Help with flats please -CCD Imaging

[Zakalwe]

I recently built a lightbox, to try out taking flats, and put it to use last night. I used the Russell Croman type of lightbox, and used some greasproof paper to diffuse the light source (battery powered LEDs), similar to here

The exposure was set at 2 seconds...any higher and FITS Liberator was showing the peak value at 65355, which I assumed meant that the flats were overexposed.

I added ten of the flats into DSS along with 14 x 240s lights, but the result was terrible, with lots of noise. The stacked image was much better without the flats, than with.

Can I have attached a jpeg version of the flats to this post (fit file loaded into FITS Liberator, saved as .tif, then loaded into PS, levels stretched to move the LH slider to the beginning of the curve, converted to jpeg and reduced by 50%). I have loaded one of the native .fit flat files into my Dropbox here.

Would someone mind casting their eye over the flats? Am I somewhere near right with the exposure, and does the flat look OK? I'm clearly doing something wrong, as the stacked image looked much worse with the flats, than without.

Thanks in advance for any help offered.

[kookoo_gr]

make sure that the diffuse paper you use is completely flat (it has no physical ripples) so that any imperfection of the paper does not show at the image. the leds at the page you provide do not contain the whole spectrum of the light so it may be a problem and being at the center does not provide a completely flat image. at my first lightbox i used ιncandescent light bulbs and i placed them at the periphery of the main hole so i will have a flat image.

now for the exposure times i found out at my atik 16 ic that at 2 secs exposure i burnt the image as well so i decreased the exposure time so the histogram will be near the center of the image. the exposure time i ended up with was about 0.1 secs at 1200mm FL. i suggest first to reduce the exposure time

[Zakalwe]

Thanks for the reply. I have built the box so the LEDs shine directly onto a white card, and the light is then reflected onto the diffuser, rather than having the light shining directly onto the diffuser, in an attempt to create a flatter field.

However, looking at the paper that I used for the diffuser, it is not totally uniform...it has a coarse texture. Would that cause an issue, or would the fact that it is physically located so close to the objective mean that the texture is totally out-of-focus?

[Paxo]

One point to remember when using flats is that you either need to take dark frams for the same exposure - "dark flats" or a minimum of bias files must also be used (this usually is sufficient as the exposure length is so short). Without it you are then adding noise to the final image - DSS takes care of this automatically if the bias or 'dark flats' is added.

The flat you have shown looks very uneven, and without any vignetting (which you would proably expect to see). What telescope did you use? You should be looking for an average of 22,000 - 24,000 on the pixel value (40% ish of the full well value of 65,000). At this level you would probably epect to see a brighter centre with a drop-off towards the edes.

Any grain in the tissue used should be so out of focus it shouldn't matter. A lot of people use white T shirts or even an LCD monitor showing white. 2 seconds seems long to me for the exposure but it is difficult to say as I don't know how intense the lights are or how thick the tracing paer is.

Hope this helps a little.

Steve

[Zakalwe]

Thanks Steve.

The 'scope is an Equinox 80, with a QHY8L CCD camera.

I think you have identified where I have gone wrong.....I did not add in any darks or to the DSS file list. I'll go and take some and see what the results look like.

[Zakalwe]

I now realise that I am amazingly thick at times! Of course, with no darks or bias all I am doing is adding noise to the image (facepalm moment....).

I have set the CCD off running on 20x240 second darks, and have also taken 20 x 0 second bias frames (the beauty of a CCD with setpoint cooling!). Once I have the files, I will restack and see what the results are like.

[Paxo]

Let us know how you get on.

Steve

[Zakalwe]

Steve,

Much, much better. Stacking with bias and darks took out a few hot pixels and improved the detail. I'm not convinced that I am doing the flats correctly....I might re-take them with a slower exposure.

This was the resulting image, and with just an hour's exposure (4 minute subs) in a not-quite-dark sky I am pretty happy with this.

Thanks for your help:icon_salut:

[Paxo]

Let us know how you get on with the shorter exposures. When I get on my obsy pc next I'll post up a typical flat from my system, there will be more vignettig on mine but it will give you an idea.

Steve

[Zakalwe]

Thanks Steve.

