AI video styles guide
How to use styles on Lucy
Pick a style, copy a phrase, generate on Lucy, or build panels via storyboard to video. Deep craft: Prompt Guide · camera moves · UGC ad path.
UGC / creator phone aesthetic
Look: Bathroom mirror or desk selfie, slight handheld sway, window light, pores visible, imperfect framing.
Prompt that works
slight handheld camera sway at eye level, soft natural phone-selfie energy — not shake-cam, keep her face readable, morning window light, iPhone UGC look, pores and peach fuzz visible
What usually fails: Beauty-filter skin, three-point studio softboxes, locked tripod perfection, logo redesign mid-clip.
Product consistency: Lock a packshot still first; hold product label toward camera in a static or tiny handheld beat.
Polished cinematic
Look: Motivated key light, shallow depth cues, one elegant move (dolly/push/orbit), filmic color.
Prompt that works
cinematic medium close-up, slow gentle push-in toward her eyes, dramatic side light with soft falloff, photoreal 4K, one move only, hold final frame clean
What usually fails: Stacking ‘8K masterpiece cinematic hyperreal’ adjectives; multiple moves in 3 seconds.
Product demo
Look: Clean desk/bathroom, product readable first, demo action second, face as proof.
Prompt that works
medium selfie-style handheld, she holds the product toward camera and turns the pack so the label reads, soft room tone, no beauty filter
What usually fails: Logo melt, shade change, floating hands. Lock product @Image still.
Product consistency: Same packaging description + same angle still across shots; animate after stills.
Unboxing
Look: Top-down or 45° desk, hands enter, lid lift, react to camera.
Prompt that works
high angle desk unboxing, static lock-off then slight tilt down to the product reveal, hands only until the logo faces camera sharp, soft daylight
What usually fails: Box morphing, brand mark rewrite, endless lid loops.
Testimonial talking-head
Look: Chest-up, eye-level, soft room, micro fidgets, spoken line optional on engines that support native audio.
Prompt that works
medium close-up talking head, slight handheld sway, soft window key, natural blinks and tiny weight shifts, she nods once then a real small smile
What usually fails: Mannequin stillness, over-wide lens distortion, perfect studio backdrop when you wanted bedroom.
Lifestyle B-roll
Look: Inserts: pour, walk, city haze, product on table — little or no dialogue.
Prompt that works
lifestyle B-roll, slow tracking beside the product on a sunlit table, dust in air, shallow focus, no faces required, one move only
What usually fails: Trying to tell the whole story in one B-roll clip.
Meme / cutty social
Look: Hard cuts between short punches; whip pan or static holds; text added later in editor (don’t ask the model for burned-in captions).
Prompt that works
fast whip pan left with motion blur between two clear sharp end-frames, start sharp and land sharp, use once only
What usually fails: Asking the model for perfect subtitles/logos; long single takes that go mushy.
Music-video energy
Look: Orbit, crane, silhouette, rhythmic cuts in the edit — sync is edited, not promised.
Prompt that works
camera slowly arcs around her, she stays centered and mostly still while the background slides past, stylized night color, one move only
What usually fails: Claiming frame-perfect beat sync from one generate. Stitch to your track.
More study hubs
FAQ
- Which AI video style should I pick for ads?
- Most paid social still wants UGC / creator phone aesthetic or product demo. Use cinematic for brand films and hero moments. Match style words to lighting and camera — not adjective spam.
- How do I keep the product consistent?
- Stills first (GPT Image on Lucy). Same packaging description, same preferred angle, bind as a product ref when the workflow supports it, then animate. See /consistent-character and the Prompt Guide.