Jump to content

Stargazers Lounge Uses Cookies

Like most websites, SGL uses cookies in order to deliver a secure, personalised service, to provide social media functions and to analyse our traffic. Continued use of SGL indicates your acceptance of our cookie policy.

sgl_imaging_challenge_banner_30_second_exp_2_winners.thumb.jpg.b5430b40547c40d344fd4493776ab99f.jpg

jager945

Members
  • Content Count

    31
  • Joined

  • Last visited

Community Reputation

50 Excellent

2 Followers

About jager945

  • Rank
    Nebula

Contact Methods

  • Website URL
    https://www.startools.org
  1. Quickie in 1.6 beta; --- Type of Data: Linear, was not Bayered, or was Bayered + white balanced Note that you can now, as of DSS version 4.2.3, save your images without white balancing in DSS. This indeed allows for reweighing of the luminance portion due to more precise green channel. However, since this dataset was colour balanced and matrix corrected, this is currently not possible. --- Auto Develop To see what we got. We can see a severe light pollution bias, noise and oversampling. --- Crop Parameter [X1] set to [1669 pixels] Parameter [Y1] set to [608 pixels] Parameter [X2] set to [3166 pixels (-2858)] Parameter [Y2] set to [2853 pixels (-1171)] Image size is 1497 x 2245 --- Rotate Parameter [Angle] set to [270.00] --- Bin To convert oversampling into noise reduction. Parameter [Scale] set to [(scale/noise reduction 35.38%)/(798.89%)/(+3.00 bits)] Image size is 794 x 529 --- Wipe To get rid of light pollution bias. Parameter [Dark Anomaly Filter] set to [4 pixels] to catch darker-than-real-background pixels (recommended in cases of severe noise). --- Auto Develop Final global stretch. Parameter [Ignore Fine Detail <] set to [3.9 pixels] to make AutoDev "blind" to the noise grain and focus on bigger structures/details only. --- Deconvolution Usually worth a try. Let Decon make a "conservative" automatic mask. Some small improvement. Parameter [Radius] set to [1.5 pixels] Parameter [Iterations] set to [6] Parameter [Regularization] set to [0.80 (noisier, extra detail)] --- Color Your stars exhibit chromatic aberration (the blue halos) and DSS' colour balancing will have introduced some further anomalous colouring in the highlights. --- Color Parameter [Bias Slider Mode] set to [Sliders Reduce Color Bias] Parameter [Style] set to [Scientific (Color Constancy)] Parameter [LRGB Method Emulation] set to [Straight CIELab Luminance Retention] Parameter [Matrix] set to [Identity (OFF)] Parameter [Dark Saturation] set to [6.00] Parameter [Bright Saturation] set to [Full] Parameter [Saturation Amount] set to [200 %] Parameter [Blue Bias Reduce] set to [1.39] Parameter [Green Bias Reduce] set to [1.14] Parameter [Red Bias Reduce] set to [1.18] Parameter [Mask Fuzz] set to [1.0 pixels] Parameter [Cap Green] set to [100 %] Parameter [Highlight Repair] set to [Off] What you're looking for in terms of colour, is a good representation of all star (black body) temperatures (from red, orange and yellow, up to white and blue). Reflection nebulosity should appear blue and Ha emissions should appear red to purple (when mixed with blue reflection nebulosity). M42's core should be a teal green (O-III emissions all powered by the hot young blue O and B-class giants in the core, this colour is also perceptible with the naked eye in a big enough scope). As a useful guide for this object, there is a star just south of M43 that should be very deep red (but I was unable to achieve this at this scale with the signal at hand). Final noise reduction (I used Denoise Classic for this one as the noise is too heavy for "aesthetic" purposes). --- Filter Parameter [Filter Mode] set to [Fringe Killer] Put stars with their halos in a mask and click on the offending star halos and their colours. This should neutralise them. --- Wavelet De-Noise Parameter [Grain Dispersion] set to [6.9 pixels] Hope you like this rendition & wishing you clear skies!
  2. Thanks for the details Andy! Bortle 4 is definitely helping you here, allowing you to easily capture fainter objects (e.g. galaxies) and nebulosity. The CLS filter will skew colours a little, as it removes part of the spectrum. Yellows are usually impacted once the image is properly color balanced (usually yielding foreground starfields that have a distinct lack of yellow stars). It's not the end of the world, just something to be mindful of. If you're using a 600D, ISO 800 seems to be recommended. For a 6D, it's 1600 or 3200 (source). As far as ST goes, if you want to reduce the Saturation, in the Color module use the slider named... Saturation If you'd like to switch color rendering from scientifically useful Color Constancy to something PI/PS/GIMP users are more used to (e.g. desaturated highlights), try the "Legacy" preset. Finally, a maintenance release update for 1.5 was released a couple of days ago with some bug fixes. Updating is highly recommended. And if you feel adventurous, you can also try the 1.6 alpha, which comes with an upgraded signal quality-aware multi-scale (wavelet) sharpener. Clear skies!
