ChatGPT Prompt for Realistic Photos
Generate a photorealistic image of [subject, described with real physical specificity], shot on a [lens, e.g. "85mm portrait lens" or "35mm wide-angle"], [lighting, described directionally and by quality -- e.g. "natural window light from camera-left, soft and diffused"], with a [depth of field, e.g. "shallow depth of field, background softly out of focus"], visible natural texture (skin pores, fabric weave, surface imperfections), unretouched, non-idealized look. Avoid symmetric or overly smooth rendering that reads as artificial.
Getting ChatGPT to produce something that reads as an actual photograph — not obviously AI-generated — depends less on saying “realistic” and more on referencing real photography terms: lens type, natural lighting, and texture.
What this produces
An image rendered with photographic conventions — natural lighting, camera-appropriate depth of field, realistic texture — reducing the “AI sheen” that gives away artificially generated images.
Variations
Two ready-to-copy versions for different subjects.
Generate a photorealistic portrait of [subject, described specifically], shot on a 50mm or 85mm lens, natural window light from one side creating gentle directional shadow, shallow depth of field with the background softly blurred, visible natural skin texture including pores and minor asymmetry, unretouched look — avoid the smooth, airbrushed quality typical of AI-generated portraits.
Generate a photorealistic image of [scene, described with concrete detail], shot as if on a DSLR camera with a standard 35-50mm lens, natural outdoor lighting appropriate to [time of day], realistic textures and cast shadows consistent with that lighting, atmospheric perspective if the scene has depth (distant elements slightly hazier/less saturated). No illustrated, painterly, or overly clean digital quality.
Pair this with ChatGPT Prompt for Image Generation, or explore AI Prompt for Kids Photoshoot if you’re looking for something related.
FAQ
Why do AI images sometimes look “off” even when they’re technically realistic?
Overly smooth skin, too-perfect symmetry, and missing fine texture are common tells — referencing real photography terms and natural texture explicitly helps avoid this.
Does this work for editing an existing photo too?
This prompt is focused on the realism of newly generated images — for editing an existing photo while keeping it realistic, combine these terms with the dedicated photo-editing prompt instead.
What camera terms are most useful to mention?
Lens type (50mm, wide-angle), lighting (natural, golden hour), and depth of field (shallow, blurred background) do the most work toward a photographic look.
