Page 1 of 1

Image Evaluation Tips

Posted: Sun Jul 17, 2016 9:01 pm
by Guy
Hi Ivo.

I was looking at that really helpful Astro Imaging Channel Startools video you did and at the end (at around 2h20m) someone asked about a methodical way of inspecting an image to see how good it is.

Also, in the past you have said that the workflow depends on the data - and that many problems stem from a sparse or budget setup.

I have been looking for a methodical approach to assessing images to answer the following questions:
- What are the strengths and weaknesses of the image?
- How should I change my processing workflow based on this assessment?

Have these issues been covered by a subsequent video or elsewhere? If not, do you have any advice you can give?

Thanks,

Guy

Re: Image Evaluation Tips

Posted: Tue Jul 26, 2016 11:26 pm
by admin
Apologies for the late reply - I've been traveling a whole lot the past few weeks...
Guy wrote: I have been looking for a methodical approach to assessing images to answer the following questions:
- What are the strengths and weaknesses of the image?
- How should I change my processing workflow based on this assessment?

Have these issues been covered by a subsequent video or elsewhere? If not, do you have any advice you can give?
That's a very good and extremely important question.

The question is not so much about the strengths and weaknesses of the image (they're subjective), rather it's about the strengths and weaknesses of the data you produce.
Something i wrote elsewhere;
You can get great images with even the simplest gear by just getting to the bottom of two things; 1. why does my image/data look like it does, 2. what should my finished image ideally look like.

Addressing #1 will help you get to know your gear and the unique challenges that your gear and circumstances pose. Once you completely understand how your gear and circumstances make #2 look like #1, you will fundamentally understand each and every challenge that you need to tackle, i.e. undoing or mitigating the individual things that make #2 look like #1. These things will be unique for everyone. Understand them though and all of a sudden image processing will become a 'zen' like experience - I promise! The goal here is to take the guesswork out of it all. All of a sudden you're free to get directly hands-on with your artistic interpretation and vision for the image.

So, try to understand all the possible sources that generate photons that make it to the wells of your CCD. Try to understand how these sources are distorted themselves (ex. atmospheric turbulence). Try to understand how the photons arrive at the aperture of your scope, how they are diffracted, how they are focused, what sort of aberrations your optical system exhibits, how you've tried to remedy these aberrations, how you're filtering any photons. Then try to understand how the photons arrive at the imaging plane, how your CCD converts them into electrons that are stored in wells, whether your CDD's resolution will be over or under sampling the scene, how your camera's circuitry then reads the amount of electrons in each well, how this reading is converted into a digital value, how this digital value is stored as a pixel in a 2D array of pixels, how the 2D array is demosaiced in case of a color filter, how multiple 2D arrays are aligned and stacked, how the stacking reduces noise, how the selected statistical rejection method works, etc.

At each of these steps (and many more), imbalances, noise, inefficiencies, etc. can arise. Identify and learn to spot them or their hallmarks, and find a software or hardware solution that eliminates or ameliorates them.

Spend some time learning about #2 as well. What is it that I'm really imaging here? What's its story? What are its interesting features? What is it made of? How do other people portray it?

Is there a central star to that planetary nebula that you should bring out? Are there any interesting gas knots or distributions in that nebula? Is my image showing emission line colors correctly in that nebula? Is my image showing star temperatures correctly? What's the star age (and correlated temperature) distribution in that galaxy's core and outer rim? Any faint features I should know about that I might have captured? Any interesting interactions between these two galaxies I'm imaging? What's that object's dynamic range? Should I even bother imaging that nebula in N2 narrow band? etc.

Know the answer to #1 and #2 and image processing becomes almost trivial and replicable - a means to an end, a means to get from A to B, not something you get lost in.
Is this useful at all?

Re: Image Evaluation Tips

Posted: Wed Jul 27, 2016 11:55 am
by Guy
Hi Ivo,

Thanks for the response. It certainly sounds like what I should always have in the back of our mind when processing the data. Such an understanding as you describe would also help in guiding how best to improve my technique and (over time) equipment.

