Photography Prompt Rules

A photographic prompt works by describing the physical act of capturing an image — camera, lens, light, framing — not by describing the finished result in abstract terms. "A photorealistic image of a woman" tells the model almost nothing useful; "35mm lens, natural window light, slight motion blur on her hand" tells it how a camera actually would have rendered that moment, which is what produces a convincing photograph rather than an illustration that merely resembles one.

Camera and lens language

• Specify a focal length appropriate to the subject and mood: wide (16–35mm) for environmental context and a sense of place, standard (35–50mm) for a natural, undistorted human perspective, telephoto (85–200mm) for compressed backgrounds and portrait isolation. Naming a specific focal length does more work than saying "close-up" or "wide shot."
• Specify aperture when depth of field matters to the image: a wide aperture (f/1.4–f/2.8) for a shallow-focus subject with a blurred background (bokeh), a narrow aperture (f/8–f/16) for a scene where foreground and background both need to be sharp.
• Name a camera format or stock when it should shape the image's character — "shot on 35mm film," "medium format," "shot on a vintage rangefinder" — each carries a real, learnable visual signature (grain, color rendition, sharpness falloff) distinct from a generic digital-clean look.
• Include real photographic terms describing the shot type (candid, environmental portrait, macro, long exposure) rather than only describing subject matter — the shot type carries technical implications (motion blur, framing distance, focus behavior) a plain subject description doesn't.

Lighting vocabulary

• Name the light source and its quality specifically: golden hour (warm, low-angle, soft shadows), overcast/diffused (soft, even, minimal shadow), hard midday sun (harsh shadows, high contrast), softbox (studio-even, controlled falloff), rim light (edge-lighting separating subject from background).
• Specify light direction relative to the subject (front-lit, side-lit, backlit) when it matters to the mood — backlighting for silhouette or rim glow, side lighting for dimensional texture and shadow detail on a face or object.
• Describe color temperature explicitly when relevant (warm tungsten interior, cool blue dusk light, neutral daylight) — mismatched or unstated color temperature is a common source of an image reading as artificial or flat.
• Avoid vague lighting descriptors like "nice lighting" or "good light" — they carry no information the model can act on; every lighting note should specify source, direction, quality, or color temperature.

Composition terms

• Use standard composition vocabulary the model has strong associations for: rule of thirds (subject off-center on a grid intersection), leading lines (a road, fence, or shoreline drawing the eye toward the subject), negative space (deliberate empty area balancing the subject), framing (a doorway or foliage framing the subject within the shot).
• Specify camera angle and subject placement explicitly (low angle looking up, eye-level, overhead/top-down, three-quarter view) — angle changes the emotional read of a subject as much as lighting does, and leaving it unstated leaves the result to chance.
• State the intended crop or framing distance (extreme close-up, medium shot, full body, wide establishing shot) since this anchors both composition and how much environmental context the model includes.

Realism anchors

• Include physically grounding details that a real camera would capture: slight imperfections (skin texture, fabric wrinkles, environmental dust or reflections), natural asymmetry, and consistent shadow direction matching the stated light source.
• Reference a specific real-world context (a described location, weather condition, time of day) rather than a generic backdrop — specificity constrains the model toward plausible, coherent detail instead of a generic composite.
• Where the subject includes hands, text, or complex reflective surfaces, keep the composition and framing choices favorable to those elements (avoid extreme close-ups on hands, avoid requiring long legible text) since these remain the most common failure points regardless of prompt quality.

What makes AI photos look fake, and how to avoid it

• Oversaturated, uniformly high-contrast color with no tonal variation is a classic tell — counter it by naming a real color grade or film stock, or explicitly requesting muted, naturalistic color rather than defaulting to whatever the model's baseline style produces.
• Unnaturally perfect symmetry, skin, or surfaces read as synthetic — counter with explicit texture and imperfection cues (visible pores, fabric creases, slight asymmetry, environmental grime) rather than leaving surfaces to render at a default airbrushed smoothness.
• Light and shadow that don't agree with each other (a shadow falling the wrong direction relative to the stated light source, or multiple inconsistent light sources) is one of the fastest ways a viewer's eye catches a fake — keep the lighting prompt internally consistent and don't stack contradictory lighting descriptors.
• Background elements that look plausible from a distance but incoherent up close (garbled architecture, impossible object arrangements) are harder to fully prevent, but are reduced by keeping busy backgrounds slightly out of focus (shallow depth of field) rather than demanding equal sharpness across a complex scene.
• A "too clean" overall composition — nothing out of place, no environmental noise — often reads as synthetic precisely because real photographs rarely look that tidy; a stray object, an uneven surface, or ordinary background clutter adds credibility a pristine scene lacks.