flats and noise

[alcol620]

Any suggestions on the following: I am imaging with an Atik383 on an ED80 scope and taking flats on a home made light box (which seems to work OK - it seems to get rid of dust motes and vignette). I took 4X600s light subs this evening before clouds arrived and have just noticed that when I calibrate them, they appear noisier after calibration than before. For illustration I have stretched the images in Pixinsight and added them below. By a process of elimination it looks as if most of the noise appears when the master flat is applied to the calibration - see image c_2 below - the master flat whilst adding noise has removed the dust mote . Simply applying a master bias and a master dark leaves the image much as it was out of the camera (bottom image) dust mote still there as no master flat applied. My flats are taken at an ADU of around 40000, I read somewhere that this was the sort of figure to use for flats when using an Atik 383 camera?

Any ideas what I might be doing wrong or is this normal?

 

 

[Uranium235]

40,000 is a bit high to be honest. Try 20-22,000 and cool the camera to the same temp as your lights - otherwise you get a false reading.

[alcol620]

Thanks Rob. Yes camera temperature at the same level as the other subs. Currently at -25/26C .  Camera is still in situ so I will take some more tomorrow at lower values. Since posting I did wonder whether what I see on the "noisier image" is simply darker pixels representing the background of the image?

Any other 383 users with views on flat exposure levels.

Regards Alec

[Uranium235]

It might be worth just stacking them and seeing how it looks. Sky noise can and does differ from sub to sub, so just flick through each calibrated sub to see if they are similar.

Another thing you can try is using DSS to calibrate your images (its what I use) and compare them, all you need are the bias and flat frames - and you can get it to output calibrated lights (filename.cal.fit) to the same folder where you lights are stored. No need to stack, just registering them is enough to perform the calibration.

[dph1nm]

What makes you think it is noisier? Visual appearance can be very deceptive, and the levels on those two shots are not the same. Looks to me like faint objects are easier to pick out on the left hand image.

NigelM

[ollypenrice]

Are you dark-subtracting your flats? If you don't, they'll introduce noise into the image. You don't need to shoot dedicated darks for flats, you can just use a master bias as a dark for flats. Rob, I don't see why exposing to 40,000 over 23,000 would add noise to the flats. I'd rather expect the reverse. (But there are lots of things I don't see!) The problem with going too high is rather to do with losing linearity (as I understand it.) Personally I don't go as high as 40,000 but I don't see that as being to reduce noise so much as to stay within the camera's most linear range. So far as I know the only thing that's special about the 8300 and flats is the mechanical shutter. Exposures need to be long enough not to include a gradient caused by the 'wipe' of the blades. This just means dimming the lightbox.

Olly

[alcol620]

Thanks Guys

Much to think about following your inputs. It could be that I am just looking for problems. It was only that when I looked at the original image it looked very "smooth" and when I calibrated it (using Pixinsight with a master bias, , master dark and master flat) it looked grainier - as per the images I put up. I took this graininess to be noise - perhaps it isn't. This is when I calibrated the sub again with and without bias, dark and flat masters and the only 1 that seemed to have an effect was the master flat. The flats are dark subtracted.

Regards

 

Alec

[alcol620]

Thanks Guys

Much to think about following your inputs. It could be that I am just looking for problems. It was only that when I looked at the original image it looked very "smooth" and when I calibrated it (using Pixinsight with a master bias, , master dark and master flat) it looked grainier - as per the images I put up. I took this graininess to be noise - perhaps it isn't. This is when I calibrated the sub again with and without bias, dark and flat masters and the only 1 that seemed to have an effect was the master flat. The flats are dark subtracted.

Regards

 

Alec

[alcol620]

By way of feedback. The Rosette image (my icon) was taken with the existing set up so perhaps I should stop looking for problems that aren't there and get on with imaging?

 

Regards everyone

 

Alec

[gnomus]

I see that these images have all been 'auto-stretched' in PI (I assume we are looking at a screen capture).  I don't know how PI decides how much it should stretch a particular image.  It must surely stretch images with different light 'levels' differently.  A calibrated image will not be the same as an un-calibrated one.  Therefore, is it possible that PI might be auto-stretching your calibrated images differently from your un-calibrated ones, making them appear noisier when they are not?  You could check this by applying the same histogram transformation to each of the linear images to see whether what you are seeing is real or a product of different auto-stretches.

Having said that I don't see much wrong with the calibrated image.  (I prefer the image without the splodge ).  4 x 10 minutes is not a lot of images or capture time.  The best way I have found to reduce noise is to capture more subs.

FWIW, I shoot my Atik 383L flats at ~24000ADU.  I calibrate my flats using a superbias and a master dark that is shot at the same length of time as the flat (..... I have no idea if this is optimal or not, but it seemed to make better sense than using a master dark that might be 20 minutes long for a 10 second flat exposure).  I 'manually' pre-process my images following something very close to this: http://www.lightvortexastronomy.com/tutorial-pre-processing-calibrating-and-stacking-images-in-pixinsight.html   It seems like a lot of work, but actually it doesn't take all that long once you've been through it a couple of times.  I'm open to suggestions if there is a better way, however, because I don't really know what I am doing. 

Regards

Steve

 

 

[alcol620]

Hi Steve

Thanks for your input. I have decided to fall more in line with inputs on here and reduce my flats to around 25000 ADU. Noted your point about PI stretching images differently in which case close comparison of individual images is pointless and should only be used to give a general idea of what the image looks like. I noted your comment about using short darks to calibrate flats - must be more accurate than scaling darks of considerable length? A lot of people suggest not bothering with darks at all when calibrating flats!

I also use Kayron's tutorials (LightVortex) which I consider to be an excellent source of information. 

Regards

Alec

[alcol620]

Just a bit of clarification. The images at the start of this thread represent only one of the four 600s subs I took. 

Have followed Steve's suggestion and equally stretched the calibrated and uncalibrated images using PI HistogramTransformation. The graininess of the images is now more similar and the calibrated image has a darker background and has more definition. So I assume that Steve's suggestion is correct that the PI "auto stretch" treats calibrated and uncalibrated images differently?

Regards

 

Alec

[alcol620]

Tried a stronger PI HistogramTransformation stretch on the calibrated and uncalibrated to look for differences. The attached images show the difference, each image being stretched with the same HT setting, albeit pushing the black point and losing about 1% of the pixels to emphasise the difference. The image on the left is the calibrated image. Note: 1 single 600s sub.

 

[andrewluck]

There are some persistent misconceptions around that calibration frames remove noise. In fact, they will all add noise to an image! But, by stacking multiple calibration frames you do get to choose how little noise you want to add.

This presentation has some issues with text missing from some of the slides but is a very good introduction to the subject: http://www.scienceandastronomy.com/wp-content/uploads/2013/11/ASAE-Calibration-131114-B.pdf

The original presentation was given at MWAIC 2009. Videos were available from http://mwaic.com but the magazine has been defunct for quite some time so I would touch base with them before ordering if you want a copy.

Andrew