To some extent I have an (albeit imperfect) understanding of the possible limitations of the envronment, equipment and my technique in using it.
What I'm trying to do is to learn the extent to which these potential limitations have impacted on the data I have collected and how best to limit their effects on the final image.

You say:
'At each of these steps (and many more), imbalances, noise, inefficiencies, etc. can arise. Identify and learn to spot them or their hallmarks, and find a software or hardware solution that eliminates or ameliorates them'

That is what I want advice on - I'm trying to find a methodical way of looking at the data collected and identify imperfections 'or their hallmarks'.
I'm using typical low-end gear so some imperfections I will have to learn to live with and to find ways to overcome as best I can.

I've looked around but not been able to find a good source of this type of advice - so I've managed to cobble some ideas together from various sources.
So perhaps I could tell you my general current approach and perhaps you could say how it could be improved:

Current Analysis
(This is aimed at a relative novice like me - but even so I may have gone into too much detail in some places)

After an imaging session - as a minimum start with a set of Lights, Darks & Bias frames.
  • - Look at the lights and discard any that are low grade or have individual artefacts that don't appear in the majority of images (e.g. Satellite trails etc, focus issues, cloud, mist - etc -etc)
    - Stack the sub-frames
Study the stacked image:

First, look for the problems:

- Check if image has been white balanced [Ivo - Is there an easy way to do this?]
- look at the overall image - are there areas where there are colour casts or vignetting? if so plan to use the Wipe tool
- Use AutoDev module and look at initial image - this will show up any colour casts and vignetting well.
- Sampling - is the resolution better than the seeing? (over-sampled) [How do you identify this from the image?]- use Bin module to reduce resolution and increase SNR
- Zoom in and look at the background - Look for shadows and marks that could indicate dust or scratches - make a note to try the Heal tool
- Look at noise:
  • Is there a lot of noise?
  • zoom in - does it have a colour bias (light pollution?) or random (equipment generated?)?
  • Are there signs hot pixels have not been eliminated before stacking?
  • Is the noise level such that a binned image would look better? [Ivo, Is this sensible - any tips on how to decide?]
    [The whole area of identifying types of noise and how best to correct it in post-processing is a bit of a 'black-art' to me. Any tips?]

- Look for banding - vertical or horizontal - Auto Dev module may show this - plan to try Band tool to fix
- Look at the stars and other features:
  • Signs of Chromatic aberration - try using Lens or Filter module later
  • Signs of Purple fringing? - try using Lens or Filter module later
- Look at the stars in the centre of the image - are they round or elongated?
  • if elongated then look back at the sub-frames and see if they are all the same or if some are worse than others - re-stack if necessary.
    - if all the same then general problem (guiding problems?) see if you can fix for next time. For now plan to try the Repair tool.
  • if round but blurred or doughnuts - see if all subs are the same - discard sub(s) and restack if necessary.
    - if all the same think about what might cause this (e.g. poor focus) and try and put it right for next time. For now plan to use the Decon module.
- Look at the stars at the edge - is there coma (stars elongated on axis towards the centre) - as opposed to star trails?
  • Make a note to use the Lens tool to reduce this before doing any cropping of the image.)
- Look at the Luminance histogram - is there much saturation? Could it have done with longer exposure/ more subs? [What other things to look for?]
- Look at the RGB histogram - is there saturation/clipping of colours? Is there much green? [What other things to look for?] This will also show up in 'Max RGB' tool in the Color module.

Now look at the qualities of the image:
The intention here is to plan about how to process the data:
[This area is not clear to me - any advice you can give would be appreciated]
- Are there any reasons to use something other than a 'standard' workflow? [If so - what are they?]
- Dynamic Range - is it very high dynamic range - are there saturated areas? - use HDR module - or merge 2 images with different exposure times (need to capture more data)
- Colours - what colours are there?
  • are they real - how to ensure they are preserved during processing? or
  • are they artificial - how to remove?
- Details - are there specific features you want to bring out?
- Image framing - does the image need to be cropped - for framing or to get rid of edge effects?

