Saturn October 18 · Processing comparisons
Three best centered 180-second captures. Evaluate each in 3–7 presets, with October 7 and October 13 included as color/detail references.
Compare captures and processing presets
Each capture keeps its own curated 3–8 processing choices, including the original unsharpened stack. Change captures independently to compare different nights or stack percentages.
Presets for the selected left capture
Processing notes, camera details and ranking
The 1:59 AM 20% Topaz blend is the preferred October 18 trial. October 7 is provided as a color reference; the captures are not geometrically registered.
October 18: all 51 SERs sampled over their duration; 26 were eligible 180-second centered captures. The 1:59 AM, 2:15 AM, and 2:12 AM videos were fully evaluated by AutoStakkert 4 and stacked best 10%, 1.5× drizzle. The pre-Topaz control used genuine RegiStax Softer10 native wavelets, saved RGB gains and 24% median denoise. The ring-whitening trials measured and applied a further, limited channel balance in the outer rings; previous gain was not applied twice.
The lighter October 7 options blend the actual Topaz master TIFF with the genuine pre-Topaz control in a 90/10, 80/20, or 65/35 ratio; they are not independently rerun Topaz models. Preserve the original Oct 7 white-ring appearance and ring detail as your preferred reference.
Topaz Clear in this lab means Medium noise reduction / High sharpening / Recover Detail 10 / Color Noise Reduction 0. Some historical capture presets use alternative RegiStax / AstroSurface / Nik/Levels; names identify them. Treat available TIFF masters as sources of truth; browser previews are 8-bit. All original SER captures and older masters are preserved.
October 18 full capture information · RGB and ring white recipe · October 7 recipe
New Oct 18 11:01/11:04 PM SER quality and server processing details
The two 180-second videos show substantial Saturn ROI drift, including clipping. Strict 10px screening found only about 5.6–5.8% centered frames. On the server, a custom Python frame-registration stacker selected the best 250 of those sampled safe frames, with the alternatives using all 641–642 eligible sampled frames. Eight available comparison choices per recording are clearly labeled as conventional, non-AI, and processed with 1.5x interpolation—not native AutoStakkert drizzle, RegiStax, or Topaz. Ring RGB balance was calibrated for this session, not inherited blindly from Oct 13. These are experimental previews, not new favorites.
October 18, 11:01 and 11:04 PM — genuine native processing details
Four new groups are from actual AutoStakkert 4 1.5× drizzle (10% or 17%), true native RegiStax Softer10, and native Topaz Clear Medium/High Recover Detail 10. The new 20%, 50%, 90%, and 100% processing choices blend the genuine native Topaz result with its pre-Topaz source, rather than simulating Topaz. The additional white-ring RGB/channel balancing is reproducible 16-bit numerical finishing. The original unsharpened AS4 TIFF remains a selectable reference. Tracking drift reduced ideal frame coverage, but these are full native stacks rather than the earlier strict-sample Python tests.
Show original full Processing Lab and all experiments
Nothing was deleted or merged. The legacy Processing Lab retains the full historical controls, tracking audits, multiple per-image processing schemes and research notes.
Open full original Processing Lab · Open full Oct 18 legacy comparison · Open full Oct 9 experimental comparison