VdB 14 and 15 are two very bright reflection nebula in Camelopardalias. They have been on my wish list of targets to capture with the newt for a few years now but haven't been able to give the targets the time they need. This fall however an opportunity came up to give it a shot and so I decided to aim for the brighter of the two pair, vdB 15 and I am so happy with how the data turned out. My favorite version of this target has to be the
two-panel mosaic by Prashant Ranganath that was the real inspiration and motivation to not letting it slip by me this year.
Caution: This image has undergone manipulation using neural network-based algorithms; users are advised to acknowledge that neural networks may infer information not strictly accurate or representative of the photographed subject, and while efforts have been made to ensure accuracy, some details may not be entirely genuine for scientific purposes. Neural Network based algorithms used: StarXterminator. DeepSNR
Equipment Overview:
* Orion 8in f4.9 / 1000mm focal length reduced to 920mm (tswynne m68 .97x reducer)
* Orion Atlas Pro AZEQ Mount
* QHY268m with Chroma LRGB 2in filters
Processing Notes:
* All individual subs were calibrated with dark masters, scaled with bias, and corrected with corresponding flat and dark flats using Image Calibration, Cosmetically Corrected using the dark masters, weighted using PSFSignalWeight in Subframe Selector, groomed using Blink to reject poor quality subs aligned and then stacked using PixInsight
Applying local normalization to the Luminance Data:
* The best subs were used to generate a local normalization reference, and that reference was then corrected using the MSGR technique outlined by Vicent Peris with separate 350mm data I collected, and then run through local normalization. I don't often use WBPP, but I do feed it my aligned subs to perform a Local Normalization reference image generation because it has a helpful interactive mode that lets you pick the best gradient-free (or reduced) subs to build the reference with. After building the reference, I applied local normalization directly to the aligned subs and stacked using image integration to build the luminance master.
Noise Reduction on Luminance:
* In addition to the normal Luminance stack mentioned above, three additional stacks were made using the luminance data by grouping the aligned subs into three independent data sets via a Round-Robin distribution sorted by the PSFSignalWeight and using the local normalization files produced by the previous step.
* These three sub-stacks were then combined into a color image (not that it is particularly colorful) so that DeepSNR could be applied. (The trick with DeepSNR is to ensure each channel has differing noise to them to minimize the introduction of artifacts.)
* DeepSNR was then applied to this three-channel luminance, and then a new synthetic luminance was extracted from the image with a weighting of 1:1:1 on each channel
* This Noise-Eliminated luminance was then blended back into the non-noise-reduced image using a mask to protect the high signal areas and to protect the star cores of the image carefully to minimize the introduction of false information.
Stretching the Luminance Data:
* First step of stretching the luminance data is to remove the Stars. I used StarXterminator to perform this task, and then set the Stars image asside.
* Next, I set the blackpoint on the image to where the minimum values are without clipping. To do this, I create a preview window on the image that excludes the stacking edges but fills most of the image, and then use Histogram Transformation's "Auto Zero Shadows" feature and then apply the value to the starless image, and re-apply the screen transformation.
* After setting the black point, I then launch GHS and then set the symmetry point to the peak of the histogram by zooming into the far left side of the image. I then bring the stretch factor up until the peak of the histogram is around 0.12 or 0.13, and then bring the intensity factor up such that the brightest part of the nebula isn't over stretched but not so much that the data looks flat.
* Background signal was further stretched by using a stretched copy of the image as a mask to protect the midtones and highlights and using Local Histogram Equalization targeting the largest scales (Radius 512px, Contrast Limit of 6 and Amount of 0.1)
RGB Processing
* The RGB stacks were combined using Channel Combination
* The gradients were modeled and corrected using the MSGR technique outlined by Vicent Peris with separate 350mm data I collected.
* The combined RGB image was Photometrically Color Calibrated
* DeepSNR was used for noise reduction: A copy of the RGB image was made and then run 100%. This copy was then blended back into the non-noise-reduced image using a mask to protect the high signal areas and to protect the star cores of the image.
* StarX was then used to separate the Stars from the nebula data for further processing and be re-added later.
Stretching the Color Data
* I use GeneralizedHyperbolicStretch (GHS) for most of my stretching needs, however before using it I like to set the black point on the image to where the minimum values are without clipping. To do this, I create a preview window on the image that excludes the stacking edges but fills most of the image, and then use Histogram Transformation's RGB/K black-point slider to find where information starts to clip by zooming in on the histogram slider to the far left and setting the black point carefully such that zero pixels are clipped. Since the image has already been color calibrated, I do not apply this technique to the R,G,B channels separately but only to the RGB/K channel.
* After setting the black point, I then launch GHS and then set the symmetry point to the peak of the histogram. Like the Histogram Transformation tool, I have to zoom in far on the left of the image to find the peak.
* I then start with a small stretch in "Colour" mode, bringing up the Stretch Factor a small amount (in this case 6.5) followed by the Local Intensity (in this case 2.2) The goal here is I want to just to start to see the colorful portion of the nebula to show up without any screen stretch applied
* After the first Colour stretch, I then perform the RGB stretch to bring out the overall contrast of the nebula. I re-apply the symmetry point at the new peak of the histogram, and then bring the stretch factor to where the information is spread across the histogram then adjust the intensity factor to where it gives me the most pleasing balance, and then tweak the symmetry point to target the background levels.
* A contrast enhancement is then made using Local Histogram Equalization targeting the largest scales (Radius 512px, Contrast Limit of 6 and Amount of 0.1)
* GHS was then used in Saturation mode to bring the out more of the color information
Combining Luminance with RGB
* Now that we have our starless luminance and color data stretched and contrast enhanced, I use the LRGB Combination tool to apply the luminance into the RGB image. Because I didn't get an aggressively high ratio of Luminance to RGB, I use a Channel Weight value of 90% for this so that the RGB data continues to contribute to the detail layer.
Additional processing
* ACDNR was used with a protective mask to target some additional background noise
* Background hints of green and blue were transitioned slightly to red using Curves Transformation for a better neutralization with a highly protective mask
* MMT was used to perform a slight amount of sharpening by increasing the bias on wavelet layers 3, 4 and 5
* Dark Structure Enhance was used to add some depth to the smaller scale shadows
Star Processing and Addition
* GHS was used to process the stars. By creating a preview window over a sampling of Stars, the Stretch Factor was captured first, and then the Local Intensity to make sure the star profiles look natural and not over stretched or under represented relative to their outer fainter glow
* Stars were brought back in using the Reverse-Midpoint addition technique described by Charles Hagen:mtf(.005,
mtf(.995,Starless)
+mtf(.005,Stars)
)
Comments