FLUX.3 Video

Name: FLUX.3
Author: Black Forest Labs

byBlack Forest Labs

Direct multi-second cinematic sequences with synchronized atmospheric audio, deterministic keyframe interpolation, and seamless video extensions

Direct cinematic scene

How FLUX.3 works

From prompt direction and reference framing to draft validation and high-fidelity rendering.

Frame the scene

Describe your visual action, camera trajectory, and desired soundscape, or upload start, end, or extension video clips.

Set duration and fidelity

Select duration from 5 to 20 seconds, choose aspect ratio, and toggle draft mode for rapid low-cost blocking.

FLUX.3 model with compact input, controls, and generated output

Generate audiovisual video

Produce your multimodal sequence up to 1080p with synchronized environmental audio and realistic physics.

What FLUX.3 is good at

Engineered for end-to-end audiovisual generation, precise keyframe control, and seamless narrative extensions.

Unified audiovisual synthesis

Generate complete video scenes from 5 to 20 seconds at up to 1080p with organic camera motion, coherent lighting dynamics, and integrated environmental soundscapes.

Keyframe interpolation and loops

Lock your opening and closing frames with start and end image inputs to interpolate controlled physical transformations or create seamless infinite video loops without drift.

Narrative video extension

Supply an existing shot to lengthen your narrative timeline up to 20 seconds while preserving spatial physics, subject identity, and acoustic room tone across the extended sequence.

Rapid draft pre-visualization

Test camera blocking, subject motion, and sound cues at 720p resolution with accelerated turnaround before committing to final high-resolution 1080p master renders.

Made with FLUX.3

A collection of cinematic takes, physical simulations, and product moments generated with native spatial audio.

Fashion editorial in high-contrast tones

Cinematic film moment with synchronized environmental audio

Industrial design hero showcase with rotational motion

Architectural interior study with shifting daylight

Game action frame with bold color pairing

What people build with FLUX.3

How directors, sound designers, and creative studios deploy extended multimodal video generation.

Cinematic film pre-production

Directors and cinematographers generate multi-second story sequences with natural lighting and synchronized foley to validate camera movement and emotional tone.

Fashion and lookbook motion

Apparel labels and creative agencies produce high-impact runway motion and textured fabric studies with rich studio lighting and ambient audio.

Product and hardware commercials

Hardware and furniture brands showcase product finishes, kinetic mechanisms, and tactile details through continuous orbital camera moves.

Seamless looping visual loops

Visual artists and dynamic branding studios design infinite seamless background loops with deterministic start and end keyframes.

Short-form creative iteration

Content teams test shot compositions, camera velocities, and audio cues rapidly in 720p draft mode before committing compute to final 1080p renders.

FLUX.3 specs

Plans and credits

Inputs: Prompt (text), Start Image (image), End Image (image), Video
Output: video
Duration: 5–20 seconds
Resolution: 720p, 1080p
Aspect ratios: 21:9, 2:1, 16:9, 4:3, 1:1, 3:4, 9:16
Audio: Generates audio
Cost: 413–14,603 credits per run

Frequently asked questions

Draft mode renders at 720p preview resolution with faster turnaround and substantially lower credit cost, making it ideal for checking camera velocity, shot pacing, and audio timing. Standard mode delivers full-fidelity 1080p output with refined surface micro-textures and pristine dynamic motion for client-ready delivery.

Use text-to-video when generating an entirely new scene from scratch. Use first-and-last frame mode when you have two distinct reference images and need deterministic interpolation between them—such as building seamless looping clips with identical start and end frames, or guiding a deliberate visual transformation across 5 to 20 seconds.

Extend mode takes an existing video clip and generates the next sequence of continuous action. To maintain visual continuity, provide stable base footage without hard cuts and describe the subsequent narrative progression and acoustic environment clearly in your prompt.

FLUX.3 synthesizes visual frames and synchronized auditory latents together in a single foundation architecture. Describing concrete acoustic details—such as mechanical hums, footsteps on gravel, or room reverberation—steers the native audio generator to match on-screen physical dynamics.

FLUX.3 was created by Black Forest Labs and introduced in mid-2026 as a unified multimodal foundation model for 5 to 20 second video generation.