These are reusable skeletons, not finished prompts. Copy one, swap the placeholders, and paste it straight into Luma Ray3 inside Dream Machine (also available in Adobe Firefly), the reasoning-driven 2026 video model. Every clip runs silent — Ray3 has no native audio, so add sound in your editor afterward.

How the placeholders work. The parts you fill in use [BRACKETS], like [SUBJECT] or [SETTING] — replace those with your own words and delete any you do not need. The brackets are fill-in markers only; they are not camera syntax. Ray3 has no bracket tags — you describe every camera move in plain English inside the sentence, such as "slow dolly in", "orbit clockwise", or "crane up to reveal". Ray3 reads every word literally, so keep one main action and avoid contradictions. Set aspect ratio, resolution, and duration in the panel. Test wording in Draft Mode (about five times faster and cheaper), then Upgrade to Hi-Fi for the final render. For the full roundup of ready-made prompts, see our best Luma prompts collection.

Advertisement

The master template

Every template below is built from one five-part skeleton: subject, action, scene, camera movement, then style and mood. Learn this shape and you can write any Ray3 prompt from scratch. For a printable reference, see the Luma prompt cheat sheet.

1. The five-part master skeleton

[SHOT SIZE] of [SUBJECT] [SPECIFIC ACTION] in [SETTING], [CAMERA MOVEMENT]. [LIGHTING], [COLOR/TONE], [MOOD], [STYLE DETAIL].

Which blanks matter most: [SUBJECT], [SPECIFIC ACTION], and [CAMERA MOVEMENT] carry the shot; keep the action to one clear beat. [SHOT SIZE] = wide / medium / close-up. [CAMERA MOVEMENT] is plain English like "slow dolly in". Filled example: Medium shot of a lone astronaut in a scuffed white suit turning slowly toward the camera in a derelict space station corridor, slow dolly in. Cold blue emergency lighting, desaturated steel tones, tense mood, cinematic film grain. (16:9, 1080p, 10s)

2. Minimal three-part variant

[SUBJECT] [ONE ACTION] in [SETTING], [CAMERA MOVEMENT]. [MOOD/STYLE].

Which blanks matter: when you want speed, keep only [SUBJECT], [ONE ACTION], and [CAMERA MOVEMENT] — Ray3 fills in sensible lighting on its own. Add detail only where the default lets you down. (16:9, 720p, 5s)

Cinematic b-roll

Atmospheric cutaways and texture inserts. Lean on one plain-English camera move and hold the action to a single element so the frame stays clean. See more in our cinematic b-roll prompts.

3. Slow reveal establishing shot

Extreme wide shot of [PLACE] with [ONE ELEMENT MOVING], crane up to reveal [WHAT IS BEHIND]. [LIGHTING], [COLOR/TONE], [MOOD], cinematic.

Which blanks matter: [WHAT IS BEHIND] is the payoff of the crane — make it worth the reveal. Filled example: Extreme wide shot of a fog-wrapped mountain village with smoke rising from chimneys, crane up to reveal a snow-capped peak behind, golden-hour light, warm amber tones, serene mood, cinematic. (16:9, 1080p, 10s)

4. Lateral tracking b-roll

Medium shot of [SUBJECT] [ACTION] through [SETTING], tracking shot alongside at a steady pace. Shallow depth of field, [LIGHTING], [COLOR/TONE] grade, [MOOD].

Which blanks matter: [ACTION] and the tracking move should agree in direction; shallow focus keeps the subject clean against a busy background. (16:9, 1080p, 10s)

5. Macro detail insert

Extreme close-up macro of [SUBJECT DETAIL] [SUBTLE ACTION], slow push-in. [LIGHTING], soft focus falloff, [MOOD].

Which blanks matter: keep [SUBTLE ACTION] tiny (steam curling, crema swirling) so the macro stays sharp. (16:9, 1080p, 5s)

Product & ad video

Clean framing on a single product with controlled motion. Use one gentle move and let the surface and light do the selling. For technique on realism, see how to prompt Luma for realistic video.

6. Studio product orbit

[PRODUCT] centered on [SURFACE], orbit clockwise slowly around it. Clean [BACKGROUND COLOR] studio background, soft [LIGHTING] with gentle reflections, premium and minimal.