  3. Hi Andy, That's an excellent start! At a glance, your images look nice and clean and well calibrated. It bodes very well! Can you tell us if you used a light pollution filter in the optical train? What software did you use to stack? Exposure times for each image and what sort of skies are you dealing with (e.g. Bortle scale)? I'm asking as it gives us an idea of what to you can reasonably expect. The stars in image of the North American nebula are suffering from the Gibbs phenomenon (aka "panda eye" effect). This is most apparent when using "dumb" sharpening filters such as Unsharp Mask. If you used StarTools deconvolution module, the choosing a "Radius" parameter that is too high will start introducing these ringing artifacts. Not using a star mask will also cause ringing artifacts around overexposing stars. Cheers!
  4. That is a rather excellent image with fantastic usage of the duoband filter! You may be interested in the first 1.6 alpha version of StarTools (just uploaded) which includes some specific Duoband functionality in the Color module. It lets you choose from different channel assignments and blends (and thus color schemes) without having to reprocess/reload the data. I love the HOO-ish rendering above though.
  5. Say no more. IMHO, that's not bad at all for Bortle 6 and a 1000D! Light pollution is such a killer for fainter objects. The mod explains how you were able to get the HII areas showing. Nice!
  6. This looks quite good to me. I would have expected a somewhat deeper and more noise free signal for 8.5h though. What instrument did you use to acquire the data? How is the light pollution at your location? I can't quite place Peter's comments about the colours and clipping however... The foreground stars show all temperatures of the black body curves equally (red->orange->yellow->white->blue are all accounted for). The yellow core is present (older stars), the bluer outer rim is correct (younger OB stars, more star formation), the all too often neglected purple/pink HII areas (as can be corroborated/seen here) are just about coming through in the outer rim (responsible for the star formation), the red/brown dust lanes are present. You have produced a scientifically valuable image in terms of coloring of temperatures and emissions. Finally, your histogram looks just fine to me and shows no clipping that I can see; I'd say nicely done!
  7. I'm the muppet here! My apologies - I should've made clear I'm the developer. I usually have this in my signature on other forums. I just updated my profile pic to avoid any confusion in the future. Clear skies!
  8. Hi Paul, You are producing some fairly good data - kudos! Now I could be wrong, but from the color signature (a lack of yellow) I get the impression you are using a light pollution filter. Is that correct? The only obvious thing I can recommend to improve your datasets, woudl be to try dithering between frames, as I can see some walking noise ('streaks'). When processing spiral galaxies, things to look out for are a yellow core (older stars, less star formation), blue outer rim (younger blue stars, more star formation) and purplish/pink HII areas dotted around the arms (you captured these too). You'd also be looking for a good even foreground distribution of star colors, ranging from red->orange->yellow (missing here)->white->blue. You are definitely well on your way! FWIW I processed your image in StarTools 1.5 as follows; --- Auto Develop Default parameters, to see what we are working with. We can see some stacking artifacts, gradients, walking noise, dust donut at the top. --- Crop Cropping away stacking artifacts and dust. Parameter [X1] set to [20 pixels] Parameter [Y1] set to [55 pixels] Parameter [X2] set to [2129 pixels (-16)] Parameter [Y2] set to [1414 pixels (-14)] Image size is 2109 x 1359 --- Wipe Getting rid of gradients. Default parameters. Parameter [Dark Anomaly Filter] set to [4 pixels] --- Auto Develop Final stretch. Clicked and dragged a Region of Interest (RoI) over part of the galaxy. Parameter [Ignore Fine Detail <] set to [2.5 pixels] Parameter [RoI X1] set to [95 pixels] Parameter [RoI Y1] set to [607 pixels] Parameter [RoI X2] set to [1963 pixels (-146)] Parameter [RoI Y2] set to [964 pixels (-395)] --- Deconvolution Auto mask. Default settings. Usually worth a try, as the module tends to "know" when and where improvements can be made. It didn't do too much in this case. --- HDR Default parameters. Shows a little more detail in the core. --- Color See notes above about what to look for. Parameter [Bright Saturation] set to [3.00] (less saturation in the highlights to hide some color channel alignment issues) Parameter [Green Bias Reduce] set to [1.09] Parameter [Red Bias Reduce] set to [2.09] Parameter [Cap Green] set to [100 %] (only when you're done color balancing) --- Wavelet De-Noise Default parameters. Parameter [Grain Dispersion] set to [7.5 pixels] Hope this helps at all!
  9. Hi Danjc, The short version of what's going on; stacking artifacts. Crop them away. They are fairly obvious in the image you posted. The long version; any StarTools tutorial or video will start by telling you to crop away stacking artefacts. The application itself will warn you about them too if they are detected (as it does with the dataset you posted). When you launch the Wipe module without cropping them away, you are effectively asking Wipe to create and subtract a light pollution model where those stacking artifacts are not being clipped (StarTools will virtually never clip your data unless you explicitly allow it to). In order to satisfy this, Wipe creates a model that locally backs off (e.g. locally subtracts no light pollution at all), causing local light pollution remnants around the edges where the stacking artefacts are located. Tutorials can be found here, including a "Quick start" tutorial if you're in a hurry. With regards to your dataset, it is fairly noisy (even when binned), which may make it hard(er) for AutoDev to lock onto the detail. A manual Develop will yield better/easier results in that case. When/if using Wipe, use the Narrowband preset (as of v 1.5) for this H-alpha data. You may not need to use Wipe if your data is well calibrated and there are no obvious gradients and/or bias signals in your dataset. Your dataset is well calibrated, with nothing too obvious in terms of bias signal. You can get something like the above with a simple workflow like this; --- Auto Develop Default values, to see what we got. We can see stacking artefacts (thin lines around all edges), noise, oversampling. --- Bin Converting oversampling into noise reduction. Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)] Image size is 1374 x 918 --- Crop Getting rid of stacking artefacts. Parameter [X1] set to [7 pixels] Parameter [Y1] set to [8 pixels] Parameter [X2] set to [1369 pixels (-5)] Parameter [Y2] set to [911 pixels (-7)] Image size is 1362 x 903 --- Wipe Optional, I'd say (see also comments/reasons above). --- Develop Final stretch. Choosing a manual Develop here, AutoDev will have trouble locking onto the finer, fainter detail to to overwhelming noise. Parameter [Digital Development] set to [96.76 %] Parameter [Dark Anomaly Filter] set to [20.0 pixels] --- Deconvolution Usually worth a try. Auto-generate star mask. Only very marginal improvement visible due to signal quality. Parameter [Radius] set to [1.3 pixels] --- Wavelet Sharpen Default parameters. Using same mask that was auto-generated during Decon. --- Wavelet De-Noise Default parameters. Parameter [Grain Size] set to [8.0 pixels] Hope this helps!
  10. AutoDev serves a dual purpose; first it is used to show any defects in your image, then it is used to perform your final global stretch after you have mitigated the issues you found earlier. In the above image, we can see gradients, dust donuts and (rather severe) walking noise. Flats are the #1 way to improve your image (and processing enjoyment! ), with dithering between frames a close 2nd (this solves the walking noise issue). They are really not optional and they don't really cost anything, except some time. Also don't forget you can (should) click and drag a Region of Interest for AutoDev to optimize for (by default the RoI is the entire image). In the case of the image above, a slice of M31, including its core would constitute a good dynamic range sample we'd be interested in. There are also cases where AutoDev simply cannot lock onto celestial detail. These cases usually involve heavy noise and/or lots of "empty" background. The latter can be solved with an RoI. The former can be solved by increasing the "Ignore fine detail <" parameter, which makes AutoDev "blind" to small noise grain. If you cannot fix or work around the calibration/acquisition issues in your dataset and cannot obtain a good result with AutoDev, you can always resort to a manual Develop. As bottletopburly already highlighted though, post-processing becomes infinitely easier and repeatable when you have "good" data. I'm putting "good" between quotes here, because it just means "properly calibrated". It doesn't have to be deep, gradient free or noise free, as long as it has been dithered and properly calibrated with - at least - flats. Hope this helps!
