Definition — The text instruction describing what a generation should depict — scene, wardrobe, mood, composition. With a trained persona, prompts describe the situation only; the identity comes from the model.
With a trained persona, the prompt's job shrinks in the best way: it no longer has to carry the character's appearance, only the situation. 'Reading a menu outside a small café, late afternoon sun, candid half-smile' is a complete prompt — the face, build and overall identity are supplied by the model.
Prompts that work tend to share a shape: a concrete action, a specific setting, wardrobe if it matters, and one or two atmosphere cues (light, mood). Concrete beats abstract — 'holding a chipped white mug' generates better than 'cozy vibes'. Prompts that fail are usually keyword soup: long comma lists of adjectives that pull the image in conflicting directions.
Prompts also interact with Forge Styles: the style fixes the look of the shoot, so the prompt should stay inside it rather than fight it — describing a beach scene inside a studio-portrait style produces confused output. The first photoshoot guide shows how prompt and style divide the work.
The practice of writing and systematically refining prompts to get reliable, repeatable results from a generative model — concrete language, deliberate structure, and one change at a time.
A secondary prompt listing what a generation should avoid — unwanted objects, styles or artifacts. The model steers away from the listed concepts instead of toward them.
A reusable style template that fixes scene, lighting, framing and mood for a generation, so a content series stays visually coherent. Presets separate the persona (trained once) from the look (chosen per shoot).
See it live with 100 free credits
No card required. Generate with showcase models on the free tier, then train your own persona on any paid plan.