This is as far as I have got - a lot of detail in some areas - little in others - and perhaps based on misunderstandings.
I have put any specific questions I have in [] but please add to or correct it as you see fit.

Thanks and regards,
Guy

Re: Image Evaluation Tips

Posted: Fri Jul 29, 2016 9:19 pm
by admin
This is actually a very good, comprehensive list!
Guy wrote:ther from various sources.
So perhaps I could tell you my general current approach and perhaps you could say how it could be improved:

Current Analysis
(This is aimed at a relative novice like me - but even so I may have gone into too much detail in some places)

After an imaging session - as a minimum start with a set of Lights, Darks & Bias frames.
  • - Look at the lights and discard any that are low grade or have individual artefacts that don't appear in the majority of images (e.g. Satellite trails etc, focus issues, cloud, mist - etc -etc)

    - Stack the sub-frames

Satellite trails don't have to be a problem, depending on the outlier rejection algorithm you choose during stacking (for example median stacking). This decisions too depends on your circumstances; have you got a lot of time and thus a lot of frames? Can you afford to be picky?
Study the stacked image:

First, look for the problems:

- Check if image has been white balanced [Ivo - Is there an easy way to do this?]
This requires knowing what your software and camera does internally. The latter only white balances if making use of an on-board processing engine (.e.g JPEG), whereas RAW will not be white balanced.
DSS insists on white balancing, whereas PixInsight does not.
Though not scientific or always accurate, it's often easy to see whether data has been white balanced or not with a lot of cameras; a green/teal bias is often visible. White balanced images often show a red or yellow bias.
- look at the overall image - are there areas where there are colour casts or vignetting? if so plan to use the Wipe tool
- Use AutoDev module and look at initial image - this will show up any colour casts and vignetting well.
- Sampling - is the resolution better than the seeing? (over-sampled) [How do you identify this from the image?]- use Bin module to reduce resolution and increase SNR
Simple - if the image, zoomed in looks "blurry" then more pixels than necessary are used to describe the detail in your image. Your image is then oversampled.
- Zoom in and look at the background - Look for shadows and marks that could indicate dust or scratches - make a note to try the Heal tool
- Look at noise:
  • Is there a lot of noise?
  • zoom in - does it have a colour bias (light pollution?) or random (equipment generated?)?
  • Are there signs hot pixels have not been eliminated before stacking?
  • Is the noise level such that a binned image would look better? [Ivo, Is this sensible - any tips on how to decide?]
In AP, the noise level is always such that a binned image would look better. :) There is always more detail just over the horizon, waiting to be brought out, if only your data was that little bit better.
At the end of the day, it's up to you to decide of course. If you want a deep, high-fidelity post stamp, that's your artistic prerogative. The cool thing is that binning gives you that freedom to make that trade-off.
[The whole area of identifying types of noise and how best to correct it in post-processing is a bit of a 'black-art' to me. Any tips?][/list]
I could fill pages on this topic, but in most cases by far the biggest noise type is shot noise (aka Poisson noise).
"Correcting" this type of noise is not really possible as such - this noise is fundamentally 'uncertainty' in your signal. We can, however, (in all sorts of clever ways) pretend that uncertainty is not there by making educated guesses about what the signal would look like if we modeled that uncertainty and then removed it. You are spot on when you say you feel it's a black art - it very much is. There are no right answers. No one can tell you what is "too much" noise reduction or "too little".
The only unique thing that StarTools does versus any other software (that I know of) is that it makes sure to keep an exact handle on that modeled uncertainty at all times, even when your data is being processed, stretched and modified. Therefore noise reduction in ST is using the right model at all times, applicable to the image as you currently see it. This in turn makes for extremely targeted noise reduction across the image. E.g. noise grain (uncertainty) in the darker areas that are, for example, stretched more than the highlights are noise reduced more precisely because those areas contain more uncertainty after they have been stretched.
- Look for banding - vertical or horizontal - Auto Dev module may show this - plan to try Band tool to fix
- Look at the stars and other features:
  • Signs of Chromatic aberration - try using Lens or Filter module later
  • Signs of Purple fringing? - try using Lens or Filter module later
- Look at the stars in the centre of the image - are they round or elongated?
  • if elongated then look back at the sub-frames and see if they are all the same or if some are worse than others - re-stack if necessary.
    - if all the same then general problem (guiding problems?) see if you can fix for next time. For now plan to try the Repair tool.
  • if round but blurred or doughnuts - see if all subs are the same - discard sub(s) and restack if necessary.
    - if all the same think about what might cause this (e.g. poor focus) and try and put it right for next time. For now plan to use the Decon module.
- Look at the stars at the edge - is there coma (stars elongated on axis towards the centre) - as opposed to star trails?
  • Make a note to use the Lens tool to reduce this before doing any cropping of the image.)
- Look at the Luminance histogram - is there much saturation? Could it have done with longer exposure/ more subs?
[What other things to look for?]
If applicable, I would add determining the optimum ISO value for your camera to the above. ISO is a bit of an artificial thing in the digital age, since digital sensors only really have one sensitivity factor. Analog and digital circuitry artificially modifies the sensor's behavior (for example emulating a higher sensitivity by" counting" 2 photons for every real photon that is recorded), but the sensor's inherent sensitivity cannot be changed. It's a bit of a long story, but in a (somewhat simplified nutshell) the ISO where the sensor is not artificially "throttled", nor "boosted" is the ISO you should be recording at.
- Look at the RGB histogram - is there saturation/clipping of colours? Is there much green? [What other things to look for?] This will also show up in 'Max RGB' tool in the Color module.
If you haven't come across this yet, the Color module documentation has some pointers, specifically;
White balancing by known features and processes

