixel math to resolve this. In its basic form, your data numerical. Don't loose sight of this fact as there are some powerful tools at your aid. Each pixel has a depth based on bit space. If you use pixel math you can increase or decrease all pixel values in an image to reach the desired ADU. This is sometimes known as data scaling, thought I don't particularly like this term as scaling is typically size sampling in the photoshop realm. Anyway, 20,000 / 11,000 gives a scaling factor 1.81. So if you scale the data by this value you will reach your desired ADU. Respectively, you can also downscale if your flats are overexposed.

There are many other uses of pixel math which I'll leave for another day.

[seeker372011 (Narayan)]

Originally Posted by Dennis

Did you either re-focus the 'scope, rotate the camera in the focuser or remove and replace the camera in the focuser between the light frames and the flats?

Doing any of these will cause the flats not to register with the lights and thus they wont work.

What software are you using to apply the flats?

Cheers

Dennis

Dennis, no I didnt refocus, rotate the camera or even touch it between lights and flats..course doesnt mean it didnt slip or something but I honestly dont think that's the case....I use Nebulosity to apply the flats

[seeker372011 (Narayan)]

Originally Posted by monoxide

nebulosity scales flats doesn't it?

yes I thought so, that;s what Craig says in the manula but I am not sure what happened here

[seeker372011 (Narayan)]

Originally Posted by jase

No, don't throw any data away. Most is salvageable if you know how.

There are a couple of ways to resolve this.

• Sum combine: If the flat is under exposed and you haven't hit your ADU (which would appear to be the case from the information you've provided - 11,000 with a target ADU of 20,000). If you've got a few subs with 11,000, after subtracting darks, sum combine two of them. In this instance your target ADU will be 22,000. Once you've sum combined the individuals as pairs, you can then median combine the summed pairs to improve the flat accuracy.

• If your flat ADU is not too far off the target, you can use pixel math to resolve this. In its basic form, your data numerical. Don't loose sight of this fact as there are some powerful tools at your aid. Each pixel has a depth based on bit space. If you use pixel math you can increase or decrease all pixel values in an image to reach the desired ADU. This is sometimes known as data scaling, thought I don't particularly like this term as scaling is typically size sampling in the photoshop realm. Anyway, 20,000 / 11,000 gives a scaling factor 1.81. So if you scale the data by this value you will reach your desired ADU. Respectively, you can also downscale if your flats are overexposed.

There are many other uses of pixel math which I'll leave for another day.

thank jase I am going to have ago at working with the flats and see what I can do -I am going to try rescaling first. I am not sure my target ADU is correct in the first instance.

thanks everyone for the suggestions- 2 and half hours of data is a crime to throw away
Narayan