몽골 사구에서 프로토케라톱스를 습격하는 벨로키라토르 — 백악기

약 7,500만~7,100만 년 전 후기 백악기 캄파니아절의 몽골 자도크타층 사구 지대에서, 깃털로 덮인 소형 드로마이오사우루스류 벨로키랍토르 몽골리엔시스가 몸을 낮춘 각룡류 프로토케라톱스 앤드루시를 급습하는 극적인 순간이 펼쳐집니다. 붉고 황금빛 바람무늬 사구와 단단한 사구 사이 평지, 드문 초기 속씨식물 관목과 낮은 침엽수성 식물이 보이는 이 건조한 환경은 물보다 모래와 바람이 지배하던 사막 분지였습니다. 벨로키랍토르의 낫 모양 발톱과 날개깃, 프로토케라톱스의 강한 부리와 목 프릴은 당시 포식자와 초식동물이 어떻게 맞섰는지를 생생히 보여 주며, 이 장면은 실제 화석 기록으로도 유명한 백악기 아시아 내륙 생태계의 한 단면을 담고 있습니다.

AI 과학 위원회

이 이미지와 캡션은 역사적, 과학적 정확성을 평가하는 독립적인 AI 모델 위원회에 의해 검토되었습니다.

GPT 이미지: 조정 캡션: 조정 Mar 30, 2026

The image depicts a Late Cretaceous-style desert with wind-rippled sand dunes, sparse low shrubs, and a burial-prone dune/sand sheet setting. The overall paleoenvironment mood is plausible for Mongolia’s Late Cretaceous (often associated with the Djadokhta/Gerbil formations’ arid conditions), and the scene contains no obvious anachronistic artifacts or human elements. The feathered dromaeosaur appearance is broadly consistent with current evidence that many dromaeosaurs were feathered. However, the feathering reads as a very large, fully fluffed wing/wing-like surface across the back and arm region, which may be more exaggerated than typically depicted for non-avian dromaeosaurs (most reconstructions show quill-like or pennaceous feathers but not a large “prop” wing at this scale). Also, the Protoceratops silhouette looks reasonably ceratopsian, but details like facial/neck frill shape and body proportions are not sufficiently constrained in the image to be confidently “Protoceratops andrewsi” specifically.

For the caption, the timing (75–71 Ma) and region (Late Cretaceous Mongolia, Northern Hemisphere) are reasonable. The ecosystem description as arid dune fields with dust and rapid burial potential is broadly compatible with interpretations of these deposits. The main issue is the specificity and implied certainty: calling the association “famous” and tying it directly to Djadokhta Formation predator-prey implies a particular Velociraptor–Protoceratops interaction as a widely documented pair, but the classic evidence is the Djadokhta/Gerbil material popularly known as “fighting dinosaurs” (dromaeosaur vs. ceratopsian), and taxonomic labels (Velociraptor vs. other dromaeosaurs; Protoceratops vs. related ceratopsians) can vary depending on the study. Additionally, the caption’s vegetation references “conifer-like plants”; the depicted plants are generic shrubs rather than clearly identifiable conifer morphology, so the text may over-specify compared with what’s actually shown.

Because there are no major, obvious scientific violations, a complete redo isn’t necessary; instead, refine the prompt/description to (1) moderate feather display to more typical non-avian dromaeosaur reconstructions, (2) avoid overclaiming taxon-specific “Velociraptor mongoliensis” and “Protoceratops andrewsi” identity unless the artwork clearly reflects distinguishing traits, and (3) soften or qualify the predator-prey claim to match the tentative nature of the classic Djadokhta/Djadokhta-group evidence.

Claude 이미지: 조정 캡션: 조정 Mar 30, 2026

The caption is largely accurate. The 75–71 Ma date range is well-supported for the Djadokhta Formation. Calling the interaction a 'famous predator-prey association preserved in Djadokhta Formation fossils' is appropriate — the 'Fighting Dinosaurs' specimen (IGM 100/25) is indeed one of the most celebrated Cretaceous fossils and unambiguously documents a Velociraptor–Protoceratops encounter. GPT's colleague raised doubt about whether this is 'widely documented,' but it is genuinely one of the most famous fossils in palaeontology, so the caption's language is defensible. The description of the landscape as burial-prone with wind-rippled red sand and sparse vegetation is well-supported. However, 'conifer-like plants' in the caption is slightly misleading given that the image shows what appear to be broadleaf shrubs — this discrepancy between caption and image should be reconciled. The species names Velociraptor mongoliensis and Protoceratops andrewsi are the correct and expected taxa for the Djadokhta Formation and their use is appropriate. I partly disagree with GPT's suggestion to soften taxon-specific claims — these are the canonical Djadokhta taxa and their use is scientifically sound. The main adjustments needed are: reconcile 'conifer-like' vegetation description with the broadleaf shrubs shown, and optionally note the feathered wing-arm display posture is speculative.

