Title: SH2-101 the Tulip Nebula in Cygnus (Reprocessed)
Description:
Monmouth, OR
September 3 and 4, 2021, Reprocessed January 14, 2022
With additional data, I reprocessed the Tulip Nebula narrowband filter images using the Hubble pallet (Sulfur-II to Red, Hydrogen-a to Green and Oxygen-III to Blue channels, SHO). Several new tools and techniques were tested in PixInsight, so the resulting false color image is a work in progress. In the first processing version, I had created a HSO color image that made the tulip glow in red-yellow hues. This time, I wanted to create more of a rainbow effect and utilize starless images in the processing workflow. While the image is no longer scientifically relevant, it does show more details in the nebula than the previous version.
Key steps in the image processing workflow:
1) I used the Ha, Oiii and Sii calibrated and integrated images generated in the previous HSO image processing version. Each channel image had been cropped and then denoised with the Mure Denoise script.
2) Russ Croman’s StarXTerminator was used to create starless and stars images for each channel in the linear state. The one issue with StarXTerminator using the Version 7 AI weighting is that very faint ghosts of stars remain in the combined starless image, resulting in subtle mottling where stars are not present in the image combined with the stars image.
3) The stars narrowband images were combined into one RGB image and the Photometric Color Calibration was used to create some semblance of acceptable star colors. Short of taking RGB data, there is probably a better way to handle the stars images to obtain more realistic colors.
4) The stars RGB image was stretched and processed to get a desired number of stars with adequate brightness and morphology. The CurvesTransformation tool was used to tweak the star brightness and how many stars were visible and to make the background black for subsequent combination with the processed starless image.
5) The Ha, Oiii and Sii channels were linearized with LinearFit to the brightest channel without saturating any one channel. To determine which image was the brightest (or dimmest), the Statistics tool was used, and the Mean and Median values defined the ranking. Each channel when combined would then have similar or enhanced brightness. Starless images provide great capability and flexibility for image processing.
6) The narrowband starless images were combined with the ChannelCombine tool using the Hubble Pallet assignment (SHO) to create an RGB color image. This made Sii and Oiii contributions strong and there was no green cast to be removed.
7) The Luminance was extracted from the RGB image.
8) The Luminance image was stretched, denoised (background) and sharpened. I downloaded and tried the Generalized Hyperbolic Stretch script to stretch the image in several passes. The tool provides excellent curve shaping parameters. The downside is that there is no real time preview when moving sliders, so I had to use the histogram after the first stretch. This script allows you to shape the curve to prevent clipping stars while being more aggressive with the midtone and shadow regions. I used this script for all image stretching, including the RGB starless and stars images. Sharpening was done with the Histogram Equalization process tool with three kernel sizes and bit depths to sharpen details for different scales. Denoising was done with the TVGDenoise process tool using a low contrast mask and a support mask.
9) The starless RGB image was stretched, denoised with the MultiscaleLinearTransform tool, and then it was blurred with the Convolution tool.
10) The Luminance image was then combined with the starless RGB image using the LRGB Combine process tool.
11) I started using a new set of PixelMath expressions to create and then blur color masks that seem to produce better masks than the Color Mask script. The LRGB starless image was adjusted with several color masks (Yellow, Green, Blue, Cyan and Magenta) and the Curves and Histogram Transformation tools. The color mask PixelMath process icons can be downloaded from the lumatico YouTube channel at
(https://drive.google.com/drive/folders/138L6Z2vbTzNAutbfe5cuhL9Pl0-iltRp). Process Icon Merge was used to integrate these icons into my PixelMath_Comos workspace. First, a color mask is created using the stretched starless image. Next, the MaskBlur is applied to the new mask. Finally, the mask is applied to the starless image and the CurvesTransformation is used to adjust brightness, hue and saturation. Although the mask has been blurred, Curves have to be somewhat subtle to prevent the creation of color artifacts.
12) The starless image was saved as 16-bit TIF and then opened in Photoshop 2022 to selectively sharpen (sharpen tool), dodge and saturate (sponge tool) different regions of the image.
13) The starless and stars images were combined in PixInsight using the PixelMath combine function with the op_screen() option [ combine(stars,starless,op_screen()) ]. Note that there are far fewer stars, and they are somewhat dimmed so that they do not overwhelm the nebula. There are some minor color artifacts on the edges of the stars that may have been created with the Ha, Oiii and Sii stars images, their combination or color calibration.
Imaging details:
Celestron 9.25" Edge HD SCT
Celestron 0.7x Focal Reducer (FL = 1645mm, f/7)
Celestron off-axis guider with a ZWO ASI 174MM mini guide camera
Losmandy G11 mount with Gemini 2
ZWO ASI 2600MM Pro cooled monochrome camera (-10oC)
ZWO 36mm Hydrogen-, Oxygen-III and Sulfur-II filters
Software: Sequence Generator Pro, ASTAP plate solving, PHD2 guiding,
Losmandy Gemini ASCOM mount control and web client interface,
SharpCap Pro for polar alignment with the Polemaster camera.
Software for image processing on a Macbook Pro:
PixInsight 1.8.8-12, StarXTerminator for PixInsight, and Photoshop 2022
Hydrogen-a 10 min x 21 subframes (210 min), Gain 100, Offset 68, 1x1 binning
Oxygen-III 10 min x 16 subframes (160 min), Gain 100, Offset 68, 1x1 binning
Sulfur-II 10 min x 15 subframes (150 min), Gain 100, Offset 68, 1x1 binning
Uploaded: ...
Comments