StarTools' Color Constancy feature makes it much easier to see colours and spot processes, interactions, emissions and chemical composition in objects. In fact, the Color Constancy feature makes colouring comparable between different exposure lengths and different gear. This allows for the user to start spotting colours repeating in different features of comparable objects. Such features are, for example, the yellow cores of galaxies (due to the relative over representation of older stars as a result of gas depletion), the bluer outer rims of galaxies (due to the relative over representation of bright blue young stars as a result of the abundance of gas) and the pink/purplish HII area 'blobs' in their discs. Red/brown (white light filtered by dust) dust lanes complement a typical galaxy's rendering.

Similarly, HII areas in our own galaxy (e.g. most nebulae), while in StarTools Color Constancy Style mode, display the exact same colour signature found in the galaxies; a pink/purple as a result of predominantly deep red Hydrogen-alpha emissions mixed with much weaker blue/green
emissions of Hydrogen-beta and Oxygen-III emissions and (more dominantly) reflected blue star light from bright young blue giants who are often born in these areas, and shape the gas around them.
Image
M101 exhibiting a nice yellow core, bluer outer regions, red/brown dust lanes and purple HII knots, while the foreground stars show a good distribution of color temperatures from red to orange, yellow, white to blue.
Dusty areas where the bright blue giants have 'boiled away' the Hydrogen through radiation pressure (for example the Pleiades) reflect the blue star light of any surviving stars, becoming distinctly blue reflection nebulae. Sometimes gradients can be spotted where (gas-rich) purple gives away to (gas-poor) blue (for example the Rosette core) as this process is caught in the act.

Diffraction spikes, while artefacts, also can be of great help when calibrating colours; the "rainbow" patterns (though skewed by the dominant colour of the star whose light is being diffracted) should show a nice continuum of colouring.


