Help With Circles in Images.

[Spacehead]

Hi All.

I have a lot of images of the Pelican from the other night. 
(I cropped these just for upload)
When I stack them, I get the image below (note I have used fade correction to bring out the issue of the rings).
A sample of a single sub also follows - again I have used fade correction to attempt to see if the rings are present in the individual subs - they arent)
So I can see the usual vignette.  But whats with these rings?
Any ideas anyone?  Stacked in DSS with no flats or darks - I would expect smooth gradient due to vignetting - but these rings are causing me a real headache.
The only thing I can think of is related to the shading of the images as the evening progressed (about 3 hours).
So additionally, please see the thumbs showing the range of brightness of the images - all taken at 1m and 400iso - but changes in the night sky show up as it gets deeper into the night.
(starting top left ending bottom right in the thumbs shows the gradient as the night moved on).
So I am thinking these rings are related to this changing overall brightness throughout the set?
Help!!

[dph1nm]

Posterization, caused by compressing a 16/32 bit image down to 8-bits?

NIgelM

[Spacehead]

NIgelM, thanks for your reply.

I have checked my input files, these are CANON RAW and showing as having a 48 Bit "bit depth" in Paintshop Pro X8.

Checking the output from DSS, I find Paintshop Pro X8 is reporting a bit depth of 48 Bits also, on the file dss creates.

One thing I note is that the resolution on the source files is 72dpi, whereas on the output file it is 100dpi - I don't know if thats related?

At the point before I save the stacked image, so when DSS has finished and just showing the result on the screen, the data is already "corrupt" as in the image below - which is a screen shot of the DSS screen image area.

 

[Oddsocks]

Your single sub does show linear horizontal banding and it looks as though D.S.S. has ‘created’ artificial structures made from a combination of the linear horizontal banding and noise in the vignetted field.

To be certain if this is a D.S.S. artefact try integrating the images in a different image processing package. Astroart would be a good application to try since it handles DSLR images well and the free evaluation period never expires, you just can’t save the results of any processing you carry out, only look at it on screen.

http://www.msb-astroart.com/default.htm

In your D.S.S. processed final image there are quite a few satellite trails and other outliers. As you have plenty of subs to work with, when you integrate ‘lights’ frames the integration method should be one of the kappa sigma reject algorithms so that spurious outliers such as aircraft and satellite trails are automatically rejected and don’t make it into the final image. Average or Median integration method should be used when you only have a few subs to stack and can’t afford to throw away any data or you are integrating a non-registered stack of frames to manually create a master bias, flat or dark.

If D.S.S. has created these artefacts from noise in the source frames try going back and adding a set calibration frames of twenty or so dark frames that match the ISO and exposure time of the lights plus forty or so bias frames. These can be created just with the camera off the scope as long as the camera aperture is capped to prevent light entering the camera. Even though the exposure conditions are not identical it may make a difference.

You could also create a synthetic flat from the existing lights, it is fairly simple with monochrome images, a bit more difficult with bayered images so all the lights and calibration subs would need debayering and converting to tiffs first, then create the synthetic flat from the lights and finally use D.S.S. to produce a calibrated image from the all-ready debayered colour tiffs. The process for creating synthetic flats is rather convoluted and most online descriptions are based around Photoshop but the process converts to Paintshop readily enough. Search Google on-line using the string: Synthetic Flats - Photoshop - Astrophotography and this will bring up many video tutorials and written descriptions.

But firstly, I would try processing the same data in Astroart and see if it comes out the same.......

 

 

[michael8554]

8 hours ago, Spacehead said:

I have checked my input files, these are CANON RAW and showing as having a 48 Bit "bit depth" in Paintshop Pro X8.

Checking the output from DSS, I find Paintshop Pro X8 is reporting a bit depth of 48 Bits also, on the file dss creates.

One thing I note is that the resolution on the source files is 72dpi, whereas on the output file it is 100dpi - I don't know if thats related?

I have PaintShop Pro X7 and it won't handle the 32bit DSS Autosave file, so I Save As a 16 bit image and PSP reports it as:

Pixels per inch = 100

Pixel Depth/color = RGB - 16 bits/channel   (which is 3 x 16 = 48 ?)

So your figures are normal I'd say.

I agree with NigelM that it's Posterization, which can happen if you have very dim subs that DSS has stretched ?

Michael