Grok 이미지: 승인 캡션: 승인 Mar 30, 2026

The image effectively captures the arid dune fields of Late Cretaceous Mongolia's Djadokhta Formation, with wind-rippled red sands, blowing dust, sparse low shrubs, and a harsh desert atmosphere that aligns with paleoenvironmental reconstructions. No anachronisms are present—no modern elements, architecture, or technology. The Velociraptor mongoliensis is depicted with appropriate feathering (proto-wings on arms, quill-like coverage on body and tail fan), agile build, and ambush posture, consistent with fossil evidence and modern reconstructions. Protoceratops andrewsi shows accurate stocky ceratopsian proportions, low-slung body, beaked mouth, modest frill, and defensive posture, matching Djadokhta specimens. Vegetation is limited to low shrubs and conifer-like plants, plausible for the period (e.g., Ginkgoales, conifers). Visual coherence is strong, with dynamic action, realistic lighting, and burial-prone sand collapse adding plausibility to the 'Fighting Dinosaurs' scenario. Geology (eolian dunes) and biology (feathered dromaeosaur, ceratopsian herbivore) are scientifically sound.

The caption is factually precise: 75–71 Ma matches Djadokhta Formation dating; the 'famous predator-prey association' directly references the iconic IGM 100/25 'Fighting Dinosaurs' fossil (Velociraptor vs. Protoceratops locked in combat), a cornerstone of paleontology. Landscape and flora descriptions (arid dunes, sparse shrubs/conifers, blowing dust) align perfectly with both image and geological evidence. No misleading claims; the level of detail provides excellent context without overstatement. Species names and ecological dynamics are spot-on for the region/period.

I agree with Claude's approval of the caption's specificity—the 'Fighting Dinosaurs' is unambiguously Velociraptor mongoliensis and Protoceratops andrewsi, making taxon claims robust, not tentative. Disagree with both colleagues' 'adjust' votes: feathering is not exaggerated (matches evidence from related dromaeosaurs like Zhenyuanlong); it's typical for artistic reconstructions. Vegetation ('conifer-like') fits the shrubs shown and Djadokhta flora; no major discrepancy. No significant issues warrant adjustment—image and caption are accurate and ready for educational use.

Gemini 이미지: 승인 캡션: 승인 Mar 30, 2026

The image and caption are highly accurate for the Late Cretaceous of Mongolia (Djadokhta Formation). The environment depicted—arid, wind-rippled dunes with sparse vegetation—is consistent with interpretations of this geological unit. The fauna, featuring a feathered *Velociraptor mongoliensis* and *Protoceratops andrewsi*, is canonical for this setting and time period (75–71 Ma). The depiction of the *Velociraptor* with significant arm feathering/proto-wings is supported by evidence from closely related dromaeosaurs and is a reasonable modern artistic interpretation, contrary to the slight hesitancy noted by GPT.

The caption is excellent. It correctly identifies the time, location, and the iconic predator-prey relationship (

Matania 요약 이미지: 조정 캡션: 조정

The committee agrees that the scene broadly fits a Late Cretaceous Mongolian desert setting: the wind-rippled dune field, blowing dust, sparse low vegetation, and general arid ambiance are all plausible for the Djadokhta Formation. The featured animals are also broadly correct in concept, with a feathered dromaeosaur and a stocky ceratopsian engaging in a classic predator-prey confrontation.

IMAGE ISSUES: 1. GPT noted that the dromaeosaur’s feathering reads as somewhat exaggerated, with a very large, wing-like display surface across the back and arms; it may be more elaborate than typical non-avian dromaeosaur reconstructions. 2. GPT and Claude both flagged that the vegetation is shown as generic shrubs, while the caption calls them “conifer-like plants”; this is an image-caption consistency issue. 3. GPT noted that the Protoceratops identity is not fully constrained by the visible anatomy, since the frill, facial details, and proportions are not sufficiently diagnostic to confirm Protoceratops andrewsi specifically from the image alone.

CAPTION ISSUES: 1. GPT flagged that the text may over-specify taxonomic certainty by asserting Velociraptor mongoliensis and Protoceratops andrewsi as the exact taxa, when the artwork itself does not clearly show diagnostic features sufficient to prove those exact species. 2. GPT also suggested the predator-prey association should be phrased more cautiously, since the classic fossil is a famous Djadokhta-group fighting-dinosaurs specimen and taxonomic assignment can be discussed with some nuance. 3. GPT and Claude both identified a vegetation mismatch: the caption says “conifer-like plants,” but the image shows generic low shrubs/broadleaf-looking plants rather than clearly coniferous morphology. 4. GPT noted that the caption’s “famous predator-prey association preserved in Djadokhta Formation fossils” is broadly correct but slightly overstates certainty/typicality without qualification. 5. The date range 75–71 Ma, Late Cretaceous Mongolia, and burial-prone dune environment were not flagged as errors and were generally accepted.

FINAL VERDICT: adjust. The scene is scientifically strong overall and does not require a full regeneration, but it should be refined to reduce overstatement and align the vegetation/taxon wording more closely with what is visually shown. The image mainly needs moderation of the feather display and better alignment between depicted flora and the caption. The caption mainly needs either a softer wording on taxon certainty and predator-prey attribution or a clearer statement that the scene is a reconstruction inspired by the famous fighting-dinosaurs fossil, plus correction of the vegetation description.

Other languages

Back

백악기