Help needed, Monochrome Data

[szymon]

Hi there, pleased to meet you all.

I am trying to get Startools to process a simple image of M101 that I took with an Altair Hypercam 183M, using a Celestron ED80. There's not a huge amount of data, but what there is has been stacked in DSS as follows:

Code: Select all

Stacking step 1 ->5 frames (ISO: -) - total exposure: 50 mn 0 s 
RGB Channels Background Calibration: No 
Per Channel Background Calibration: No
Method: Median
-> No Offset -> Dark: 1 frames (ISO : -) exposure: 10 mn 0 s
-> Dark Flat: 10 frames (ISO : -) exposure: 4 s
Method: 
Median
-> Flat: 10 frames (ISO: -) exposure: 4 s
Method: Median

Yes there should be more darks, but it was late and I was tired, and this was just a test. In any case, even DSS seems able to stretch this and turn it into something worth looking at.

Here is an example stretched in DSS: https://www.dropbox.com/s/vkid21wr1hv9g ... h.png?dl=1
Here is an example stretched in Gimp: https://www.dropbox.com/s/sq6jjxgfvf6ky ... h.jpg?dl=1

However I can't get anything resembling that from StarTools, and I've been trying for hours! Autodevelop seems to bring in huge amounts of noise, developing manually doesn't seem to give me the control that I do with gimp curves or levels (e.g. bring down the middle, then bring up the black point to compensate), no matter what I do the best I can do looks like a print out on an old dot-matrix printer (those of my age will understand this!).

All of the example videos I've seen are either for colour images, barring one monochrome Ha image which was already very high quality data. This is data through a CLS filter in London (Bortle 8). Is it the case that Startools just doesn't work with monochrome data? Am I doing something badly wrong?

If someone could try taking my data and see if they can at least approximate the poorly processed examples above from DSS and Gimp, and then tell me how they did it, that would be awesome . Here is the stacked output:

https://www.dropbox.com/s/0qzewas7pi019 ... s.FTS?dl=1

Many thanks,

-simon

[almcl]

I have had a quick go with your data, Simon and put the result below. To save space and reduce noise it's binned 50%.

Not sure if it's much of an improvement, see what you think?

me101_dss_noadjustments.jpg (187.39 KiB) Viewed 6228 times

[szymon]

Wow, that's a definite improvement on what I was able to achieve! (I'm not looking for perfection, just something better than ridiculously bad, which is what I was getting). The data is very imperfect, this is LPLondon with just a CLS filter, but I was expecting to at least match what DSS/Gimp can do.

Can you please paste exactly what you did in StarTools to achieve this? I must be missing something. I'll try again now

[almcl]

It was only a quick and dirty version of Richard Konrad's work flow (https://www.youtube.com/watch?v=tNidlGR ... e=youtu.be and https://www.youtube.com/watch?v=PPoJ-lM ... e=youtu.be) and I missed a few steps along the way; masking could definitely be improved and so could decon and sharpen, but here is the logfile (it's too big with the masks and the board doesn't allow text files, so you will have to work at masking the nebula before the first Life module):
----------------------------------------------------------
StarTools 1.4.352
Sat Jul 06 07:55:44 2019
-----------------------------------------------------------
File loaded [E:\Astronomy\someone elses files\me101_dss_noadjustments.FTS].
Image size is 5435 x 3643
---
Type of Data: Linear, was not Bayered, or was Bayered + white balanced
--- Auto Develop
Parameter [Ignore Fine Detail <] set to [Off]
Parameter [Outside RoI Influence] set to [15 %]
Parameter [RoI X1] set to [4152 pixels]
Parameter [RoI Y1] set to [1586 pixels]
Parameter [RoI X2] set to [4219 pixels (-1216)]
Parameter [RoI Y2] set to [1678 pixels (-1965)]
Mask used (BASE64 PNG encoded)



--- Life
Parameter [Detail Preservation] set to [Linear Brightness Mask]
Parameter [Compositing Algorithm] set to [Multiply, Gamma Correct]
Parameter [Inherit Brightness, Color] set to [Off]
Parameter [Output Glow Only] set to [No]
Parameter [Airy Disk Sampling] set to [128 x 128 pixels]
Parameter [Airy Disk Radius] set to [64 pixels]
Parameter [Glow Threshold] set to [0 %]
Parameter [Detail Preservation Radius] set to [20.0 pixels]
Parameter [Saturation] set to [100 %]
Parameter [Strength] set to [99 %]
Parameter [Mask Fuzz] set to [40.0 pixels]
Mask used (BASE64 PNG encoded)


--- HDR
Parameter [Small Detail Precision] set to [Max]
Parameter [Channels] set to [Brightness Only]
Parameter [Algorithm] set to [Optimize Hard]
Parameter [Dark/Bright Response] set to [Full]
Parameter [Detail Size Range] set to [21 pixels]
Parameter [Strength] set to [1.2]
--- Bin
Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)]
Image size is 2717 x 1821
--- Life
Parameter [Detail Preservation] set to [Linear Brightness Mask]
Parameter [Compositing Algorithm] set to [Multiply, Gamma Correct]
Parameter [Inherit Brightness, Color] set to [Off]
Parameter [Output Glow Only] set to [No]
Parameter [Airy Disk Sampling] set to [128 x 128 pixels]
Parameter [Airy Disk Radius] set to [4 pixels]
Parameter [Glow Threshold] set to [0 %]
Parameter [Detail Preservation Radius] set to [20.0 pixels]
Parameter [Saturation] set to [100 %]
Parameter [Strength] set to [80 %]
Parameter [Mask Fuzz] set to [1.0 pixels]
Mask used (BASE64 PNG encoded)


