Video model · Black Forest Labs ·
FLUX 3 is a video model by Black Forest Labs. In Endless it costs from 90 credits per video and is available on the Hobby plan and up.
FLUX 3 Video is Black Forest Labs' first video model, making 5 to 20 second clips at up to 1080p with synced sound from text, images or a clip.
FLUX 3 Video is the first video model from Black Forest Labs, the team behind the FLUX image models. It generates clips of 5 to 20 seconds at up to 1080p with a synchronized soundtrack of dialogue, effects and ambient sound. Write a line of dialogue in quotes and a character speaks it with lip sync, in 13 languages that include English, Spanish and Portuguese. One clip can also hold several shots and camera angles.
You can start from a prompt alone, a starting image to animate, a first and last frame for the model to connect, or an existing clip to continue from its final frames. Teams use the frame pair to control exactly where a product reveal begins and ends, and continuation to add a beat to a shot they already like. Audio can be switched off for a silent clip.
FLUX 3 Video accepts a short phrase, a long paragraph, a timeline or a labeled list. A short prompt like "a red fox leaping through fresh snow, telephoto" leaves framing, motion and mood to the model, which can give unexpected results. A longer prompt gives you more control over the scene, camera, pacing and sound. Black Forest Labs suggests you start short to explore an idea, then add only the details you want to control. Length alone does not improve a clip: a prompt with too many instructions can make the motion less coherent.
Write the prompt as directions for a scene, not as a list of objects. Say what happens, how the subject moves, what the camera does and what the place feels like. Use concrete nouns and verbs the camera can see. Vague adjectives leave those choices to the model.
For most clips, one sentence that follows the same pattern is the easiest to revise:
[camera] shot of [subject] [action] in [environment]. [motion and light details]. Audio: [sounds].
When the action has to happen at specific moments, write it as a timeline. Keep each beat achievable: two or three beats for a 5-second clip.
0–2s: locked wide shot of a still harbor at dawn
2–4s: a slow push-in as gulls lift off the water
4–5s: the sun clears the horizon and the camera settles
FLUX 3 Video can include several shots and camera angles in one clip. For a multi-shot sequence, number each shot, give it a time range, and mark a hard cut where the angle changes. Describe the main character once with the same words in every shot so their look stays consistent. Close with one line on the overall style and palette.
Start with one framing term, one camera movement and one clear subject action. Black Forest Labs notes that too many camera instructions in one sentence usually make the shot harder to read. Terms the model responds to include:
The model also covers looks beyond cinematic: camcorder footage, dashcam, thermal imaging, cel animation, claymation, motion design. Name the medium early in the prompt, and describe on-screen typography as part of the scene.
Audio is generated with the frames and is on by default. A scene full of visible actions, like footsteps, impacts, rain or engines, produces clearer audio than an abstract scene. When the sound matters, name each layer you want: speech, ambience, effects tied to visible actions, and music with its place in the mix. Name specific sources ("rain ticking on a metal awning") instead of mood words. Ask for a sound rather than for silence: "quiet room tone" can turn into static.
Put spoken words in quotation marks and say who says them. A visible speaker gets lip sync. An off-screen voice needs a "voiceover" cue, or the quoted line may appear as text in the frame. Add "no on-screen text, no subtitles" when you want to avoid this. Direct the voice with a few concrete details (age, register, recording setup, delivery) rather than words like "engaging" or "professional", which tend to produce a polished, ad-like read. To use another language, name the language next to the line. You can quote the line in its native script.
Speech takes time. A short line in a longer clip is safer than a script that fills every second.
In Endless you can set:
In your workspace, open Video and pick FLUX 3 in the model picker.
Write a prompt with the subject, camera and motion, or start from an image you upload or generate.
Generate, then refine in chat or on a canvas. Every clip is saved to your workspace assets.
Grouped by use case. Each one opens in Endless with FLUX 3, ready to edit and generate.