Finally, star temperatures, in a wide enough field, should be evenly distributed; the amount of red, orange, yellow, white and blue stars should be roughly equal. If any of these colors are missing or are over-represented we know the colour balance is off.
Now look at the qualities of the image:
The intention here is to plan about how to process the data:
[This area is not clear to me - any advice you can give would be appreciated]
- Are there any reasons to use something other than a 'standard' workflow? [If so - what are they?]
A "standard" workflow is very personal. I you need to deviate from your standard workflow, it's a sign that something is out of the ordinary (not necessarily something bad of course).
Things "out of the ordinary" may be;
  • shooting from a different location (light pollution levels)
  • shooting at different times (moon out)
  • atmospheric conditions
  • equipment change or retuning (for example mount, flexure, insertion of filter in optical train, etc.)
  • object characteristics (high dynamic range, low dynamic range, faintness, etc.)
  • non-celestial detail (mountain range, trees, dust specks/donuts)

...or whatever makes an imaging session/circumstances markedly different to the ones preceding it.
- Dynamic Range - is it very high dynamic range - are there saturated areas? - use HDR module - or merge 2 images with different exposure times (need to capture more data)
- Colours - what colours are there?
  • are they real - how to ensure they are preserved during processing? or
  • are they artificial - how to remove?
- Details - are there specific features you want to bring out?
- Image framing - does the image need to be cropped - for framing or to get rid of edge effects?

This is as far as I have got - a lot of detail in some areas - little in others - and perhaps based on misunderstandings.
I have put any specific questions I have in [] but please add to or correct it as you see fit.

Thanks and regards,
Guy
Again, this is a fantastic list and quite possibly worth turning into an article or stickied post. :thumbsup:

Thank you!

Re: Image Evaluation Tips

Posted: Sun Jul 31, 2016 11:50 am
by Guy
Hi Ivo,
Thanks for the really comprehensive response.
I'm happy to do the compiling/editing of this list if you, and anyone else, will chip in with advice.
I'm not sure how best to do this - but I'll start with this revised version below with a view to refining this entry as I get more feedback.
For a downloadable PDF version see this external link to PDF.

Image Analysis
This is aimed at a relative novice like me - it is meant as a starting point and guide - and not a set of strict rules - add your own knowledge of your equipment and the conditions to guide in assessing the image and finding the best remedy for problems.

Note: If there isn't a link to an explanation of a note then the default place to get an understanding of how to use the modules is either the Module features and documentation web pages, or the individual module notes that you might find useful which are accessible through StarTools Main Window Use, or the Unofficial User Manual.

