The highest-quality AI video prompts read like a director's brief: subject, setting, action, then camera setup + motion + look. OpenAI's Sora 2 prompting guide explicitly recommends going "ultra-detailed" with camera setup, grading, soundscape, and motion cues to lock the aesthetic.
This pack gives you ready-to-use prompts for the four most common cinematic moves: Push-In (Dolly-In), Dolly-Out, Parallax, and Handheld. Works with Sora 2, Kling AI, and Veo 3.1 inside QuestStudio.
Best for: Photo → video and short cinematic clips. Internal links: Pillar: Image to Video AI • TikTok workflow • Reels workflow
Quick Start: Universal "Director Blueprint"
If you only copy one thing, copy this:
[SUBJECT] in [SETTING], [CLEAR ACTION].
Camera: [SHOT SIZE], [CAMERA MOVE], [SPEED], [STABILIZATION], [LENS/FOCAL LOOK], [FOCUS BEHAVIOR].
Lighting: [MOTIVATED LIGHT SOURCE], [CONTRAST LEVEL].
Look: cinematic color grade, natural skin tones, realistic motion, clean details.
Output: [ASPECT RATIO], [DURATION] seconds.
Negative: no subtitles, no on-screen text, no watermark.
This works because you're explicitly describing camera work using cinematography terminology (dolly, handheld, tracking, etc.).
Pack 01: Push-In (Dolly-In) "Intimacy + Focus"
A push-in (dolly-in) physically moves the camera closer, which feels natural and builds emotional intensity.
Short Prompt
Medium shot of [SUBJECT] in [SETTING], [ACTION]. Slow, smooth dolly-in (push-in) toward the subject, shallow depth of field, cinematic lighting, realistic motion. [AR], [DURATION]s.
Director Prompt (More Control)
Start wide, slowly dolly-in to a medium close-up on [SUBJECT] as they [ACTION].
Lens look: 35mm, shallow depth of field, soft highlight roll-off.
Focus: locked on eyes, gentle background blur.
Lighting: motivated key light from [WINDOW/LAMP/NEON SIGN], natural shadows.
Cinematic grade, clean details, no text. [AR], [DURATION]s.
Photo → Video Version (Safer Geometry)
Animate a single photo of [SUBJECT] in [SETTING] with a subtle dolly-in and gentle parallax depth, preserve identity and composition, realistic micro-movement, cinematic lighting, no morphing. [AR], [DURATION]s.
Pack 02: Dolly-Out "Reveal + Scale"
Use dolly-out to reveal context: the world around the subject. This is the reverse emotional effect of intimacy.
Short Prompt
Close-up of [SUBJECT] [ACTION]. Slow dolly-out, revealing the larger [SETTING]. Cinematic grade, realistic motion, no text. [AR], [DURATION]s.
Director Prompt
Begin tight on [DETAIL: hands/face/object], then dolly-out to a wide shot that reveals [SETTING DETAIL].
Motion: smooth, steady, no wobble.
Lighting: [LIGHT SOURCE] with soft contrast.
Look: filmic, natural colors, crisp subject separation. [AR], [DURATION]s.
Pack 03: Parallax "Depth From a Still"
Parallax is the quickest way to make a photo feel like a shot with real depth. Your goal is stable geometry with gentle foreground/background separation.
Short Prompt
Create a parallax depth effect on [SUBJECT] in [SETTING]. Slow camera drift, subtle perspective change, realistic depth layers, cinematic lighting, no warping, no text. [AR], [DURATION]s.
Director Prompt (Best for Realism)
Photorealistic scene of [SUBJECT] in [SETTING]. Add foreground/midground/background separation and gentle parallax.
Camera: slow lateral move (tiny left-to-right) + slight push-in, steady.
Focus: subject sharp, background softer.
Look: cinematic grade, natural motion, clean edges, no distortion, no text. [AR], [DURATION]s.
Photo → Video "Preserve Image" Version
Animate this image with parallax only. Preserve the original subject, colors, and composition. Add subtle depth and camera drift. No morphing, no extra objects, no text. [AR], [DURATION]s.
