Au pied d’une falaise noir brun presque plus sombre que le charbon, la surface apparaît déchirée en blocs anguleux, corniches en surplomb et fractures verticales, tandis qu’une étroite fissure éclairée par le Soleil projette un jet pâle de gaz et de poussière dans le vide. Dans cette gravité infime, les grains scintillants et les cailloux arrachés à la croûte riche en organiques et en glace sale montent en lentes trajectoires balistiques, comme suspendus un instant sous un ciel d’un noir absolu où aucune atmosphère n’adoucit la lumière. De minces traces de givre bleu-blanc soulignent les bords des crevasses et les recoins ombragés, témoignant de la sublimation de glaces volatiles chauffées près du périhélie, un processus qui sculpte ici des falaises abruptes, des pitons acérés, des terrasses effondrées et des dépressions fraîches sur un noyau d’à peine quelques kilomètres. Les ombres, nettes comme des lames, accentuent l’étrangeté du paysage : on a l’impression de se tenir sur une ruine cosmique en activité, où la matière primitive du Système solaire s’échappe silencieusement de la roche sombre vers l’espace.
Comité de revue scientifique
Chaque image est examinée par un comité d'IA pour sa précision scientifique.
GPT
Image: Ajuster
Description: Ajuster
Scientific accuracy: The image is consistent with an active comet surface in a low-gravity, vacuum environment—there are localized, jet-like dust/gas emissions emerging from a fractured, dark nucleus surface. The very dark, low-albedo terrain and strong contrast from direct illumination also fit the general expectation for Wild 2’s nucleus imagery. However, several specifics do not align well with the requested “cliff vent in eruption” scenario: (1) the dominant activity looks more like broader, fountain-like plumes and bright surges rather than a narrow, vent-driven jet from halfway up a cliff crack; (2) the vent geometry and “cliff wall” framing are ambiguous—most of the scene reads as a boulder-strewn slope/area rather than a clearly jagged vertical cliff with a vent on the wall; (3) the sky contains a bright, clearly prominent central light (interpretable as the Sun) while also showing a star field; Wild 2 nucleus images typically emphasize the Sun direction but a “perfectly black sky with faint stars still visible” is not inconsistent, yet the overall lighting suggests more terrestrial-style or stylized illumination. Scale is also not verifiable, though the terrain spans only a few km is plausible for a close-up concept. Visual quality: The rendering is high quality and largely photorealistic in texture detail (craters/blocks, dustiness, sharp shadows). The main issue is physical realism of the plume: the ejection appears overly bright, structured, and “liquid-like” in places (and the particulate arcs are not clearly slow-graceful dust grains); this suggests a visual-art effect more than sublimation-driven grain trajectories. Caption accuracy: The caption describes a narrow pale jet emerging from a crack on a cliff face, lifting glittering dust/pebbles in graceful arcs, with thin bluish-white frost tracing the vent. In the image, emissions are present, but they are not clearly constrained to a narrow wall vent, and the dominant plume brightness/shape differs from the described tight jet. Also, the scene shows more global surface dust/frost-like highlights than specifically “thin frost and exposed icy seams” tracing a single vent. Overall: good comet-like concept and darkness/contrast are on point, but the vent morphology and plume physics/placement are off enough that both the image and caption need adjustment rather than full approval.
Claude
Image: Ajuster
Description: Approuver
I largely agree with GPT's assessment but want to refine several points and add new observations. SCIENTIFIC ACCURACY: The surface albedo is convincingly low — the dark charcoal-brown coloration is consistent with Wild 2's measured geometric albedo of roughly 0.03, one of the darkest objects in the solar system. The sharp, knife-edged shadows in a vacuum (no atmospheric scattering to soften them) are correctly rendered. The presence of exposed bright icy/frost patches contrasting against the dark matrix is scientifically defensible — Stardust flyby data confirmed heterogeneous surface composition with localized volatile-rich regions. The layered cliff structure with overhangs is plausible given Wild 2's observed mesas, pinnacles, and steep scarps. The Sun rendered as a point-like but intensely bright source against a black starfield is accurate for heliocentric distances of ~1.5–2 AU. HOWEVER: the plume dynamics are the primary scientific concern. The ejected material appears too massive, too bright, and too 'turbulent' — sublimation jets on comets produce collimated, low-density streams of sub-micron to millimeter particles, not the churning, quasi-liquid eruption column depicted here, which visually resembles a geyser or volcanic fumarole more than cometary outgassing. Pebble-sized fragments in the ejecta are scientifically defensible in low gravity (~0.001 m/s²), but the dense, opaque nature of the central plume overstates particle concentration. The frost deposits and icy seams are well-executed and appropriately restrained. VISUAL QUALITY: Texture work on the regolith, boulders, and cliff faces is excellent and genuinely photorealistic. The color grading — cold bluish whites against warm reddish-brown rocks — adds compositional realism and is consistent with Wild 2's observed reddish organic-rich surface material alongside exposed ice. The main artifact is the plume itself: it has an almost fluid-simulation quality with internal swirling structure that breaks photorealism and looks more like a particle system from a game engine than a vacuum sublimation event. The particle arcs around the base also appear too symmetrical and too bright. The Sun's lens flare is acceptable stylistically but slightly overdone. CAPTION ACCURACY: I diverge from GPT here and rate the caption more favorably. The caption accurately describes: the fractured dark crust, angular blocks, a jet from a cliff crack, lifted grains and pebble-sized fragments in slow arcs, thin bluish-white frost on ledges, the low-reflectivity surface, hard sunlight with knife-sharp shadows, mesas/pits/overhangs, black sky with faint stars, and the sublimation mechanism. The image does depict all of these elements — the vent IS emanating from a cliff face crack approximately mid-wall, frost patches ARE visible, the sky IS black with stars. The caption's description of a 'tight, pale jet' is the weakest match since the rendered jet is neither particularly tight nor pale, but the overall correspondence between description and image is strong enough for 'approve' rather than 'adjust.' RECOMMENDATION: Adjust the image primarily to narrow and de-densify the plume column, reduce its internal turbulent structure, and make the particle trajectories more sparse and graceful — consistent with actual cometary jet observations from Rosetta/67P and Stardust/Wild 2 data. The cliff geometry and surface texture are strong and should be preserved.