Which blanks matter: [PRODUCT] and [SURFACE] set the whole frame; the orbit is the only motion, so nothing else should move. Filled example: A matte-black wireless headphone centered on a polished stone slab, orbit clockwise slowly around it. Clean off-white studio background, soft key light from upper left with gentle reflections, premium and minimal. (1:1, 1080p, 5s)

7. Product hero push-in

Close-up of [PRODUCT] on [SURFACE] with [SUBTLE PROP MOTION] around it, slow dolly in toward the label. [LIGHTING], [COLOR ACCENT] rim light, sleek and aspirational.

Which blanks matter: [SUBTLE PROP MOTION] adds life without stealing focus; the dolly should end on [PRODUCT], not the props. (16:9, 1080p, 5s)

8. Product in-use lifestyle

Close-up of [HANDS OR PERSON] using [PRODUCT] to [ACTION] in [SETTING], handheld push-in. Shallow focus on the product, [LIGHTING], natural [MOOD].

Which blanks matter: [ACTION] should be one continuous motion (pouring, wiping, opening); the handheld move reads authentic for lifestyle. (9:16, 1080p, 5s)

Characters & people

People need one clear expression or gesture and stable framing so faces and hands hold together. Keep camera moves gentle.

9. Static portrait with emotion

Close-up of [PERSON WITH 2-3 DETAILS] [ONE FACIAL ACTION], looking [DIRECTION], static locked-off frame. [LIGHTING], [EMOTION] mood, natural skin texture.

Which blanks matter: [ONE FACIAL ACTION] must be singular (a slow smile forming); a locked-off frame keeps the face stable. Filled example: Close-up of an older man with a grey beard and wool cap, a slow warm smile forming, looking off-camera left, static locked-off frame. Soft window light, quiet reflective mood, natural skin texture. (16:9, 1080p, 5s)

10. Character walk-and-talk

Medium shot of [CHARACTER] [WALKING ACTION] through [SETTING] while [GESTURE], tracking shot alongside. [LIGHTING], [COLOR/TONE], [MOOD].

Which blanks matter: pair [WALKING ACTION] with [GESTURE] as one beat; the clip is silent, so add any dialogue as voiceover later. (16:9, 1080p, 10s)

11. Two-character moment

Medium shot of [CHARACTER A] and [CHARACTER B] [SHARED ACTION] in [SETTING], slow push-in between them. [LIGHTING], [MOOD], cinematic.

Which blanks matter: [SHARED ACTION] keeps both figures busy with one event so neither drifts; the push-in draws the eye to the interaction. (16:9, 1080p, 10s)

Advertisement

Nature & landscapes

Landscapes reward a slow move and one weather or light element in motion. Ray3's native 16-bit HDR handles bright skies and deep shadow in the same frame.

12. Sweeping aerial landscape

Aerial wide shot of [LANDSCAPE] with [WEATHER OR LIGHT ELEMENT] moving across it, slow drone glide forward. [TIME OF DAY] light, [COLOR/TONE], [MOOD], cinematic.

Which blanks matter: [WEATHER OR LIGHT ELEMENT] (rolling mist, drifting clouds, moving shadows) gives the frame life; the drone glide should stay slow. (21:9, 1080p, 10s)

13. Nature detail with ambient motion

Close-up of [NATURAL SUBJECT] in [SETTING], [AMBIENT MOTION], gentle handheld drift. [LIGHTING], [COLOR/TONE], [MOOD].

Which blanks matter: [AMBIENT MOTION] (leaves trembling, water rippling) is the whole shot — keep it small and let the light carry the mood. (16:9, 1080p, 5s)

Action & physics

These push weight, momentum, and collision. Keep it to one physical action so the motion stays believable and limbs stay stable.

14. Weighted action beat

Medium shot of [CHARACTER] [ONE PHYSICAL ACTION] in [SETTING], wearing [WARDROBE], tracking shot following the motion. [LIGHTING], [MOOD].

Which blanks matter: [ONE PHYSICAL ACTION] must be a single move (spins once and holds) so the body reads correctly across the clip. (16:9, 1080p, 10s)

15. Impact and collision

[OBJECT] [COLLIDES OR FALLS] onto [SURFACE] and [RESULTING PHYSICS], static locked-off frame, high detail. [LIGHTING], [MOOD].

Which blanks matter: [RESULTING PHYSICS] (shards scattering, splash arcing) is where Ray3's simulation shows — a held frame lets it resolve cleanly. (16:9, 1080p, 5s)

16. Chase through a crowd

[SUBJECT] [RUNS OR MOVES] through [DENSE SETTING] as [CROWD REACTS], handheld push-in with slight shake. [LIGHTING], [MOOD], urgent.

