I was experimenting with a variety of methods to stack my ~2h of Andromeda data. Since the results can vary considerably between methods I figured I'd share the results here.
- Version A uses DSS's Bayer drizzle.
- Version B uses RawTherapee to convert RAW files to 16 bit TIFF and then stacked in DSS.
- Version C uses RawTherapee to convert RAW files to 32 bit floating point TIFF and then stacked in DSS with 2x drizzle.
- Version D uses RawTherapee to convert RAW files to 32 bit floating point TIFF and then stacked in DSS without drizzle.
- Version E uses DSS to stack directly from raw files with 2x drizzle.
- Version F uses DSS to stack directly from raw files without drizzle.
- Version G uses Sequator to stack.
- Version H uses RawTherapee to convert RAW files to 16 bit TIFF and then stacked in DSS with 2x drizzle. (Missing from original submission)
All versions had the following steps done after stacking:
Siril:
1) Crop
2) Rotation
3) Background Extracion
4) Starnet star removal
5) Stretching of starless and starmask
6) Save as 16 bit TIFF
Lightroom:
7a) Starless: Final black point and saturation adjustment
7b) Starmask: Purple defringe and saturation adjustment
Siril:
8) Conversion back to FIT
9) Recomposition of starless and starmask with fine tuning of starmask brightness.
10) Save as JPEG
I attempted to make all versions as close as possible but using the exact same settings wasn't possible due to the differences in stacking results. I did NOT use Topaz Denoise AI for this since I wanted a more honest comparison.
- To my eye, the 32 bit float RawTherapee TIFF conversion gives the best overall balance of detail and color although some odd green halos were produced around some bright stars. Also, the color has a slight green tinge.
- 16 bit stacking has excellent color but the image is "blocky" and lacks smoothness in color/brightness gradation presumably due to the lower bit depth. This is somewhat alleviated with 2x drizzle but it is still very noticeable.
- The Bayer Drizzle seems to produce slightly better detail, especially when using a denoise program, but sacrifices color. Bayer drizzle also produces some odd comet like tails to the left of brighter stars. (I had average rather than Kappa-Sigma clipping selected as advised by DSS)
- All versions stacked with DSS directly from RAW show poor color performance but have low noise. and are smooth
- Sequator gives what is arguably the best color performance but is the noisiest image by far.
Comments