Checking sub-frames and stacking options
After an imaging session - as a minimum - start with a set of Lights, Darks, Flats & Bias (or Dark Flat) frames.
- Check the exposure length of the sub-frames:
  • Only stack subs with the same exposure time and ISO setting - if necessary combine them later in Startools using the Layer module- See Creating a multi exposure length HDR composite.
  • If combining subs using median or sigma modes (anything but additive) the sub exposure must be long enough to catch some signal photons on most of the light frames so that the signal is not considered as noise by the stacking process.
  • Light pollution can tempt you to reduce the exposure time but often the signal is lost. Stacking and post-processing techniques can help to reduce the effect of light pollution, and compensate for saturated areas, but there is no way to make up for lack of signal.
  • If there are some over-exposed stars they can be improved by using the Shrink (Magic) module.
- Look at the luminance histogram - See External Links SNR and Histograms and Interpreting Histograms.
  • Is there a high count on left of the histogram and nothing on the right? If so, it may benefit from a longer exposure time.
  • Is there much saturation (a high count on the far right of histogram) showing the exposure time was perhaps too long?
  • If both left and right have high counts then the object has a high dynamic range (HDR) - consider collecting sets of subs with different exposure times - stack them separately and combine later in StarTools.
- Look at the light sub-frames:
  • If there are hot or stuck pixels or fixed pattern noise:
    - If you do dithering between sub-frames (which is recommended) then these will be managed if you stack using a form of Sigma clipping with at least 10 subs. See this external article on Dithering.
    - Without dithering, dark frames will get rid of the hot or stuck pixels.
  • If there is horizontal or vertical banding:
    - If you do dithering this should be much reduced after stacking - also stacking with a set of bias frames may help with banding. If you don't currently dither between sub-frames then consider doing so in the future.
  • If there are satellite trails, cosmic rays and other small blemishes in individual frames:
    - These will be managed if you stack using Median, or a form of Sigma clipping with at least 10 subs.
    - Otherwise remove the sub-frame with the defect - or, if you only have a few sub-frames, see if you can remove it using the Heal module during post-processing.
  • If there are any that are low grade or have any other individual artefacts that don't appear in the majority of images (e.g. focus issues, cloud, mist - etc. - etc.) discard these sub-frames.
  • Discarding the subs is better than introducing errors which will be difficult to remove later - however, if you only have a few subs leaving in subs with minor problems may improve the overall signal to noise ratio (SNR) more than leaving them out.
- For OSCs and DSLRs:
  • When debayering, use Bilinear interpolation rather than VNG or AHD - This allows StarTools to control noise better.
  • Improve detail by avoiding debayering - if you have used spiral dithering during capture and have a large number of sub-frames you can stack using Bayer-drizzle (DSS) - See the topic Using debayered integrated images.
  • For noisy images - consider whether the image is sufficiently oversampled that you can afford to reduce the resolution by a factor of 4 - if so consider stacking using Super-pixels (DSS) or colour binning (Nebulosity) after reading this note by Michael Covington.
- If using DSS: - For DSLRs, check whether the optimum ISO was used:
  • The optimum ISO for a camera is one where the sensor output is not artificially amplified to emulate improved sensitivity.
  • Above this setting you are not gaining any benefit in capturing detail and the dynamic range may be reduced.
  • See the external links: DSLR Exposure, Best ISO setting for Astrophotography, DSLR Camera Data.
  • If necessary, make a note to change it for next time.
- Stack the sub-frames.

Studying the stacked image - first, look for the problems:

- Check if image has been white balanced:
  • RAW sub-frames will not be white balanced, JPEG will.
  • Stacking can introduce white balancing. Deep Sky Stacker (DSS) before v4.2.3 always white balances when using RAW subs - v4.2.3 and later have an option not to, PixInsight and Regim do not white balance by default.
  • White balanced images often show a red, brown or yellow bias. Non white-balanced images are teal or blue-green.
  • Ideally sub-frames should not be white-balanced before being processed by StarTools.
This will affect how you open the image in StarTools:
  • - If it is a (stacked) RAW image - either monochrome or white-balanced colour - select 'Linear, was not Bayered or is white balanced'.
    - If it is a (stacked) RAW colour image that is not white balanced - select 'Linear, was Bayered, is not white balanced'.
    - If it is a JPEG use 'Modified and not linear'.
- Use the AutoDev module and look at the initial image - this is specially designed to show up any colour casts, vignetting and artificial patterns such as banding. See this example using M51.
  • Are the corners darker? - this is a sign of vignetting - using flats will help.
  • Can you see darker blotches? - These may be due to dust specks on the sensor - best way to remove these is to use flats. In the absence of flats try this method for removing dust bunnies using the Heal module.
  • Is there a Zipper or Checkerboard pattern - a possible indication of a debayering issue - see the topic Checkerboard Pattern in image.
  • Zoom in to see the noise: -
    - Is there a lot of noise? - For images with low signal to noise - increase the "Ignore Fine Detail" setting to stop AutoDev responding to the noise. See also later references to noise.
    - Is the background quite uniform or are there coloured or darker single-pixel spots? If there are dark spots then consider increasing the Dark Anomaly Filter when using the Wipe module.
  • Look for stacking artefacts - be sure to Crop the image to remove these before using the Wipe module.
  • Look for any banding - vertical or horizontal:
    - Ideally, if you can, restack using a set of bias frames.
    - Otherwise plan to try the Band module to fix this: - use as first step, turn tracking off - use Band module - turn tracking back on and continue - see this topic discussing the Band module.
    - In the future stacking with bias frames, or dithering during capture, may help.
- Look at the background for Skyglow
  • Does the background have a colour bias (skyglow or light pollution):
    - Red or yellow/brown cast - skyglow that has been white balanced.
    - Teal, blue or green cast - skyglow that has not been white balanced.
    - Bright blue-green cast - skyglow filtered using a light pollution filter.
    - Missing yellow (e.g. no yellow stars) - indicates use of light pollution filter.
  • Try the Wipe Module - if this doesn't work it may be that you need more subs to increase the total integration time - or your subs aren't long enough - to get a reasonable SNR. See CN discussion Deep Sky Imaging, Aperture and Sky Fog Limits.
- Look at the stars in the centre of the image:
  • Are the stars oval and all oriented in the same direction?
    - Look back at the sub-frames and see if some are worse than others - if so discard sub(s) and re-stack.
    - If all the same this may indicate guiding problems or the imaging plane is not perpendicular to the telescope axis. For now plan to try the Repair module.
  • If round but blurred or doughnuts - see if some are worse than others - if so discard sub(s) and restack.
    - If all the same think about what might cause this (e.g. poor focus) and try and put it right for next time. For now plan to use the Decon module.
  • If there are some stars (or other features) that are over-exposed then they can be improved by:
    - shrinking stars using the Shrink (Magic) module or ...
    - using HDR techniques to combine a stack of long exposures with a stack of short exposures - See Creating a multi exposure length HDR composite.
- Look at the stars at the edge of the image:
  • Are the stars elongated on axis towards the centre of the image (Coma)?
    - Make a note to use the Lens module to reduce this before doing any cropping of the image. For the future a Coma Corrector will reduce this effect.
  • Are the stars at the edge more blurred than at the centre - or vice versa?
    - This is most likely caused by field curvature. For the future using a field flattener may help this.
- Sampling - is the resolution better than the seeing? (over-sampled) - You can improve the signal to noise ratio by reducing the resolution with the Bin module.
  • To check - look at the smallest stars - if they occupy more than a 3x3 block of pixels then the image is over-sampled.
  • When binning, reduce the resolution so this blurryness is gone. See Bin module usage.
- Zoom in and look at the background - Look for shadows and marks that could indicate dust, scratches, cosmic rays or satellite trails - make a note to try the Heal module.
  • Remember to mask out dust specks (as well as the subject) when using the Wipe module.
- Look at noise:
  • Is there a lot of noise in the background? - try using the 'Isolate' preset in the Life module later using a full mask as described in the topic First Astro Photograph.
  • Zoom in - does the noise have a colour bias (skyglow or light pollution?) or is it random colours (equipment generated?)?
  • Are there signs hot pixels have not been eliminated by stacking? Go back and use Sigma or Kappa Sigma stacking to eliminate stuck pixels.