Pack 04: Handheld "Human, Documentary, Real"
Handheld means slightly unstable but controlled movement — it adds a human, documentary feel to your clips.
Short Prompt
Handheld medium shot of [SUBJECT] in [SETTING] [ACTION]. Natural micro-shake, controlled framing, cinematic lighting, realistic motion, no text. [AR], [DURATION]s.
Director Prompt
Handheld tracking feel: camera follows [SUBJECT] as they [ACTION]. Slight handheld micro-shake, but subject stays centered.
Lens look: 24–35mm documentary vibe.
Lighting: practical lights from [SOURCE], natural shadows.
Look: realistic, slightly gritty, clean details, no subtitles. [AR], [DURATION]s.
Photo → Video Version (Avoid "Rubber Face")
From a single photo of [SUBJECT], add very subtle handheld micro-movement and minimal parallax depth. Preserve facial features and edges, no warping, no text. [AR], [DURATION]s.
Bonus: "Combo Moves" That Feel Expensive
These are high-performing combos because they mimic real cinematography blocking.
Combo A: Push-In + Rack Focus
Slow push-in toward [SUBJECT]. Rack focus from [FOREGROUND OBJECT] to the subject's eyes. Cinematic lighting, clean details, no text. [AR], [DURATION]s.
Combo B: Handheld Follow + Reveal
Handheld follow shot behind [SUBJECT] walking through [SETTING]. Slight micro-shake, then a gentle reveal to a wide shot. Cinematic grade, realistic motion, no text. [AR], [DURATION]s.
Model Notes (QuestStudio: Sora 2 / Kling / Veo 3.1)
Sora 2 / Sora 2 Pro
Sora's own guidance encourages "director-style" prompts with camera setup, lensing, grading, and motion cues. Use the Director prompt versions above for maximum control.
Kling AI
Kling prompt guides heavily emphasize camera moves like dolly/push-in and explain the difference between a dolly and a zoom. Stick to one primary move (push-in OR handheld OR parallax) per clip for stability.
Veo 3.1
Replicate's Veo guide recommends plain-language detail and structuring your prompt (subject, context, camera work, mood). Note: Veo 3 outputs 16:9; generate 16:9 and reframe for TikTok/Reels inside your workflow.
Troubleshooting: Why "Cinematic Motion" Fails (and Fixes)
Fail 1: Warping faces or "melting edges"
Fix: Reduce motion intensity. Use "subtle dolly-in" instead of "fast push-in". Add "preserve identity and composition" for photo-based prompts.
Fail 2: Camera feels chaotic
Fix: Remove competing moves. Keep only one: dolly-in OR handheld OR parallax.
Fail 3: Scene looks like it's zooming, not dollying
Fix: Explicitly say "dolly" and avoid zoom language. Dolly is distinct from zoom.
Fail 4: Unwanted subtitles or on-screen text
Fix: Add negatives like "no subtitles, no on-screen text, no watermark".
Copy/Paste Prompt Pack (Ready to Ship)
Use these exactly, then swap placeholders:
Push-In
Medium shot of [SUBJECT] in [SETTING], [ACTION]. Slow smooth dolly-in (push-in), shallow depth of field, cinematic lighting, realistic motion, no on-screen text, no subtitles. [AR], [DURATION]s.
Dolly-Out Reveal
Close-up of [SUBJECT] [ACTION]. Slow dolly-out revealing [SETTING DETAIL]. Steady camera, cinematic grade, realistic motion, no text. [AR], [DURATION]s.
Parallax Depth
Create subtle parallax depth on [SUBJECT] in [SETTING]. Foreground/midground/background separation, gentle camera drift, preserve composition, no warping, no text. [AR], [DURATION]s.
Handheld Documentary
Handheld medium shot of [SUBJECT] in [SETTING] [ACTION]. Controlled micro-shake, subject stays centered, practical lighting, cinematic grade, no subtitles, no text. [AR], [DURATION]s.
