Help me improve this M42 image!

[michael.h.f.wilkinson]

Flatfielding is certainly the better solution to your problem

[Matthew.Blake]

Presented for your amusement - my first attempt at adding flats. At least it's a different gradient problem...

[Matthew.Blake]

For a bit of fun, I thought I'd use Photoshop's magic wand selector to select the background of my original image, apply some blurring to it and use that as my flat. This is the result. It's not brilliant, but I can make out much more of the detail of the nebula.

[michael.h.f.wilkinson]

For a bit of fun, I thought I'd use Photoshop's magic wand selector to select the background of my original image, apply some blurring to it and use that as my flat. This is the result. It's not brilliant, but I can make out much more of the detail of the nebula.

I tried something similar with the image you posted, but it contains many zeros in the background. This leads to div-by-zero errors in the flatfielding process. Are you simply subtracting the background, or are you dividing by the estimated white frame?

[Matthew.Blake]

I tried something similar with the image you posted, but it contains many zeros in the background. This leads to div-by-zero errors in the flatfielding process. Are you simply subtracting the background, or are you dividing by the estimated white frame?

I wouldn't know the difference!

[Starman]

While there's no substitution for getting the best light and calibration frames to obtain the best results, the image you posted still has a lot of great detail. I manually created some artificial flats and had a quick go at pulling out a bit more detail. It's a bit close to the threshold to be honest and looks quite noisy on my brighter laptop screen but ok on my normal desktop. Some nice detail in the nebula captured.

[michael.h.f.wilkinson]

Presented for your amusement - my first attempt at adding flats. At least it's a different gradient problem...

Can you post your flats so I can check the differences between subtraction and division?

[Matthew.Blake]

Can you post your flats so I can check the differences between subtraction and division?

How many of them do you want? I took lots to make sure the average was reasonable as I used my monitor for a light box so there was some variation between each flat.

[Matthew.Blake]

This is the master flat that DSS generated from the set that I took.

[michael.h.f.wilkinson]

This will do, I will get back to you later

Michael

[michael.h.f.wilkinson]

Here is my attempt using pixel-wise division (top) and your original. Though there is some improvement, it is not really perfect. This is partly because the original already seems to be clipped to black (0,0,0) at the right-hand side of the image. Some processing using curves, such as contrast stretching is present.

The reason you need to divide is that vignetting is a multiplicative effect, because it changes the effective sensitivity of each pixel, unlike dark noise, which is additive. Thus, a dark frame needs to be subtracted, but the image needs to be divided by the white frame (and multiplied pixel-wise by some value to retain brightness). Because the effect is multiplicative, it must be done AFTER dark-field subtraction, but before ANY other processing.

[Matthew.Blake]

I think the best thing for me to try now you guys have helped imrpove my understanding may well be a new set of lights, flats and darks for M42! All I need is a clear night... Watch this space.

[michael.h.f.wilkinson]

While there's no substitution for getting the best light and calibration frames to obtain the best results, the image you posted still has a lot of great detail. I manually created some artificial flats and had a quick go at pulling out a bit more detail. It's a bit close to the threshold to be honest and looks quite noisy on my brighter laptop screen but ok on my normal desktop. Some nice detail in the nebula captured.

I Tried killing the noise with the k-flat area attribute filter both in the version of Starman and in my earlier attempt. Hows this:

[Matthew.Blake]

It's really quite interesting to see how much more information you can get out of the 8bits/ch image than I can get out of the 16bit/ch image! Just shows you the benefit of experience. I've just got to wait for this wretched cloud to clear so I can get my next attempt in.

[michael.h.f.wilkinson]

It's really quite interesting to see how much more information you can get out of the 8bits/ch image than I can get out of the 16bit/ch image! Just shows you the benefit of experience. I've just got to wait for this wretched cloud to clear so I can get my next attempt in.

You could send me the 16bit/channel version (png supports that). Regarding experience, I am not a particularly experienced deep sky imager, but it's my job to develop new image processing techniques. Of course you could not have known this technique, it has not officially appeared in the scientific literature.

[Matthew.Blake]

Attached (hopefully) is a RAR containing a 16bit/ch TIFF of the central portion of the picture. I've had to reduce the size of it somewhat to get it down to an allowed size. I've also had to put it into a ZIP file as RARs are not allowed!

M42 low res.zip

[michael.h.f.wilkinson]

Sorry for the delay, I was just processing some of my own stuff. I did not get the flats of the same area, so could not do proper flat-fielding, but a rolling ball filter does get out more detail (after contrast stretching in ImageJ) the latter stretches rather indiscriminately, so the colour balance needs a bit of work. But there is much more detail than before.

The zip contains a rar which contains the tiff file (16 bit). The 8-bit png is just for a quick view.

M42 low resrollingballstretch.zip

[Matthew.Blake]

Awesome! I'm hoping tonights clear sky will give me an opportunity to get some more data. I'm going to try collection some flats on the field as well, will be interesting. I might even try defocussing the telescope and getting some darks with the cover off (dark-glows?) to see if I can use those to subtract the light pollution whilst leaving the stars as they won't be in focus. I'll be in a better site tonight as well (Coombe Abbey).

[Matthew.Blake]

I've posted my next attempt with flats and darks in http://stargazerslounge.com/imaging-image-processing-help-techniques/96355-m42-sgl-improvements-what-next.html#post1368197 if anyone wants to continue helping me, please take a look!