How should a video prompt be structured?

Without Higgsfield: Set subject, style, camera, lighting, mood and aspect ratio together, and for video add references cited by name, character locks and a shot list that gives each shot its framing, depth of field, lighting, palette and action. A common order is subject and action first, setting and lighting second, camera move third and mood last, as in "A barista pours latte art at a café counter … Slow push in. Calm, cozy mood."

Axis: fixed labelled blocks in a set order ... one free-form paragraph

Answers by date: 10 records in 9 sources, newest first; the answers above are the current ones. The advice was refined (2026), never reversed, in these records.

date	who	answer	source
2026-10-24	Adil	The Seedance prompt Claude returns with the Higgsfield prompt skill lists the named elements, then optics for framing, camera for angles and movement, action timing second by second, and lighting, sound and acting.	The Complete AI Filmmaking Masterclass Everyone’s Missing (Full Tutorial) 00:06:07
2026-10-13	Higgsfield	Higgsfield's blog structures a two-character scene prompt as a reference block (one line per reference with name, age, height, traits and an instruction to keep its identity), a count lock, a scale lock, a position lock, timed shots, quoted dialogue with voices, and locations as geography only.	How to Put Two Consistent Characters in One AI Video Scene
2026-10-11	Higgsfield	Higgsfield's blog gives a prompt structure most platforms handle well: subject and action first, setting and lighting second, camera move third, mood or atmosphere last, as in "A barista pours latte art at a café counter … Slow push in. Calm, cozy mood."	How to Get All Your AI Video Models on One Platform?
2026-10-07	Higgsfield	Higgsfield's blog says a strong AI video prompt is closer to a shot list a crew could work from than to a description, built in six ordered steps.	Why Your AI Video Doesn't Follow the Prompt (And How to Fix It)
2026-10-05	Higgsfield	Higgsfield's blog offers a prompt template: "Generate a [length]-second [vertical or horizontal] video of [subject] [action] in [setting]", then Camera, Light and Mood, ending with a request to show the credit cost first.	How to Generate AI Videos Inside ChatGPT in 2026 (Full Guide)
2026-10-05	Higgsfield	Higgsfield's blog says a good video prompt covers six things, subject, action, place and time, shot size and movement, light and mood, each removing a guess the model would otherwise make.	How to Generate AI Videos Inside ChatGPT in 2026 (Full Guide)
2026-09-23	Amina	The first block of the team's video prompt is scene context and style: what happens in the shot in plain sentences, plus the style rules and the physics of the world.	How To Animate a Short Film with Blender + Higgsfield (Breakdown) 00:17:03
2026-10-22	Higgsfield	Higgsfield's prompt generator page says a strong prompt sets subject, style, camera, lighting, mood and aspect ratio together, and a video prompt adds references, character locks and shot lists.	AI Prompt Generator

date	who	answer	source
2026-10-22	Higgsfield	For MiniMax H3, Higgsfield advises citing each reference by name in the prompt and describing the shot like a director.	MiniMax H3 Every input. One take.
2026-10-22	Higgsfield	For maximum consistency, Higgsfield's Sora 2 prompt guide advises detailing each shot like a storyboard, giving its framing, depth of field, lighting, palette and action.	Free Sora 2 Prompt guide & Templates for Viral Videos