  11. No problem - at the end of the day it's all about what gives you the result you are most happy with! While it is of course futile to argue aesthetics, it is important to understand that noise and signal fidelity are physical properties of a dataset (as well as its intermediate states towards a final image). Workflows, tools and filters all affect noise propagation in different ways as you process your image. Noise and noise propagation, in turn, directly affect how much you can push a dataset; as you know it is far easier to dig out more detail with a cleaner dataset and signal. For that same reason, we endeavor to get longer exposure times - so we can show more detail. Us deep sky AP'ers deal with very faint signal. Noise is the bane of our existence. Noise mitigation is an extremely important part of signal processing. The speckles are just a visual manifestation, whose presence can indeed be left to personal taste. However its mathematical significance is omnipresent with far reaching consequences for the effectiveness of processing steps and their cumulative behavior. TL;DR Noise mitigation strategies employed by your software and its algorithms while you process, allow you to dig out more detail in your final image - it's not just about leaving or removing the speckles in your final image. Glad it helped someone! There is actually a 'home in' button that, once clicked a few times, will bounce between a range suitable for your image. Personally, I rarely use the Develop module (I use AutoDev with an RoI instead), except in cases where I really need to correct something or where AutoDev fails entirely. The latter can happen in the presence of noise, or when data was already stretched prior (AutoDev assumes a linear dataset). Clear skies,
  12. Thank you for uploading the dataset. It is much appreciated. I don't see anything immediately problematic, that would have prevented you from achieving the same result in StarTools only. The dataset is fairly bright, almost as if it has been pre-stretched and is no longer linear. I could be entirely wrong here though. Regardless, here is a simple workflow that was meant to emulate the coloring and result you produced and - presumably - are happy with. Simultaneously, it should hopefully demonstrate the major reason why you would want to include StarTools in your wokflow; superior noise mitigation and signal fidelity. If time vs result is a measure you are primarily concerned with then processing the image took ~10 minutes on a 6-core Xeon from 2010 with an SSD drive. Your mileage may vary obviously depending on hardware and how comfortable you are with operating the software. However, many default parameters and presets applied. StarTools 1.4.352 was used. --- Auto Develop To see what we're working with. We can see the image is quite bright, has a strong green bias + severe gradients, and is oversampled. Noise is already visible. Some stacking artifacts are visible. --- Bin To convert oversampling into noise reduction. Parameter [Scale] set to [(scale/noise reduction 50.00%)/(400.00%)/(+2.00 bits)] Image size is now 2329 x 1782 --- Crop To remove stacking artifacts and frame the galaxy better. Parameter [X1] set to [97 pixels] Parameter [Y1] set to [79 pixels] Parameter [X2] set to [2180 pixels (-149)] Parameter [Y2] set to [1711 pixels (-71)] Image size is now 2083 x 1632 --- Wipe Vignetting preset. Parameter [Dark Anomaly Filter] set to [6 pixels] in order to help Wipe ignore dark noise better. --- Auto Develop We're doing a 2-stage stretch here to achieve a similar stretch to your current image. Parameter [Ignore Fine Detail <] set to [3.3 pixels] to make AutoDev ignore noise Parameter [Outside RoI Influence] set to [5 %] Parameter [RoI X1] set to [755 pixels] Parameter [RoI Y1] set to [568 pixels] Parameter [RoI X2] set to [1336 pixels (-747)] Parameter [RoI Y2] set to [998 pixels (-634)] --- Develop Second stage, bring brightness down again. Parameter [Gamma] set to [0.58] --- Deconvolution Attempting some modest deconvolution. The earlier binning will have created some small areas that now have a high enough signal-to-noise ratio. Parameter [Radius] set to [2.6 pixels] --- HDR Reveal preset. Totally optional, but demonstrates the value of