  • Is the noise level such that additional binning of the image would look better - in spite of the loss of resolution? See Bin Module Use.
  • If noisy data then the deconvolution (Decon) module needs to be used with caution - it works best when the image has little noise.
  • Finally, the De-noise module (Wavelet De-Noise - when you switch Tracking off) will help eliminate noise - adjust the Grain Size until the noise grain can't be seen - don't worry about the signal - StarTools has been tracking it and will protect it.
  • If noise is predominantly colour blotches - set 'Brightness Detail Loss' to 0% in the Wavelet De-Noise module.
- Look at the stars and other features:
  • Signs of Chromatic aberration - try using Lens or Filter module later - See the Fringe and halo killer - or
    -Make a synthetic Luminance channel from the green channel to reduce bloat from CA in the blue channel as described in the topic Sh2-82 and reducing stars.
  • Signs of Purple fringing? - try using Lens or Filter module later - or consider using one of the approaches for processing the blue channel separately as discussed in the topic fringe killer filter add blue back to central star.
    - This may throw automatic colour calibration off - try using 'Max RGB' mode when colour correcting.
- Look at the RGB histogram:
  • Is there saturation/clipping of colours (high count on far right of histogram)?
  • Is there much green? This will also show up in 'Max RGB' tool in the Color module - Plan to fix this using Color Module - See Color module - colour balance.
Now look at the qualities of the image:
Are there any reasons to use something other than your 'standard' workflow?
- Subject - modify process depending on the subject.
  • Imaging Planets - perhaps from a video source or jpeg - try using the process described Planetary Images and LRGB stacks.
  • Imaging Comets - remove the blurry stars in the background - either when stacking by using sigma clipping, or using the Heal module as described in the topic Comet Lovejoy/2014 close up or as described in the Background Notes in Heal Module use.
  • Lunar imaging - when using the Decon module set 'Image Type' to 'Lunar/Planetary' - see the discussion in the topic Colorful Aaristarchus.
- Object characteristics (high dynamic range, low dynamic range, faintness, etc.) :
  • Is it very high dynamic range - are there saturated areas? - merge 2 images with different exposure times using Layer Module (need to capture more data) - See Creating a multi exposure length HDR composite.
  • Busy star field in background - try using the Isolate preset in the Life Module to de-emphasise it.
- Image framing - does the image need to be cropped - for framing or to get rid of edge effects?
- Are there special non-celestial elements in the image? - Trees, mountains, dust specks - Remember to mask out these object(s) when using the Wipe module.
- Atmospheric conditions - e.g. seeing is not good - Using Decon module will help here - see Decon Module use.
- Details - are there specific features you want to bring out?
  • Some nebulosity for example - See Using Heal to process stars and background independently
  • Faint Detail - Try creating a synthetic luminance channel using this approach or this alternative approach.
  • Small to medium structures - use the HDR module to locally adjust the stretch to bring out small to medium details.
  • Medium to large structures - use the Contrast module to locally adjust the stretch to bring out medium to large details.
  • Large Structures - Try using the Life module to emphasise large structures.
- Colours - what colours are there?
  • Are they real?
    + Full range of star colours - Older (red, orange, yellow) and younger (white and blue)
    + Hot ionised gas clouds - HII regions Red (Ha) or Purple (Ha & Hb), SII and NII Red, OIII cyan
    + Reflection Nebulae - Dust and cool (non-ionised) gas - can scatter light from nearby source (e.g. star) - blue scattered better than red
    - How to ensure they are preserved during processing? See Colour balancing techniques or
  • Are they wrong - like green in the wrong place? - For green use Cap Green in the Color module. For other colours try using the Filter module or ...
  • Are you using the Hubble Palette or similar? - you may want to adjust the channel balance in the Color module using the Bias Sliders.
- Combining RGB and Ha data - consider using the approach described in the topic H,R,G,B data -> synthetic luminance.
- Shooting at different times (moon out) - try using the techniques described above to handle Skyglow.
- Shooting from a different location - do you have to work to control the effect of light pollution? Are you using a light pollution filter? - See Colour Balancing using a light pollution filter which is more fully described in the topic Color balancing of data....
- Equipment change or retuning (for example mount, flexure, insertion of filter in optical train, etc.).

Re: Image Evaluation Tips

Posted: Sun Sep 04, 2016 8:18 am
by Guy
I have updated the entry above with clarifications and added links to discussions and articles.

To make this as complete as possible can anyone help me by suggesting any other:
- useful image analysis techniques?
- useful image processing techniques to resolve specific problems?
(please include links to where they are described)

Also if there are any errors, or if anything is unclear, please let me know.
Thanks,

Guy

Re: Image Evaluation Tips

Posted: Sun Sep 04, 2016 11:47 pm
by admin
This has shaped up to be a great post. Thanks again!