AI user turns “classified” text into a four-era fake evidence photo archive
MrJuart · reddit · 2026-07-25
A Reddit user describes an AI image project built around a fictional constraint: the model was told it had access to classified material it could not quote in text, only reinterpret visually.
Using ChatGPT 5.6 Sol in a high-effort mode, the user generated a four-era visual archive that looks like recovered evidence: damaged prints, archive scans, surveillance frames, field photos, leaked contact sheets, and forgotten negatives. They say some images unintentionally echo UFO-culture themes and real reported cases, making the series feel like a “forbidden photo archive” rather than polished concept art.
More from Multimodal
- Midjourney makes V8.2 its default model, touting bigger leaps in style and personalization — midjourney · 2026-07-25
- Uber details closed-loop evals for a multimodal food agent at scale — AI Engineer · 2026-07-25
- Reve image model 2.1 adds direct color picking and beats rivals on palette accuracy — DavidmComfort · 2026-07-25
- A Reddit user shares a Jarvis stack built from Whisper, Qwen, and Bernini — Alive_Ad_3223 · 2026-07-25
- Opus 4.1’s art-session energy becomes a shareable AI joke — liminal_bardo · 2026-07-25
- SuperSplat Leverages WebGPU and Voxel Collision for Massive Point Cloud Rendering — willeastcott · 2026-07-25