having these sorts of local optimisation tools in your arsenal. Parameter [Detail Size Range] set to [1000 pixels] Parameter [Strength] set to [1.7] --- Color Final color calibration, emulating the style of your current image. Parameter [Style] set to [Artistic, Not Detail Aware] Parameter [Dark Saturation] set to [2.00] Parameter [Bright Saturation] set to [2.30] Parameter [Saturation Amount] set to [365 %] Parameter [Blue Bias Reduce] set to [1.01] Parameter [Green Bias Reduce] set to [1.09] Parameter [Red Bias Reduce] set to [1.42] --- Develop I noticed you prefer a pedestal in your background. Parameter [Skyglow] set to [4 %] --- Wavelet De-Noise Final noise reduction - StarTools has now had the longest time to track (see Tracking feature) noise propagation and is ready to autonomously snuff noise out with per-pixel accuracy. Default parameters, except parameter [Smoothness] set to [82 %] After this short workflow, you should hopefully end up with this; (full res image here) Any questions, do let me know! I hope this short example has helped demonstrate the value of using StarTools in your current workflow, or at the very least demonstrates StarTools to be capable (and lower cost) alternative to other software. Clear skies,
  13. Thank you for the explanation. I'm still holding out hope there this may just be a simple understanding or maybe you are confusing StarTools with another application or action sets for PS? In the case you are indeed talking about StarTools, something appears to be terribly wrong. Have you ever processed an image from start to finish in StarTools in the past, or have you been using it for just one or two features? With regards to coloring, it appears you may have missed the Style and LRGB Method Emulation settings and/or the Saturation Amount slider in the Color module? If you don't like the default Color Constancy rendering style, it's a single click to change it to the style of other applications (e.g those of PI or APP) or even effortlessly emulate the LRGB compositing styles/techniques of Kredner/Kanevsky or Gendler with another click. Color calibration can only be performed on linear data, so perhaps that's where things are going wrong? Are you are aware StarTools performs other - for astrophotography - useful things like deconvolution, local contrast optimisation, local dynamic range optimisation, wavelet sharpening, color constancy and more, none of which are currently - by my knowledge - offered by either PS or APP? Are you familiar with ST's noise propagation Tracking feature? You are certainly not obliged to do so, but if you wish to share your dataset/stack with me, I'd love to prove StarTools' worth in your workflow. Indeed, especially in the area of noise mitigation, the present image may benefit greatly. Depending how challenging the gradients are, Wipe may also be able to clean up the red and green gradient remnants currently visible in the background. Do let me know!
  14. Hi, I'd love to know why you wouldn't recommend anyone use StarTools any longer? Is there a dataset you are having trouble with? (happy to do a personalised tutorial if so!) Thank you! EDIT: I also noticed on astrobin the difference in exposure time between the two images is 5h50m for the new image vs just 1h15m for the old image?
  15. @bottletopburly You're so right. It's kind of unfair for newbies; there is so much to tackle and if acquisition is poor, then post-processing is so much harder. It's a double whammy. What you did was the perfect approach; divide and conquer. i.e. decoupling learning post-processing from learning data acquisition. Perhaps not a video/tutorial, but I thought I'd post this animated GIF here as well. It shows what the years of development between 1.3.5 and 1.4 (to be released soon) have yielded in terms of signal fidelity. It's a 400% enlarged crop of a Jim Misti M8 Hydrogen-alpha dataset that has been intentionally non-linearly "mangled" to put Tracking through its paces. Specifically, it has been stretched, Contrast enhanced, then linearly deconvolved (using Tracking time travel and precognition of future signal evolution), then noise reduced (also using Tracking). Workflow, parameters and settings were kept identical between 1.3.5 and 1.4. The only difference is the algorithms in Decon and Denoise making increasingly more sophisticated use of the Tracking data/time travel over the years (with one unreleased quality bump applied here that hasn't been released yet). ("Original" is signal as visible, without any Tracking-enabled modules applied yet)
×
×
  • Create New...

Important Information

By using this site, you agree to our Terms of Use.