Photography Signal: Quiet Authority

Hook: Most AI photography looks expensive and says nothing. Here's the difference between a render and a photograph.

The signal

Photography is the largest cluster in our library — and the easiest to fake badly. The signal that separates a real photographic frame from a glossy AI render isn't resolution. It's restraint: one light, one subject, one decision the image is making.

Reference breakdown

• Lighting — single dominant source, soft falloff, shadows allowed to go dark instead of being "fixed." The image isn't afraid of black.
• Composition — subject off-center, generous negative space, one clear point where the eye lands. Nothing fights for attention.
• Mood — quiet, observed, slightly withheld. It feels like someone was there, not like a prompt asked for "beautiful."
• Material — real grain, lens character, gentle imperfection. Texture you could almost touch.

Where slop happens

AI photography fails when everything is lit, everything is sharp, and everything is "perfect." Perfect is the tell. Real photographs choose what to lose.

Prompt translation

1. One-light direction — describe a single source and its quality (window light, late sun), not three.
2. Permission to be dark — ask for deep shadow, low key, retained blacks. Stop over-lighting.
3. Lens character — name grain, shallow depth, optical imperfection instead of "8k ultra sharp."
4. Subject restraint — one subject, lots of empty space. Describe what's NOT in frame.

Keep: one light, negative space, real grain.

Kill: "ultra detailed, perfect, 8k," everything-in-focus, decorative clutter.

AIDGART curates the references. You decide what's worth generating.

→ Drop01 — a curated photography direction, so you start from taste, not noise.