Grok
Image: Ajuster
Description: Ajuster
I concur with both GPT and Claude on the core strengths and issues, adding targeted insights from the provided image. SCIENTIFIC ACCURACY (adjust): The dark, low-albedo (~.03) charcoal-brown surface with angular blocks, fractures, mesas, and overhangs aligns excellently with Stardust mission imagery of 81P/Wild 2's rugged, ~5 km nucleus—heterogeneous mix of dust, organics, and patchy ice is spot-on. Knife-sharp shadows from direct sunlight at ~1.5-3 AU are accurate (no atmospheric diffusion), and the black sky with faint stars fits deep-space vacuum. Low-gravity (~10^-4 m/s²) arcs of ejecta are plausible for micron-to-pebble grains. However, the 'cliff vent' is marginally defined—the jagged wall exists, but the vent emerges from a broad fracture/cavity at the base rather than a 'narrow crack halfway up the wall'; plume is collimated but too dense/opaque/turbulent, resembling a fluid geyser (e.g., water-like flows at base) over sparse sublimation jets seen in Wild 2 data. Frost is present but more patchy/scattered than 'thin bluish-white seams tracing the vent.' Sun appears as a small, bright disk (realistic size at comet distance), not a lens flare artifact. VISUAL QUALITY (adjust): Photorealistic textures (regolith dustiness, rocky fractures, icy glints) are outstanding, with convincing color grading (dark reds/browns vs. cold whites). No major artifacts, but plume physics breaks immersion—internal glow/swirls and base 'puddles' look like VFX simulation errors, not vacuum outgassing; particle arcs are graceful but overly luminous/dense. Scale feels right for a ~100m cliff on a km-scale body. CAPTION ACCURACY (adjust): Description mismatches on key specifics: no clear 'tight, pale jet' from 'halfway up' a cliff (it's basal/broad/white-hot); arcs are present but not 'glittering grains/pebbles lingering'; frost is visible but not distinctly 'tracing the vent/ledges'; 'perfectly black sky with faint stars' matches, but 'silent, active, uncannily alive' tone fits while surface isn't 'exceptionally dark, low-reflectivity' enough amid bright ejecta. Disagreement with Claude: caption overstates vent precision/placement. RECOMMENDATION: Preserve surface/cliff/sky; regenerate plume as narrow, dimmer, sparser jet from mid-cliff crack with ballistic dust trails (ref: actual Wild 2 jet photos). Minor caption tweak for plume geometry.
Gemini
Image: Ajuster
Description: Ajuster
I concur with the committee's consensus regarding the high quality of the surface texture and the successful capture of the comet's low-albedo, rugged morphology. The geological features (angular blocks, mesas, and sharp-edged shadows) are excellent. However, I must emphasize the critical failure in 'plume physics': the emission is rendered as a dense, turbulent, and semi-opaque vapor column, which is physically incompatible with a vacuum environment where gas expands ballistically without fluid-like vortices. The plume needs to be re-rendered as a transparent, collimated stream of discrete particles (dust and ice grains) to achieve scientific accuracy. Regarding the caption, the term 'tight, pale jet' is a specific scientific descriptor that the current image fails to realize; the caption should be adjusted to reflect the 'broad, basal emission' currently visible, or the image must be adjusted to match the 'narrow crack' description. I recommend a tighter focus on the vent morphology to better represent the localized sublimation process observed in Stardust mission data.
Matania — Synthèse
Image: Ajuster
Description: Ajuster
The committee agrees the image is strong in overall cometary realism: Wild 2's dark low-albedo surface, sharp vacuum shadows, fractured cliffs, boulders, and patchy frost are all convincing and scientifically appropriate. The main weakness is the eruption plume, which is rendered too dense, bright, and fluid-like for cometary outgassing; it reads more like a geyser or volcanic vent than a sparse sublimation jet in vacuum. The cliff-vent geometry is also not as clearly defined as the prompt requires. On the caption, most reviewers found it broadly aligned with the scene, but the wording overstates the jet's narrowness, placement, and frost-tracing details relative to the actual image. Overall, the surface and lighting are successful, but the plume physics and vent specificity need refinement.