The CCD is finishing off a series of darks , so I'll give the flats another go to see how they turn out with 1 second exposures. Obviously these newer flats won't be any use for calibrating Friday night's images (camera was taken off the 'scope since then), but it'll show me if I am barking up the right tree.

Thanks again for your help.

Regards

Steve J

[Paxo]

Here's my master flat from about a month ago. This is from a full frame 35mm chip, so it takes a lot of illuminating, through a 75mm scope. Most of the vignetting (and the dust specks) comes from the filter wheel. The dark rectangle at the bottom comes from my off axis guider. The average pixel value was 24,000 on this, it looks dark but it performs very well.

Steve

[roundycat]

In post #6 above you say that without darks or bias your flats will be adding noise to the image. Whenever you subtract darks or bias or apply a flat you are adding noise to the image. That is why we shoot quite a few of each calibration frame, to minimise the added noise.

You should make at least a simple attempt to establish the photon transfer curve for your camera. That will tell you when it starts to go non-linear. If you go above that point your flats will not properly correct. The histogram does not tell you everything.

Dennis

[Zakalwe]

I've re-taken some at 1 second....they are starting to look more like a flat now. The optics look clean as the kit is all pretty new (all bought this year).

Last question, if I may? I use Deep Sky Stacker for stacking. Once I stack the files, it creates a Master Dark and Master Bias. Can I delete the bias and darks, and just add in the Master version from then on?

Thanks again

Steve J

[Zakalwe]

You should make at least a simple attempt to establish the photon transfer curve for your camera. That will tell you when it starts to go non-linear. If you go above that point your flats will not properly correct. The histogram does not tell you everything.

Dennis

Thanks Dennis,

How do I do that? And is 20 flats/bias/darks a decent number to shoot for? I can easily add more if required (again the beauty of cooled CCDs!)

[roundycat]

what you need to do is to take a picture of an artificial light source (a very small one) or shoot a star at different exposures and measure each succeeding exposure to establish the ADU of the object. Doubling the exposure will double the ADU up to the point where it starts to go non-linear due to the anti-blooming gate on the chip. Even if you use a non-anti-blooming camera this is useful because the excess exposure willl bleed off the charge well and form a blooming spike. The light source does need to be quite dim.

I use 20 darks and flats (bias subtracted flats normally) and thirty bias. Lots of users shoot more than that but 30 gets you down to about 2.5% added noise and if you add that to the light frame there will be virtually no difference. See here http://www.dens-astropics.org.uk/page%2024.htm

Dennis

[Zakalwe]

what you need to do is to take a picture of an artificial light source (a very small one) or shoot a star at different exposures and measure each succeeding exposure to establish the ADU of the object. Doubling the exposure will double the ADU up to the point where it starts to go non-linear due to the anti-blooming gate on the chip. Even if you use a non-anti-blooming camera this is useful because the excess exposure willl bleed off the charge well and form a blooming spike. The light source does need to be quite dim.

Dennis

Thanks Dennis,

If I understand that properly, I can use this to judge how exposed the flats can be. Over this exposure level, and the flat is not corrected as it has "blown" the lighter parts?

The QHY8L has -100Db anti-blooming (whatever that means:icon_confused:). How do I judge when it has gone non-linear?

[martin_h]

Which capture software are you using?

I use Nebulosity2 and aim for a ADU reading of 22-25000.

The image shows my capture screen, you can see the histogram on the upper right showing about 1/3 across and the pixel stats box showing the entire image mean pixel count of 23544

[AndyUK]

This is very useful information for me too - Thanks all...!

[Demonperformer]

I don't tend to take flats, so regard my comments in that light.

I'm sure I've read somewhere that you should take half your flats, then rotate the light box through 90 degrees and take the rest, putting them all into DSS together, as this eliminates, in the master flat produced, any variations produced by the light source itself?

I'm sure one of our AP gurus will say (very nicely) that I'm talking rubbish if I dreamed it .

[Zakalwe]

Which capture software are you using?

I use Nebulosity2 and aim for a ADU reading of 22-25000.

The image shows my capture screen, you can see the histogram on the upper right showing about 1/3 across and the pixel stats box showing the entire image mean pixel count of 23544

I'm using EZCap, the software that comes with the camera. I'm sure there is a function in there to tell me what I should be looking for...