I've had PixInsight for about 2-1/2 years and got it shortly after taking up astroimaging. I'm currently using modded Nikon DSLRs (D5300 and D5500) but just recently picked up an ASI 533MC-Pro.
Question is on my workflow. As time has gone by, I've gotten better at processing, but there's always room for improvement. Finding the time to sit down to watch all 653,242 PixInsight YouTube tutorials (each spanning a length of geological time) is admittedly a hurdle. I mean, I've picked up at least a few things along the way. Personally, I find VisibleDark's the easiest to follow and understand, but to each their own.
Basically, I'm wondering if you were taking OSC images (I have ZERO desire for mono imaging), how would you judge the following workflow?
For stacking, I just use Deep Sky Stacker and take the unmodified Autosave.TIF file straight into PixInsight. Why not use PixInsight? Because DSS is just too easy and learning everything in PXI is 10X harder than it needs to be.
Once it's in PixInsight, here's what I do. ***NOTE–I haven't posted any photos with this method yet, but I have reprocessed several images using this method. I'll post some of these in the next few days. I've just seen the improvement in the images using the following method, but am wondering if I can still improve upon it.***
LINEAR STAGE:
1. Screen Transfer Function / autostretch.
2. Dynamic crop for stacking artifacts/margins.
3. ABE/DBE, depending on the gradients involved.
4. Photometric color calibration (NOTE–if the image is OSC Narrowband, I'll do photometric color calibration assuming this)
5. NoiseXterminator (80% Denoise / 15% Detail. The 90% / 15% default seems a bit too "polished" or "wax"-looking to my eyes)
6. Starnet++ V2
on Stars-only image (labeled "STARS")
7. Curves transformation to boost star saturation
8. Curves transformation in RGB/K to reduce star size. Literally grab the 50% point and drag it ~10% to 20% towards the lower-right corner.
on starless image (labeled "GLOW", for the object)
9. Clone stamp in case there are dust motes flats didn't take care of.
10. Stitch it back together in PixelMath, GLOW+STARS becomes "PS1" for "Pre-Stretch Image 1", saved as TIF. I save this intermediate step in case I need to go back to it at a later time.
NON-LINEAR STAGE:
1. Series of histogram stretches, though not mimicking the auto-stretch. Tried GHS, but I'm getting error messages when I tried running it.
2. Stretch until histogram peak is ~20% over from left.
3. Bring shadows back in and stretch more, maintaining 20% peak position without clipping.
4. Run Starnet++ V2 again.
On stars-only image:
5. Curves transformation for another color saturation boost.
on starless image–
6. Extract luminance channel.
7. Clip luminance until ~50% of pixels are clipped. Basically at the peak of the histogram. Clip point varies on how cleanly the background/nebula split is.
8. Label the result as luminance mask and invert mask on starless image.
9. Compress / darken starless image a bit. Generally prefer background to be ~0.1, as 0.15 looks too bright and 0.05 looks too dark to my eyes.
10. Keep luminance mask on, but remove inversion.
11. Adjust saturation and/or hue in CurvesTransformation
12. A bit of unsharp masking on just the starless image if appropriate. ALWAYS lower than the 2.00/0.80 defaults.
13. Stitch back together with PixelMath
14. Final adjustment on background with either CurvesTransformation RGB/K or arcsinh stretch.
15. Run EZ Star Reduction, set # morphological transformation iterations to 7 – NOTE: I do star reduction in linear stage because aggressively hitting a non-linear image with repeated rounds of EZ Star Reduction just doesn't look right.
I'm not saying this is the best method. It's just the one I'm using now.
I'd be curious to know what the PixInsight veterans out there think could be improved upon. If astrophotography has taught me anything, there's ALWAYS a better way to do things.
Thoughts?
Clear Skies,
Phil
Question is on my workflow. As time has gone by, I've gotten better at processing, but there's always room for improvement. Finding the time to sit down to watch all 653,242 PixInsight YouTube tutorials (each spanning a length of geological time) is admittedly a hurdle. I mean, I've picked up at least a few things along the way. Personally, I find VisibleDark's the easiest to follow and understand, but to each their own.
Basically, I'm wondering if you were taking OSC images (I have ZERO desire for mono imaging), how would you judge the following workflow?
For stacking, I just use Deep Sky Stacker and take the unmodified Autosave.TIF file straight into PixInsight. Why not use PixInsight? Because DSS is just too easy and learning everything in PXI is 10X harder than it needs to be.
Once it's in PixInsight, here's what I do. ***NOTE–I haven't posted any photos with this method yet, but I have reprocessed several images using this method. I'll post some of these in the next few days. I've just seen the improvement in the images using the following method, but am wondering if I can still improve upon it.***
LINEAR STAGE:
1. Screen Transfer Function / autostretch.
2. Dynamic crop for stacking artifacts/margins.
3. ABE/DBE, depending on the gradients involved.
4. Photometric color calibration (NOTE–if the image is OSC Narrowband, I'll do photometric color calibration assuming this)
5. NoiseXterminator (80% Denoise / 15% Detail. The 90% / 15% default seems a bit too "polished" or "wax"-looking to my eyes)
6. Starnet++ V2
on Stars-only image (labeled "STARS")
7. Curves transformation to boost star saturation
8. Curves transformation in RGB/K to reduce star size. Literally grab the 50% point and drag it ~10% to 20% towards the lower-right corner.
on starless image (labeled "GLOW", for the object)
9. Clone stamp in case there are dust motes flats didn't take care of.
10. Stitch it back together in PixelMath, GLOW+STARS becomes "PS1" for "Pre-Stretch Image 1", saved as TIF. I save this intermediate step in case I need to go back to it at a later time.
NON-LINEAR STAGE:
1. Series of histogram stretches, though not mimicking the auto-stretch. Tried GHS, but I'm getting error messages when I tried running it.
2. Stretch until histogram peak is ~20% over from left.
3. Bring shadows back in and stretch more, maintaining 20% peak position without clipping.
4. Run Starnet++ V2 again.
On stars-only image:
5. Curves transformation for another color saturation boost.
on starless image–
6. Extract luminance channel.
7. Clip luminance until ~50% of pixels are clipped. Basically at the peak of the histogram. Clip point varies on how cleanly the background/nebula split is.
8. Label the result as luminance mask and invert mask on starless image.
9. Compress / darken starless image a bit. Generally prefer background to be ~0.1, as 0.15 looks too bright and 0.05 looks too dark to my eyes.
10. Keep luminance mask on, but remove inversion.
11. Adjust saturation and/or hue in CurvesTransformation
12. A bit of unsharp masking on just the starless image if appropriate. ALWAYS lower than the 2.00/0.80 defaults.
13. Stitch back together with PixelMath
14. Final adjustment on background with either CurvesTransformation RGB/K or arcsinh stretch.
15. Run EZ Star Reduction, set # morphological transformation iterations to 7 – NOTE: I do star reduction in linear stage because aggressively hitting a non-linear image with repeated rounds of EZ Star Reduction just doesn't look right.
I'm not saying this is the best method. It's just the one I'm using now.
I'd be curious to know what the PixInsight veterans out there think could be improved upon. If astrophotography has taught me anything, there's ALWAYS a better way to do things.
Thoughts?
Clear Skies,
Phil
