Astrophotography diagnosis of Lune: Over-denoising
ProcessedAucun4s03 mars 2026
The Doc examined this image of Lune (processed, Aucun, 4s). Estimated overall technical quality: 7/10. 1 defect found: Over-denoising (severity 2/5).
&w=1920&q=75)
Info
- Cible
- Lune
- Date
- 03 mars 2026, 21:30
The target is the Moon itself at full phase (100% illumination), so lunar brightness is never a handicap here; on the contrary it provides plenty of signal that makes the 4s exposure more than sufficient. Full phase does flatten the relief: frontal lighting reduces shadows and thus topographic contrast along the terminator (absent here). For more striking crater detail, aim for a quarter or gibbous Moon where grazing shadows reveal relief. The reported 2.13 seeing is fine for short-video lunar capture.
Setup
- Type d'image
- Traitée
- Télescope
- Unistellar Odyssey
- Caméra
- Unistellar Odyssey
- Filtre
- Aucun
- Monture
- Unistellar Odyssey
- Exposition
- 4s
- Phase de lune
- 100%
- Seeing
- 2.13"
- Notes
- L'image est générée par le process automatique Unistellar en interne par l'Odyssey à partir d'une capture video de 4s. Puis traitée vite fait dans Darktable
The Unistellar Odyssey is an alt-az smartscope with on-board processing, perfectly suited to a quick lunar capture from a 4s video. On such a bright target, the short exposure avoids saturation and freezes turbulence, which is consistent. The lunar disk fills the field nicely without being clipped, a sign of framing well matched to the instrument's focal length. The downside of the automatic pipeline is uncontrollable denoising that smooths flat areas; to go further, a longer raw video processed via lucky imaging (AutoStakkert then wavelets) would yield finer detail.
The diagnosis in detail
The image is technically sound for this type of gear and target. Focus is correct, the major ray systems and crater rims are sharp across most of the disk, and the limb is clean with no optical distortion or stacking doubling. No sign of tilt, coma or tracking issue is expected or visible: the target is point-like at the sensor scale, very bright, and frozen by a 4s exposure.
The only real area for improvement is post-processing. The Odyssey's automatic pipeline applies denoising that, combined with a Darktable edit, gives the maria a slightly smoothed, plastic look with some softened crater edges in dark areas. It is not penalizing at this display size but it caps the genuinely achievable level of detail.
To step up, the main lever is acquisition and processing: capturing a longer video and processing it via lucky imaging (selecting the best frames, stacking, wavelet sharpening) will recover far more fineness than the smoothed automatic render.
Priority actions
- Reduce or disable additional denoising in Darktable to preserve mare texture
- Capture a longer lunar video and process it via lucky imaging (AutoStakkert then wavelets in RegiStax/Siril)
- Aim for a quarter or gibbous phase to benefit from grazing terminator shadows
