Saturn — October 12–13, 2026

Four selected captures, including the preferred 11:44:48 PM recording. Use the menus to compare RGB-balanced and original color at either stack percentage. All 38 videos and 833,854 source frames were screened from beginning to end. Original captures are preserved.

Oct 14 double-sharpen audit: Two new comparison controls (light blend / medium wavelets) use 24% median denoise, with NO unsharp mask and NO Nik. These are baseline controls, not yet genuine light-wavelet or genuine Nik revisions. Existing medium-strong finished + Nik may include an extra 82% unsharp mask before Nik. Processing manifest.

Oct 14 workflow update: All 16 no-extra-sharpen comparison controls are available in the menus above (four captures x two stack percentages x light/medium). Matching lossless TIFF files are prepared for a fresh genuine Nik pass. Those new Nik results are pending, not published yet. Earlier genuine Nik options remain accessible for comparison.

New 10% finishing test: Smoother gentle-detail and smoother slightly-crisper options come from the clean RegiStax wavelets + denoise baseline. Both are modest scripted finishing experiments (not genuine Nik). Select either to lock the stack to 10%; assess natural rings, globe smoothness, and any halos at 1x. Exact processing settings. Genuine Nik on this cleaner baseline is still pending.

New real Nik 9 comparison: RegiStax clean 10% + genuine Nik (gentle, Structure 8%) is available for the preferred 23:44:48 capture only. The actual Nik output uses Adaptive 12%, Output Strength 27%, Structure +8%, Local Contrast 0%, Focus 0%. It was verified to differ from the baseline. Switching to another capture or 17% stack automatically returns to the clean RegiStax baseline. Export record.

New Nik comparison: Structure +24% version for the preferred 10% stack is a stronger experimental test, not the selected best. Adaptive 12%, Output Strength 27%, Structure 24%, zero local contrast/focus. Settings.

Actual RegiStax 6 preset test: Two new options use the same unsharpened 10% stack; 2021 Saturn preset retains historic per-layer denoise; reconstructed control preset is approximate and has no separate median denoise. Test record.

Reproduced the control wavelets exactly: The corrected native RegiStax preset uses Initial Layer 2, Step 1, Default Linear filter, sliders 35/22/10/1/1/1. The new intermediate option is RGB-matched and 80% control + 20% 2021, not a native preset effect. The separate median denoise remains a finishing step. Audit.

New actual-pixel color options: for the preferred 10% Saturn capture, choose Control or 80/20 intermediate with subtle, pop, or richer color. These are precise color-only adjustments (saturation, faint warm color separation, subtle midtone contrast), using the original image pixels; no generative AI and no added sharpening. The processing record lists the numerical settings. Your original control remains unchanged.

Final color test: Rich Color control with gentle chroma-only bilateral denoise, wavelet sharpening unchanged. It smooths color fluctuations while keeping brightness-channel detail almost identical (maximum 1/255 luminance rounding). Exact processing recipe.

Actual Photoshop export: Saturation +22 and gentle Curves, both real native editable Photoshop adjustment layers. This option has no chroma denoise and does not replace our richer-color favorite. Settings.

Time selector references: Select 2021 best or Uploaded internet sample to switch directly between the historical/reference images and the 2026 captures. Both references are rotated and resized onto 480×300 black canvases. Processing dropdown does not alter a reference image. The uploaded reference uses a compressed transfer copy; source capture and geometry have not been AI-generated.

True 16-bit TIFF comparisons: The dropdown now includes three browser previews of newly reprocessed 16-bit TIFFs (not converted from 8-bit PNGs). Browser previews are 8-bit for fair comparison; download the TIFFs for continued high-precision editing: Control 16-bit TIFF · Rich Color 16-bit TIFF · Rich + Chroma 16-bit TIFF.

Genuine softer RegiStax wavelets (10% lower): Full 16-bit reprocessing of the original drizzle TIFF in RegiStax 6, using verified Linear Default settings 31.5 / 19.8 / 9 / 0.9 / 0.9 / 0.9, plus separate 24% median blend. The second version adds our Richer Color and chroma denoise, without extra sharpening. Processing record · RegiStax wavelet TIFF · Softer control TIFF · Richer-color TIFF · Wavelet preset.

RGB-balance correction (Oct 15): Both softer RegiStax variants now use RGB gains measured from the matched balanced drizzle stack, applied to the genuine 16-bit wavelet TIFF before median denoise and finishing. This post-wavelet approach avoids a RegiStax import/level failure with the balanced TIFF. The previous unbalanced versions are backed up. Exact values and processing record.