Which blanks matter: [CROWD REACTS] gives the shot depth; the handheld shake adds energy — keep it "slight" so faces do not warp. (16:9, 1080p, 10s)

Vertical social (9:16)

Vertical clips need the subject centered and one punchy action in the first second. Keep the camera move short so it reads on a phone.

17. Vertical hook shot

Vertical framing, [SUBJECT] centered [BOLD ACTION] immediately, quick slow-motion push-in. [LIGHTING], [HIGH-ENERGY MOOD].

Which blanks matter: [BOLD ACTION] must land in the first beat to stop the scroll; center the subject for 9:16. (9:16, 1080p, 5s)

18. Vertical talking-head b-roll

Vertical medium close-up of [PERSON] [SUBTLE ACTION] in [SETTING], static locked-off frame. [LIGHTING], [MOOD].

Which blanks matter: [SUBTLE ACTION] (nodding, gesturing lightly) keeps it natural; the clip is silent, so add voiceover in your editor. (9:16, 1080p, 5s)

19. Vertical POV walk-through

Vertical first-person POV moving forward through [ENVIRONMENT] with [ONE ELEMENT REACTING], steady walking pace. [LIGHTING], [MOOD], immersive.

Which blanks matter: [ONE ELEMENT REACTING] (steam rising, lights flickering past) sells the movement; POV reads well on mobile. (9:16, 720p, 10s)

Image-to-video motion

With image-to-video the uploaded still already supplies subject, setting, and lighting, and the clip follows the source aspect ratio and runs 5s. These skeletons describe motion and camera only — nothing else. See more in our Luma image-to-video prompts.

20. Gentle camera on a still

[ONE SUBJECT] [SUBTLE MOTION], slow dolly in. Everything else stays still.

Which blanks matter: name only [ONE SUBJECT] to move; swap "slow dolly in" for "pan left" or "tilt up" to match your framing. (matches source, 1080p, 5s)

21. Living-still micro loop

Only [AREA] moves: [SUBTLE LOOPING MOTION], camera locked off. Everything else frozen.

Which blanks matter: [SUBTLE LOOPING MOTION] (water flowing, hair drifting) plus a locked camera reads like a clean cinemagraph — pair with Loop for a seamless cycle. (matches source, 1080p, 5s)

22. Slow atmospheric drift

Camera trucks right across the scene while [AMBIENT ELEMENT] [SUBTLE MOTION]. Subject stays in place.

Which blanks matter: [AMBIENT ELEMENT] (dust motes, falling snow) adds depth; the lateral truck moves the camera without altering the subject. (matches source, 720p, 5s)

Frequently Asked Questions

How do I use these Luma templates?

Copy a template, replace every [BRACKETED] placeholder with your own words, and delete any placeholder you do not need. Paste the finished sentence into Dream Machine. Set aspect ratio, resolution, and duration in the interface, not in the text. The brackets are fill-in markers only, not camera syntax.

How do I write camera moves for Luma Ray3?

Describe the move in plain English inside the sentence, such as slow dolly in, orbit clockwise, crane up to reveal, tracking shot alongside, handheld push-in, or overhead top-down. Ray3 reads every word literally, so use one clear move and avoid contradictory directions.

What order do the parts of a Luma prompt go in?

Follow Subject, then Action, then Scene, then Camera movement, then Style and Mood. The master template uses the shape shot size of subject specific action in setting, camera movement, followed by lighting, color, mood, and one style detail.

What duration and resolution can I use in Ray3?

Text-to-video runs 5s or 10s, and image-to-video runs 5s. Resolution options are 540p, 720p, and native 1080p, with 4K available through upscale. Set duration and resolution in the panel, never in the prompt text.

Should I use Draft Mode or Hi-Fi?

Draft Mode renders about five times faster and cheaper, so use it to test wording and camera moves. Once a clip looks right, use Upgrade to Hi-Fi on that same generation for the final render. Ray3 also supports keyframes, Extend to about 30s, Loop, and native 16-bit HDR.

What is the difference between text-to-video and image-to-video here?

Text-to-video builds the whole scene from your words. Image-to-video starts from an uploaded still that already supplies subject, setting, and lighting, so those prompts are 5s and describe only motion and camera, and they follow the source aspect ratio.

Can I add audio in Luma Ray3?

No. Ray3 is silent and has no native audio, so every clip renders without sound. Add voiceover, music, or sound effects afterward in your editor. Writing sound cues into the prompt has no effect.

Advertisement