Optimizer

Make the prompt do the work

Veo, Seedance, Kling, Sora — anywhere a shot has to be described before it can be rendered.

Your current prompt

**Overall Concept and Narrative**
A three-act treatment for the music video of {{song_title}}, organized around {{visual_theme}} as the single visual thread that recurs and evolves across verse, chorus, and bridge. The emotional beat tracks the song's own arc: restraint in the verses, release in the chorus. Scale is carried by the visual theme intensifying each time it returns, not by increasing production size.

**Key Scenes and Transitions**
Act 1 (verses): start frame introduces {{visual_theme}} in its smallest, quietest form - movement is slow and observational - end framing holds on {{artist_name}} noticing or engaging it for the first time. Transition: a hard cut timed exactly to the first chorus downbeat into Act 2. Act 2 (choruses): start frame is {{visual_theme}} at full scale, movement is full-body performance choreography or camera movement matching the song's energy - end framing is a wide shot showing the theme's full transformation. Transition: a slow dissolve into Act 3. Act 3 (bridge and outro): start frame returns to the intimacy of Act 1's framing but with {{visual_theme}} now changed by everything that happened in Act 2 - movement settles - end framing is a still, held final image.

**Visual Style and Aesthetics**
Aesthetic references Spike Jonze's handling of a single sustained visual metaphor - restrained until it isn't. Primary light source shifts across acts: a cool 5600K daylight-balanced source in Act 1, shifting to warmer 3200K practicals and colored gel washes in Act 2, returning to neutral 5600K in Act 3. Shadow depth deepens through Act 1, flattens under performance lighting in Act 2, deepens again in Act 3. Light haze is introduced only in Act 2 to add scale to the lighting. Grade moves from desaturated in Act 1 to highly saturated in Act 2 and back to a muted, slightly warmer version of Act 1's palette in Act 3.

**Motion and Dynamics**
Camera in Act 1 is slow, observational, mostly static with rare slow pushes. Act 2 camera work is dynamic - whip pans, handheld energy matched to the beat, occasional speed ramps on hits. Act 3 camera returns to Act 1's stillness. {{artist_name}}'s physical performance escalates from stillness to full choreography and back to stillness. What is explicitly NOT moving across the whole video: {{visual_theme}} itself never leaves frame for more than a few seconds at a time, anchoring the story. Tension curve rises steadily through Act 1, peaks and sustains through Act 2, and releases slowly through Act 3.

**Audio and Timing**
Duration matches the full runtime of {{song_title}}, delivered at 24fps for a cinematic feel. Dialogue is the song's vocal performance, lip-synced by {{artist_name}} where the framing shows the face. Three cues: the song's actual mix drives all timing decisions above; a brief moment of the instrumental isolated and slowed at the Act 2-to-Act 3 transition; and room tone or practical sound design audible only in the non-vocal seconds of Act 1 and Act 3.

**Technical Details**
3840x2160 downsampled to 1920x1080 delivery, 24fps, anamorphic-style 2.39:1 letterboxing for Act 2's performance sequences only, returning to standard 16:9 for Acts 1 and 3. Light bloom and particles (dust, confetti, or theme-appropriate debris) are used deliberately in Act 2 only. No HDR master required for standard platform delivery.

**Negative Prompts**
inconsistent visual theme that disappears for long stretches, mismatched lip-sync, flat even lighting across all three acts, low-resolution or upscaled-looking footage, generic performance choreography unrelated to {{visual_theme}}, jitter from over-stabilized handheld, fast cuts faster than the song's actual tempo, oversaturated skin tones, watermarks, crowd shots not called for in the treatment, continuity errors in wardrobe across acts, audio sync drift

Template · 3 variables

3 of 3 still blank. Anything left blank is sent to the model as {{placeholder}}.

Describe your ideal output (optional)

ModeModel

7 blocks · 3 clarifying questions · Auto rotates through the models your plan unlocks, switching every two runs per mode.

⌘ + Enter

Video frame

What you will get back

01Overall Concept and Narrative
One paragraph: the emotional beat, the scale or contrast that carries it, and a summary of the camera arc. A single shot if the prompt says so, otherwise a three-scene arc.
02Key Scenes and Transitions
Start frame → movement → end framing. Multi-scene reads "Scene 1 … Transition: <named move> into Scene 2 …".
03Visual Style and Aesthetics
Genre language, the primary light source and its color temperature, shadow depth, atmosphere (fog, dust, haze), the grade (teal-versus-red and the like), and named references.
04Motion and Dynamics
Camera verb plus pace, the subject's gait, what is explicitly NOT moving, and the tension curve across the duration.
05Audio and Timing
Duration, fps (24 or 30 for cinematic, 60 for slow motion), dialogue yes or no, and three concrete ambient or score cues.
06Technical Details
Resolution, aspect ratio, texture targets, light bloom / volumetrics / particles, HDR. Every concrete item from the ideal-output description reappears here.
07Negative Prompts
A comma-separated list of roughly 10–12: wrong time of day, crowds, wrong color, handheld or shaky, fast cuts, low resolution, wrong mood (smiling), jitter, artifacts.

Plus 3 clarifying questions you can answer to re-run with more facts.