Loss of colour during wipe.

Questions and answers about processing in StarTools and how to accomplish certain tasks.

Loss of colour during wipe.

Postby shaun_slade » Tue Mar 24, 2020 4:04 pm

https://drive.google.com/file/d/1Xcap9w ... sp=sharing

Here is my stacked image of NGC 2903. This was stacked and exported as per "Starting with a good dataset guide". The link above is FITS32.

There is a really bad LP gradient in the corner. If I wipe aggressively a lot of data is lost and the image becomes very noisy AND loses all colour data.

I have to forgo the wipe to retain some colour data in the image, and still I need to increase it to 800% saturation to see any.

Is there something wrong with my workflow? Or is my stacked image that bad. It's nearly 7 hours of 2' subs with a crappy DSLR.

Here is what I have managed with ST so far:
https://drive.google.com/file/d/1jhbmCF ... sp=sharing

Kind regards,

Posts: 10
Joined: Wed Nov 13, 2019 6:10 pm

Re: Loss of colour during wipe.

Postby shaun_slade » Tue Mar 24, 2020 4:59 pm

I read the guide more carefully, then masked out all of the very bright stars and the galaxy. This has helped me do a harsher wipe while not affecting the galaxy.
Posts: 10
Joined: Wed Nov 13, 2019 6:10 pm

Re: Loss of colour during wipe.

Postby admin » Wed Mar 25, 2020 12:06 am

Hi Shaun,

Did you take flats? The gradients are quite bad in the corners and are (as an educated guess) unlikely celestial in origin. I can also see some dust specks. Per the "Starting with a good dataset" (do's and don'ts) guide;
  • Flats are really not optional - your dataset must be calibrated with flats to achieve a result that would be generally considered acceptable

If you are tacking flats, then it may be the case they were somehow applied incorrectly or do not (no longer?) correctly represent the uneven lighting and dust on your sensor.
For this dataset, however, because the gradients are bad in the corners, you can simply use the Vignetting preset without having to use masks. We're lucky here that dust specks are not obscuring the galaxy.

Color is not something you should be concerned with until you hit the Color module (indeed, if you you use the second option when importing your dataset, you are mostly processing the image in mono until you hit the Color module).

I could not replicate the issue you were having with the coloring/saturation. However, the default color balance was too green. You didn't mention what DSLR you are using, so I didn't know which matrix to use (left it at "Identity/Off").
You don't say which version of ST you are using, but I was able to produce an OK result with a standard and - intentionally - extremely short workflow with mostly default parameters in ST 1.6;

Type of Data: Linear, was not Bayered, or was Bayered + white balanced
--- Auto Develop
To see what we got. We can see a bias, oversampling. Slight stacking artefacts
--- Bin
To make use of oversampling.
Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)]
Image size is 2035 x 1344
--- Crop
To get rid of slight stacking artefacts.
Parameter [X1] set to [10 pixels]
Parameter [Y1] set to [15 pixels]
Parameter [X2] set to [2017 pixels (-18)]
Parameter [Y2] set to [1328 pixels (-16)]
Image size is 2007 x 1313
--- Wipe
Vignetting preset.
Parameter [Dark Anomaly Filter] set to [7 pixels] to catch smaller dust specks.
--- Auto Develop
RoI over galaxy. (note: you may or may not want to tweak Ignore Fine Detail < to make AutoDev ignore noise)
Parameter [RoI X1] set to [975 pixels]
Parameter [RoI Y1] set to [584 pixels]
Parameter [RoI X2] set to [1037 pixels (-970)]
Parameter [RoI Y2] set to [724 pixels (-589)]
--- Deconvolution
Conservative mask. Default parameters. It is usually worth tweaking parameters, but I'm keeping things super simple.
--- Color
The color balance was not correct by default. I used the MaxRGB color balancing method, using the galaxy's core as a reference that should-not-be-green-dominant.
Parameter [Saturation Amount] set to [400 %]
Parameter [Blue Bias Reduce] set to [1.29]
Parameter [Green Bias Reduce] set to [1.95]
Parameter [Cap Green] set to [100 %] to kill any left over green
--- Wavelet De-Noise
Parameter [Grain Dispersion] set to [17.6 pixels]
Parameter [Smoothness] set to [75 %]

NGC2903_2FITS.jpg (277.17 KiB) Viewed 764 times

I hope this helps!
Ivo Jager
StarTools creator and astronomy enthusiast
User avatar
Site Admin
Posts: 2212
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne

Re: Loss of colour during wipe.

Postby shaun_slade » Wed Mar 25, 2020 3:55 am


The issue was the gradient in the corner from a nearby street light.

I found that by masking the stars and galaxy, a very hard wipe helps get rid of it while preserving the colour data.

Additionally, I stacked the images again (and again with more subs) without using camera white balance, and saved as FITS32 as mentioned in the guide. This gave a better balance when using Star Tools.

I used Topaz Denoise for a final blast of denoising which works quite well and preserves all of the detail, but makes it too black I think.

Here is the final result (297 x 2"):


Posts: 10
Joined: Wed Nov 13, 2019 6:10 pm

Return to Image Processing Troubleshooting

Who is online

Users browsing this forum: Google [Bot] and 5 guests