Slow orbit around a brushed copper kettle on a stone countertop at dawn. Steam begins to curl from the spout as low window light slides across the metal. Audio: a rising whistle, a ceramic cup set down on stone, birdsong outside.
Macro shot of a fountain pen nib touching cream paper. Ink spreads into a single handwritten word, "Monday", as the camera pulls back to reveal a tidy oak desk. Audio: the scratch of the nib and a clock ticking in the room.
Locked-off top-down shot of a white sneaker on wet asphalt at night. A puddle beside it ripples as rain starts, and passing headlights sweep orange light across the leather. Audio: rain on the street and a car hissing past.
Lazy Susan shot of a glass bottle of cold-brew coffee turning on a matte black turntable under one hard spotlight. Condensation beads on the glass and catches the light on each turn. Audio: a soft electronic pulse, low in the mix, and one ice cube clinking.
Tracking shot beside a folding bicycle being unfolded on a train platform in four quick movements. The rider clicks the frame shut, mounts and rolls out of frame as the train doors close. Audio: metal clicks, a platform announcement in the distance, the doors sliding.
A cardboard box on a kitchen table opens its own flaps one by one, and a small potted fern rises out on a column of soil. Warm morning light, playful stop-motion feel. Audio: paper rustling, a tiny chime when the pot settles.
Medium close-up of a night-shift baker dusting flour off her hands at a steel table. She looks at the apprentice beside camera and says, "Twenty more minutes. Don't open the oven." Dry, tired delivery. Audio: oven fan hum and a radio playing faintly in the back room. No on-screen text, no subtitles.
Locked shot of a rain-streaked bus window at night, city lights smearing past. An off-screen voiceover says once, "I moved here for a year. That was nine years ago." A woman in her forties, low voice, recorded close and dry. Audio: bus engine and rain. No music, no on-screen text.
Two-shot at a fishing harbor: an old fisherman in a yellow coat mends a net while his grandson in a red cap holds the end. The fisherman says in Portuguese, warm and patient, "Devagar. O nó tem que ficar firme." The boy nods and pulls the rope tight. Audio: gulls, hull creaks, water against the pier.
Close shot of a tailor in a narrow workshop, chalk in hand, looking up from a jacket. He says in Spanish, quietly amused, "La manga está bien. Eres tú el que creció." Natural Spanish delivery, no announcer voice. Audio: a sewing machine stops, scissors set down on wood. No subtitles.
A park ranger crouches at the edge of a frozen lake, tapping the ice with a pole. She turns to camera and says, "Hear that hollow sound? That means stay off." Practical, unhurried delivery, recorded outdoors. Audio: wind across the ice and a low, distant crack.
Over-the-shoulder shot of a call-center agent at a dim desk with a headset on. The agent in a gray cardigan speaks first in English, "Can you read me the number again?", then, after a pause, switches to Hindi in the same calm voice. Audio: keyboard clicks and air-conditioning hum. No on-screen text.
Shot 1 [0–3s]: wide shot of a lighthouse keeper climbing a spiral stair with a lantern. Shot 2 [3–6s]: hard cut to a close-up of his hand turning a brass key. Shot 3 [6–10s]: the beam sweeps across a stormy sea from a high angle. Audio: wind, footsteps on iron, a heavy mechanical clunk, waves.
Shot 1: handheld follow of a skateboarder rolling through an empty parking garage at night. Shot 2: low angle as she ollies over a painted curb. Shot 3: wide shot as she glides out into the street under sodium lights. Audio: wheels on concrete, the board's pop, distant traffic.
Shot 1 [0–4s]: establishing shot of a mountain hut in falling snow at dusk. Shot 2 [4–8s]: medium shot inside as a climber pours tea for two friends by a stove. Shot 3 [8–12s]: close-up of the frosted window as their laughter fades. Audio: wind outside, kettle hiss, a crackling stove.
