
Original document
GPT Image 2 · en
Scene:
A square follow-up field image from the same site, slightly closer and lower than the reference anchor, with careful documentary framing and the same material palette. The camera should feel physically present inside the archaeological space, using natural light, soft depth separation, and one clear human figure to explain scale without turning the image into lifestyle photography.
Subject:
Show the same site identity established in the reference anchor, now with an archaeologist, conservator, or field researcher moving through the scene in practical neutral clothing. The person should be small relative to the monument and clearly secondary to the architecture, water, or painted wall system. Keep the main forms, structural rhythm, and civilizational markers unchanged.
Important details:
Preserve the exact site identity, base geometry, material behavior, and cultural signatures from the reference model. Only vary vantage, gesture, and scale cues. Keep the site recognizably Khmer, with disciplined tower silhouettes, naga-era stone logic, gallery rhythm, and roots that wrap around built forms instead of replacing them with fantasy tree sculpture. Human presence should clarify stair height, wall size, carving depth, or spatial compression while respecting conservation protocol and historical seriousness. Keep any tools understated and generic, with no readable labels. The image must still look like a rigorous archaeological record first and a dramatic scene second.
Use case:
Angkor Wat Overgrown Symbiosis series study 2/9.
Constraints:
No readable text, no watermark, no logos, preserve exact 1:1 aspect ratio, preserve exact site identity from the reference image, no fantasy elves, no glowing magic runes, no steel catwalks, no modern park signage, no fashion editorial styling, no crowd, no cinematic props.How to use
- Copy the complete original prompt, including any parameters.
- Replace the subject, text and composition with your own requirements.
- Use a compatible external tool; check its current access conditions and review the result.
Before you start
- Check whether a reference image is required. A compatible model label does not mean this example was generated with that model.
- Keep names, figures and visible text accurate; results can vary between model versions.