6 replies to this topic

[Beast1987]

Hi all,

 

i'm relatively new to narrowband DSO imaging.

Equipment list:

• ZWO AM3 mount
• svbony 102ED scope. with flattener the effective focal length is about 568 mm, so ~F5.5
• ASI 533MM mono camera
  filter wheel,  antlia 4.5 nm SHO filters
• ASI 120mm guide cam and the 120mm F4 guide scope
• asiair plus for control and capture

I'm playing around with exposures times of filters, calibration frames, processing, etcetera.

 

Recently i have imaged the soul nebula IC 1848 in SHO using  this process, imaging Ha for 300s,  Sii for 300s, and Oiii for 300s.

I've captured roughly 1.5hr per filter for now and stacked them using DSS using dark, flats and dark flats.

 

After only basic stretching in photoshop i find that the Ha looks perfectly fine and sharp.

The Oiii and especially the Sii however show these kind of "jumps" in between area's that have more vs less signal (i'm not sure how to call this effect, i'm sure it has a name), see attached pic for a zoomed in sample.

 

My guess is that there is not much signal2noise in the Sii image and what I see is basically the bitnoise of the camera because I have to overstretch it to get signal, resulting in only a very limited number of bits to distinguish whites from blacks.

The improvement I guess would be to increase my exposure times for Sii and Oiii.

 

Am I on the right track here or could this be due to another rootcause?

 

 

 

Thanks in advance, any help is much appreciated!

Attached Thumbnails

• 

Edited by Beast1987, 14 August 2024 - 02:05 AM.

[acrh2]

You need to save your stack fits files in unsigned 32-bit floating point format. Stacking multiple 16-bit files produces more data than 16 bits can hold. 

[kathyastro]

Not enough bit depth in the signal.  Even if your SNR was infinite, this would still be a consequence of a weak signal.  Definitely use 32-bit floating point, as mentioned above.  But if the problem then persists,the only remedy is longer exposure times and/or more gain.  Since this target is not exceptionally weak, I am guessing that paying attention to the bit depth of your files will fix the problem.

Edited by kathyastro, 14 August 2024 - 07:55 AM.

[unimatrix0]

 

After only basic stretching in photoshop i find that the Ha looks perfectly fine and sharp.

That's the issue right there. I think you are converting your 32bit raw files (fits) into something like a .tiff I guess? Or even worse, a .jpeg? 

 

 

the Sii however show these kind of "jumps" in between area's

Those are not jumps, but you essentially reduced the color bit depth from 32bit image to something like 4bit which is only 16 colors. 

 

So basically it has nothing to do with your images as raw, everything to do with your processing.   You probably have all the good files there, but after stacking, you converted our final stacked image into something lesser, essentially throwing away about 95% of what your camera captured. 

 

Photoshop is not a good tool especially not even able to process 32bit raw images, everything has to be 16 bit or less. 

 

I recommend finding and learning a processing software like Siril (free) or Astrophotography tool or Pixinsight. 

Edited by unimatrix0, 14 August 2024 - 09:13 AM.

[idclimber]

If you want to see what can be done with this data, consider uploading the raw integrations to a shared drive and post links in this thread. Then consider investing in PixInsight.

[Beast1987]

My thinking before was that a < 16 bit sensor could not achieve better bit depth. So I was indeed using 16 bit TIFF files.

I realise now that when stacking the "averaged" values can reach better bit depth.

 

 

So I just tried saving my stacked file from DSS to 32bit integer FITS file and 32 bit rational FITS file.

Then I loaded these files into pixinsight and compare the auto stretch to the 16 bit TIFF file.

 

Both the 16 bit TIFF and 32 bit files look quite a lot better than the photoshop stretch that I did previously, so maybe I messed up something in photoshop.

So then I tried saving the DSS stack to 32 bit TIFF file, loaded that into photoshop and stretched with a couple of levels iterations. Also this looks a lot better than the original example. Still quite noisy, but with much better bit depth.

 

 

I will look further into this and also play around with longer exposure times and/or higher gain (currently using gain=100 on the ASI 533MM) to fine tune my imaging settings.

 

Thanks for the feedback

[Oort Cloud]

Data is probably fine.

Ditch DSS, learn Siril.

Problems tend to vanish when you do that.