New neutral richer-color test: Uses the RGB-balanced softer-wavelet control, with stronger saturation and gentle chroma denoise but NO added red/blue warmth offsets. The previous warmer Richer Color version is preserved for comparison. Settings · 16-bit TIFF.

Topaz DeNoise AI 3.7.2 Clear tests: processed from the RGB-balanced 16-bit softer RegiStax control on markm75o with Recover Original Detail 0 and Color Noise Reduction 0. Three native 16-bit TIFF exports: Medium denoise / High sharp · Medium denoise / Low sharp · Low denoise / Low sharp.

Topaz Clear Recover Original Detail 10 comparison: both new variants start with the same RGB-balanced, 10%-softer RegiStax TIFF. Clear: Medium denoise / High sharpness / Recover Detail 10 / Color Noise Reduction 0. Second variant adds saturation 1.47x without red/blue warmth offsets or additional sharpening. Both have 16-bit TIFF masters. Recover 10 TIFF · Recover 10 + Rich Neutral TIFF.

October 7 recipe replay (12:30 AM EDT): The capture selector now includes the October 7 best 10% / 1.5x drizzle stack, reprocessed from the genuine original 16-bit TIFF with saved RegiStax 6 softer wavelets (31.5/19.8/9/0.9/0.9/0.9), post-wavelet RGB balance, 24% median finishing, then Topaz DeNoise AI Clear / Medium denoise / High sharpen / Recover Detail 10. Select Topaz Clear: Medium noise / High sharp / Recover detail 10 for a direct comparison with October 13. 16-bit TIFF · Exact recipe. Previous gallery options preserved.

2021 archival favorite recipe replay (Topaz complete): The capture selector now includes the archived 2021 best 28% / 1.5x drizzle TIFF, truly reprocessed from original 16-bit RGB data using the native softer RegiStax wavelet scheme and our 24% median finishing. Genuine Topaz DeNoise AI 3.7.2 Clear / Medium denoise / High sharpening / Recover Original Detail 10 / Color Noise Reduction 0 was also run on the same 16-bit control and exported natively. Select that Topaz processing stage on the 2021 capture to compare with Oct 7 and Oct 13. 16-bit control TIFF · 16-bit Topaz Recover10 TIFF · Settings/progress manifest. Original 2021 reference and all 2026 options preserved.

2021 RGB balance correction (Oct 16): The earlier 2021 processing used the 2026 RGB balance and appeared greenish. A new calibration measured the archived 2021 favorite JPG globe region, preserved luminance and recalibrated the original 2021 native RegiStax 16-bit TIFF before median denoise and a fresh genuine Topaz Clear / Medium / High / Recover 10 pass. Rebalanced Topaz master TIFF · Rebalanced pre-Topaz control TIFF · Exact measured color gains and recipe. The previous greenish Topaz version and all other options remain.

Download displayed PNG · Download matching 16-bit drizzle stack · Download matching 16-bit processed TIFF

Real AutoStakkert 4 stacking with genuine 1.5× drizzle: 320×200 capture → 480×300 output. Corrected AstroSurface profiles: HF size 4, strength 120 / 180; LF size 12, strength 8 / 12. Earlier soft size-1/2 variants have been replaced. Compare actual AstroSurface medium and strong passes with actual RegiStax Linear wavelets (initial layer 2, step 1; sliders 35/22/10, described here as medium–strong). Each finished version is freshly rendered from its corresponding sharpening result using the same despeckle, levels, sharpening and color finishing. Finishing on original stages is Go-inspired. Separate Nik dropdown choices use genuine DxO Nik 9 Sharpener Output (Display, Adaptive 15%, Output Strength 39%) on finished AstroSurface TIFFs; Photoshop itself was not used for those Nik variants. No generated detail. RGB-balanced variants use scripted channel gains measured from bright outer rings in the unsharpened stack, with the same correction across 10% and 17% for each capture. This is ring-reference white balance, not RegiStax Auto Balance. RGB balance measurements.

Download original 12:47:21 AM SER (1.47 GB) · Safe frame indices · SER checksum and capture record · FireCapture settings

See the edge clipping

This three-minute capture around 11:22 PM keeps the original 384×206 image bounds and Saturn's original position. The footer is outside the camera image. Playback is sampled at 25 fps while preserving the original duration; the full-frame audit examined every source frame.

Download / open MP4

Genuine Nik Sharpener Output processing details