--- Wavelet Sharpen
Parameter [Structure Size] set to [Large]
--- Wavelet Sharpen
Parameter [Intelligent Enhance] set to [Yes]
Parameter [Scale 1] set to [100 %]
Parameter [Scale 2] set to [100 %]
Parameter [Scale 3] set to [100 %]
Parameter [Scale 4] set to [100 %]
Parameter [Scale 5] set to [100 %]
Parameter [Mask Fuzz] set to [8.0 pixels]
Parameter [Amount] set to [260 %]
Parameter [Small Detail Bias] set to [75 %]
Mask used (BASE64 PNG encoded)



--- Wavelet De-Noise
Parameter [Filter Type] set to [Distance Weighted Outlier Rejection]
Parameter [Grain Size] set to [3.6 pixels]
--- Wavelet De-Noise
Parameter [Scale 1] set to [90 %]
Parameter [Scale 2] set to [90 %]
Parameter [Scale 3] set to [90 %]
Parameter [Scale 4] set to [90 %]
Parameter [Scale 5] set to [0 %]
Parameter [Mask Fuzz] set to [1.0 pixels]
Parameter [Scale Correlation] set to [6]
Parameter [Color Detail Loss] set to [12 %]
Parameter [Brightness Detail Loss] set to [12 %]
Parameter [Grain Size] set to [3.6 pixels]
Parameter [Read Noise Compensation] set to [Off]
Parameter [Smoothness] set to [75 %]
File saved [E:\Astronomy\someone elses files\me101_dss_noadjustments.jpg].

[szymon]

Ok, so that's interesting. It's not dissimilar to what I was trying previously. I tried just what you did, and got the same result -- on the MacBook I'm using at the moment. But on the Windows machine that I was using yesterday, the results were nothing like this. Is it possible that exactly the same actions on Windows produce a different result to those on OSX? I can't reboot to Windows at the moment, but I'll try it out again with exactly these instructions and post the result when I get some time. Thanks for your help!

[almcl]

Sorry! Don't know anything about Apple machines, but Ivo might drop by and give his take on this. I would have expected identical results.

Anyway glad if it has helped a bit.

[admin]

Hi,

Sorry to hear you're having trouble (and thanks everyone for helping!)
Which version have you been using on macOS? The 1.4.x versions are highly recommended for macOS, as they use a 64-bit signal path and are 64-bit native.
There may be some discrepancies in the output of 64-bit and 32-bit (1.3.5 is 32-bit only on macOS) versions due to some loss of precision and some memory saving measures, but they should not be too dramatic...

I can't see anything hugely problematic with the dataset, except of course the noise and calibration issues you mentioned.
The standout issue is perhaps the incorrect application of flats (right now, they make things worse!). It appears the entire frame has shifted from when the flats were made, yielding some sort of "embossing" effect. E.g. this is what an initial AutoDev shows (after binning to 25%);

me101_dss_noadjustments_initialAutoDev.jpg (521.53 KiB) Viewed 6205 times

StarTools especially is very sensitive to such anomalies; it will do its best to preserve them, as - normally - they constitute faint detail. This is also the reason why you AutoDev and Develop will not give you the "control" you need to unceremoniously black-clip these anomalies.

The Wipe module - whose job it is to fix up your very uneven background on a per-pixel-basis (e.g. rather than just using black point silder for a global subtraction) is very sensitive to these sorts of anomalies. It's likely an easy fix to just have the flats take care of it, but for the sake of completeness, here's how you deal with this issue;

Make sure you have cropped any stacking artefacts. Then go into the Wipe module and use the Lasso tool to create gaps in the mask where Wipe should not sample the background (e.g. where the anomalies are).
Then let Wipe do its thing. You'll end up with something like this;

me101_dss_noadjustments_initialAutoDev_WipeAutoDevGamma042.jpg (321.14 KiB) Viewed 6205 times

The background is nice and even, but Wipe has been allowed to clip the worst of the anomalies. Again, normally Wipe is not allowed to clip your data like this - everything is carefully treated as bonafide celestial detail by default.
There is even a way to fix up the dark/clipping anomalies, but really, your efforts should be focused not to have those in your dataset in the first place, rather than trying to fix this in post-processing.

Hope this helps & do let us know if you continue to have trouble!

[szymon]

Ivo,

I was using exactly the same version on both Windows and OSX. My good results were on OSX, my bad results were on Windows -- but I've a feeling that it was actually user error! I think I was using the wrong files on Windows. I run both Windows and OSX on my MacBook, using Apple's "Boot Camp". I capture data under Windows, because it has better support for my equipment. I then stack in DSS, and then move back to OSX to process, usually in Gimp 2.10, just because the trackpad works better under OSX than it does in Windows! However, this time I stacked in DSS and then the clouds cleared up and I decided to take some more images (of a different target), and was processing in StarTools under Windows while the capturing was going on in the background. I've a feeling that instead of loading the stacked FITS output that I uploaded here, I was processing a different FITS file -- more than likely a single frame. So it's not surprising that the results were much worse than expected -- that's a lot less data!

In any case, this post has thoroughly convinced me to buy StarTools; good job, I will be in touch

-simon