True, it's not the right time of year to image this subject. The data is old, but the processing was done today.
I have been using PixInsight for a while, but recently spent some time digging in to the tools I use most (especially the noise reduction tools), and learning about some of the recently added features. After quite a bit of reading, watching video tutorials and experimenting, I decided to reprocess some old data, consisting of two sets of raw DSLR images I collected about one year apart of B33 (37 subframes of the subject taken with a Canon 350D, totaling about 3.5 hours, along with twilight sky flats and darks taken during those two imaging sessions).
Here is a summary of the processing steps - all processing was done in PixInsight:
1. Raw images pre-processed with the
BatchPreProcessing script, using a SuperBias, dark optimization with an optimization threshold, cosmetic correction and Bayer Drizzle. Used the "Add Custom" feature to group the two sets of light, dark and flat frames so that all could be processed in one pass.
2. SubFrameSelector script to eliminate two poor subframes (based on FWHM) and to select the best reference image for ImageIntegration.
3. ImageIntegration - tried Winsorized Sigma Clipping and Linear Fit clipping. WSC gave substantially better SNR with these calibrated subframes.
4. DrizzleIntegration - subsequent steps were perfomed on the drizzled result.
5. DynamicCrop - this integrated image had to be cropped considerably because of different alignment and rotation between the two imaging sessions.
6. BackgroundNeutralization,
ColorCalibration. No
DynamicBackgroundExtraction, as there isn't a lot of background in this image, and it was pretty flat after the preceeding steps. I believe this is the first image I have processed with PixInsight that didn't need DBE.
7. Noise reduction with
MulitscaleLinearTransform, utilizing the Linear Mask feature.
8. MaskedStretch9. One pass of
ACDNR, with stdDev set to 10 for both Lightness and Chrominance and Robustness set to 5x5 Weighted Average, to reduce minor "blotchiness" in the low signal areas. I didn't save the process icon and don't recall the Amount settings, but they were less than 1 (around 0.7 IIRC).
10. LocalHistogramEqualization11. Resample, then
AssignICCProfile (sRGB) prior to saving as JPEG for browser presentation.
12. Small contrast and saturation increases using
CurvesTransormation to reverse effects of assigning the sRGB profile.
My own critique of this resulting image is that the noise reduction may be overdone a little, the sky background may be a little too dark, and the low signal nebulosity in the lower left corner appears to have been smoothed too much. But this is a much better result than I have achieved in the past. It pays to do some homework.
The full size result can be seen here:
http://dandjreed.homedns.org/images/b33_final_fullsize.jpgDon
