Autodev query

Questions and answers about processing in StarTools and how to accomplish certain tasks.
User avatar
admin
Site Admin
Posts: 3382
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Autodev query

Post by admin »

Hi,

AutoDev is indeed showing that the Wipe wasn't wholly successful because the background is too irregular.
When the background is very irregular (for example due to noise as seems to be the case here), then Wipe will be 'scared off' when trying to figure out what the true background value is - if it encounters a pixel that is much darker (due to noise or due to a dead pixel on your CCD) aka a 'Dark Anomaly', it will assume that that pixel is real information and will calibrate against that pixel.
There are a number of ways to make Wipe perform better in the face of Dark Anomalies.
The first way is the 'Dark Anomaly Filter', which does exactly what it says on the tin :) ; it filters out dark pixels. The larger the value, the more aggressive it is with filtering out dark anomalies.
The second way is to create a mask that masks out any dark anomalies. This is useful if you have any dark dust donuts in your image.
The third way is to make sure noise is greatly reduced in the first place before handing over your image to Wipe.
From the screenshot I can tell the resolution of your image is way too high for the detail that is actually captured. i.e. real detail is smeared out over multiple pixels (due to seeing, your scope's resolving abilities and/or focusing). Fortunately there is a way to make use of the 'useless' resolution; by binning multiple pixels into a single pixel, you can reduce uncertainty (noise) in your signal. By binning exactly at the amount where detail and resolution meet (i.e. where 1 pixel contains exactly 1 unit of resolution), you do not lose any detail, but you do gain fidelity.
The latter will greatly help with your processing and using modules such as Wipe.

Do upload your image - I'd be more than happy to do a personalised tutorial!

EDIT: I should also say that if AutoDev doesn't yield any usable results despite your best efforts, you can always 'go manual' using the Develop module!
Ivo Jager
StarTools creator and astronomy enthusiast
ebacon
Posts: 9
Joined: Sun Apr 21, 2013 2:34 pm

Re: Autodev query

Post by ebacon »

The original data file is attached. I certainly appreciate your help and insight. If a photo editing dummy like me can learn then anyone can. :D

The other evening I did play with "Dark Anomaly" a little bit but did not have a feel for what it was trying to do. For example I was not sure if it was looking for anomalies darker than the average background or simply anomalies in the dark background. :think:

edit: hmmm. The attachment will not upload. Maybe 91MB is over a limit?
User avatar
admin
Site Admin
Posts: 3382
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Autodev query

Post by admin »

ebacon wrote:The original data file is attached. I certainly appreciate your help and insight. If a photo editing dummy like me can learn then anyone can. :D

The other evening I did play with "Dark Anomaly" a little bit but did not have a feel for what it was trying to do. For example I was not sure if it was looking for anomalies darker than the average background or simply anomalies in the dark background. :think:

edit: hmmm. The attachment will not upload. Maybe 91MB is over a limit?
Try Dropbox? (Free, works great!)
Ivo Jager
StarTools creator and astronomy enthusiast
ebacon
Posts: 9
Joined: Sun Apr 21, 2013 2:34 pm

Re: Autodev query

Post by ebacon »

admin wrote:Try Dropbox? (Free, works great!)
Done. That was my first time with DropBox too. Hopefully it went OK. I added the e-mail address that you sent the StarTools registration from.
ebacon
Posts: 9
Joined: Sun Apr 21, 2013 2:34 pm

Re: Autodev query

Post by ebacon »

admin wrote:. . . From the screenshot I can tell the resolution of your image is way too high . . .
Now I see what you mean about the Binning function and my image having too much data.

When I first used StarTools and played with the sliders I noticed that some of the units were in fractions of a pixel. My mind thought, "Fractions of a pixel? I have ten pixels of mess for each star!"

Here is a closeup of some stars from my photo. I agree that there is way too much data there.

Image
User avatar
admin
Site Admin
Posts: 3382
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Autodev query

Post by admin »

ebacon wrote:
admin wrote:Try Dropbox? (Free, works great!)
Done. That was my first time with DropBox too. Hopefully it went OK. I added the e-mail address that you sent the StarTools registration from.
Ok, I received the image no problem.
First off, the most obvious problem that I can see is that your data is not linear - it has been stretched by something.
This is the either because you shot in JPEG (big no-no!) or because you stretched in another application.
For best results, StarTools requires linear (unstretched) data. Else it cannot perform its tracking properly.
Furthermore it appears (but I could be wrong) that it has had some sort of noise reduction applied to it (possibly because it was shot in JPEG and the camera applied noise reduction). This makes it hard to recover any faint detail which is typically obliterated by 'terrestrial' noise reduction.

I used a trick to 'relinearise' the data (assuming the stretching was due to JPEG conversion); when loading I indicated the image was stretched. Then I launched the Develop module and applied a gamma correction of approx 0.45. This undoes the typical 2.2 gamma correction that JPEG applies conform the sRGB profile. I then proceeded with clicking the Track button, indicating that the data was now linear.