Shot 1: aerial view of a vineyard at harvest, workers moving between rows. Shot 2: close-up of shears cutting a heavy bunch of dark grapes. Shot 3: tracking shot of a crate carried to a tractor. Shot 4: wide shot of the tractor driving into the golden evening. Audio: clipping shears, chatter, an engine.
Shot 1 [0–5s]: a young chess player sits alone at a park table, moving pieces against herself. Shot 2 [5–10s]: reverse shot as an old man sits down across from her without a word. Shot 3 [10–15s]: close-up of his first move, then her small smile. Audio: pigeons, a distant fountain, the click of the pieces.
Shot 1: dashcam view of a delivery van entering a narrow village street at sunrise. Shot 2: cut to the driver leaning out to hand a parcel to a baker. Shot 3: wide shot of the van reversing slowly between flower boxes. Audio: engine idle, a shop bell, the van's reversing beep.
Start from the attached photo and bring it to life: the woman at the window turns her head toward the camera as steam rises from her cup, and curtains move in a light breeze. The camera pushes in slowly. Audio: street sounds from below and a spoon against porcelain.
Animate the attached product photo: the camera arcs a quarter turn around the watch while a light reflection travels across the glass. Everything else stays still. Audio: a soft mechanical tick, no music.
Use the first image as the opening frame and the second as the closing frame. The empty studio fills with people arriving, setting up easels and starting to paint, as the light shifts from morning to afternoon. Locked-off wide shot, natural sound of chatter and chairs.
Start on the first image of the closed flower bud and end on the second image of the open bloom. Timelapse look, the petals unfold in one continuous motion while the background stays fixed. Audio: a low, soft ambient tone.
From the start frame of the sealed envelope to the end frame of the open letter on the table: hands open the envelope, unfold the letter and lay it flat. Close-up, eye level, warm desk lamp light. Audio: paper tearing and unfolding.
Continue the attached clip without a cut: the runner keeps the same pace, crosses the finish line and slows to a walk with her hands on her head. The camera keeps tracking alongside. Audio: the crowd swells, then breathing and footsteps.
1990s hand-drawn cel animation: a paper airplane flies through a sleepy town at dawn, delivering itself through an open window onto a sleeping child's pillow. Bold ink lines, painted backgrounds. Audio: a light flute melody and wind.
Camcorder footage from 2003 of a family birthday party in a backyard: a slightly shaky zoom on a boy blowing out candles, timestamp in the corner, flash of a disposable camera. Audio: off-key singing, laughter, a dog barking.
Motion design title card on a flat warm-yellow background: the word "OPEN" builds itself from sliding black bars, then the letters bounce once and settle in the center. Clean sans-serif type. Audio: crisp clicks timed to each bar, a short synth chime at the end.
Claymation-style scene of a small clay fox baking bread in a tiny kitchen, the dough visibly rising in the oven. Visible fingerprints on the clay, slightly choppy frame rate. Audio: a ticking timer, a cheerful ding, the oven door creaking.
Thermal camera view of a city rooftop at night, bodies glowing white as two people set up a telescope, cold sky in deep blue. The palette shifts as a warm chimney puffs smoke. Audio: wind and distant traffic, no music.
Hand-painted watercolor animation of a rainstorm moving across rice fields: colors bleed as the rain falls, and a farmer under a straw hat walks along a dike. The camera pans slowly left. Audio: rain growing heavier, frogs, distant thunder.
by ByteDance
30-second clips in one pass, guided by up to 50 references
From 51 credits per video
by ByteDance
Clips of 4 to 15 seconds steered by image, video and audio references
From 16 credits per video
by Kuaishou
Multi-shot clips with spoken dialogue, motion transfer and a 4K mode
From 27 credits per video
FLUX 3 next to other models: price, maker and plan side by side.
Start free, create a project, and bring your brand’s references together. Or book a demo to explore Endless with your team.