From there, I performed the following;

I binned the data to 50%
I cropped the image;

Parameter [X1] set to [475 pixels]
Parameter [Y1] set to [344 pixels]
Parameter [X2] set to [2135 pixels (-321)]
Parameter [Y2] set to [1320 pixels (-312)]

I then used Wipe in an attempt to get rid of the worst of the light pollution and vignetting. I used the Vignetting preset with;

Parameter [Dark Anomaly Filter] set to [13 pixels]

Wipe left some blotches, possibly caused by some sort of noise reduction or filter. AutoDev wasn't much help, so I used the manual Develop module;

Parameter [Digital Development] set to [50.73 %]
Parameter [Dark Anomaly Filter] set to [16.0 pixels]

I proceeded to use the Life module's 'Moderate' preset, which has the effect of 'bulking up' a cluster's glow a little.

Next up I performed some Color calibration;
Parameter [Cap Green] set to [To Yellow]
(the above parameter setting makes sure that no green tints remain in the image)
Parameter [Bottom End Saturation] set to [1.50]
(the above parameter setting keeps the bottom end saturation response in check, keeping the background nice and neutral)
Parameter [Top End Saturation] set to [9.90]
(the above parameter setting makes sure bright things, i.e. stars, benefit maximally from saturation increase)
Parameter [Saturation] set to [326 %]
Parameter [Green Ratio] set to [1.10]
(the above parameter, in my opinion, added a little bit of yellow that some stars were lacking - when performing color calibration all star colors should be equally well presented in your image)

Finally I switched tracking off and performed final noise reduction using the data that Tracking gathered during my processing;
Parameter [Scale 1] set to [50 %]
Parameter [Scale 2] set to [98 %]
Parameter [Scale 3] set to [95 %]
Parameter [Scale 4] set to [89 %]
Parameter [Scale 5] set to [80 %]
Parameter [Color Detail Loss] set to [50 %]
Parameter [Brightness Detail Loss] set to [37 %]

And this is what we ended up with;
M13_Single_from_RAW_ST13.jpg
M13_Single_from_RAW_ST13.jpg (148.73 KiB) Viewed 9960 times
Hope this helps!
Ivo Jager
StarTools creator and astronomy enthusiast
ebacon
Posts: 9
Joined: Sun Apr 21, 2013 2:34 pm

Re: Autodev query

Post by ebacon »

You are the master! :bow-yellow:

I wonder how the data was previously stretched? The data is from a Sony A57 RAW file that was simply converted to TIFF. My suspicion is that the stretching happens inside the camera. The reason I think that is that I did an experiment to see how much amp glow came from the camera. No matter how high I set the ISO or how long my exposures were they showed zero amp glow. There must be image processing happening in the camera and there is no way to turn it off.
gboulton
Posts: 122
Joined: Thu Dec 06, 2012 12:40 am

Re: Autodev query

Post by gboulton »

admin wrote:We're here for any (and I really mean ANY) questions you might have! :thumbsup:
1 horse sized duck, or 100 duck sized horses? :lol:
-- Gordon

Image

My Astrobin
User avatar
admin
Site Admin
Posts: 3382
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Autodev query

Post by admin »

ebacon wrote:You are the master! :bow-yellow:
That's too much praise! :D You're not far off from getting such results yourself, and totally to your individual taste!
I wonder how the data was previously stretched? The data is from a Sony A57 RAW file that was simply converted to TIFF. My suspicion is that the stretching happens inside the camera. The reason I think that is that I did an experiment to see how much amp glow came from the camera. No matter how high I set the ISO or how long my exposures were they showed zero amp glow. There must be image processing happening in the camera and there is no way to turn it off.
Aha! :evil:
Yes, that terrible Sony tool has caught out users before. It performs all sorts of unwanted operations and modifications to your image (including stretching and noise reduction).
For single frame conversion, try a tool like RawTherapee, which is the best tool that I know of for 'anything RAW'->TIFF conversion.
DeepSkyStacker uses the same code (called DCRaw) for conversion of RAW frames when stacking, so when you're ready to start stacking multiple frames you don't have to bother with RAW->TIFF conversion at all.
Ivo Jager
StarTools creator and astronomy enthusiast
User avatar
admin
Site Admin
Posts: 3382
Joined: Thu Dec 02, 2010 10:51 pm
Location: Melbourne
Contact:

Re: Autodev query

Post by admin »

gboulton wrote:
admin wrote:We're here for any (and I really mean ANY) questions you might have! :thumbsup:
1 horse sized duck, or 100 duck sized horses? :lol:
Common question. I should probably make this a sticky.
The answer is; it depends on your light pollution, but the rule of thumb is 110 duck sized horses, reduced by approx. 12 duck sized horses for each Bortle class of light pollution you're experiencing. Only go for single horse sized duck if imaging from excellent dark skies (e.g. class 1 which is approximately 1 horse sized duck). The granularity of the smaller horses allows you to fine tune.
Ivo Jager
StarTools creator and astronomy enthusiast
Post Reply