Make a short video from words or animate one starting image. Describe the action, camera movement, and sound, then choose the length and resolution for your next shot.
FLUX 3 is a video model from Black Forest Labs. This page focuses on short videos with generated audio, using either a written scene or one starting image. It is a separate workflow from FLUX image generation. Begin with what the viewer should see happen during the clip: an object moving, a camera revealing a space, or a small change in the environment. A clear sequence gives you something concrete to evaluate when the result arrives.
Choose an action that fits the selected duration. A hanging mobile turning slowly, a droplet falling into water, or a fabric fold settling on a surface can each create a complete short shot. Explain where the movement begins and how it should develop. Avoid requesting several scene changes when a single continuous view would serve the idea. Watch the result for unexpected pauses, repeated movements, or an ending that interrupts the action before it becomes understandable.
State whether the camera stays still, moves toward the subject, or travels across the scene. A small camera move can reveal shape and material without requiring the subject to change position. Give the model a clear direction and pace, then check the result for a consistent viewpoint. Background lines, edges, and reflections can distort during movement. If the shot must match another clip, compare the horizon, framing, and direction of travel before placing them together in an edit.
Light can make a simple scene readable. Describe a soft window reflection on glass, warm side light on folded fabric, or a diffuse shadow beneath a paper object. Keep the lighting compatible with the location and materials. After generation, check how highlights and shadows behave as the scene moves. An appealing still frame can hide reflections that slide incorrectly or surfaces that change texture. Judge the whole clip, especially where thin, transparent, or reflective objects overlap.
Describe the sounds that belong to the action, such as a gentle chime, a small splash, or quiet room ambience. Specify whether speech or music is wanted when that distinction matters. The soundtrack is generated and needs review alongside the picture. Listen for unwanted voices, abrupt changes, and sound effects that do not align with visible events. For exact dialogue, music, or timing, plan to replace or edit the audio rather than treating the first result as a finished soundtrack.
Image-to-video uses one starting picture to establish the subject and composition. The prompt describes how the shot should move from that point. This can be useful for a product concept, an illustration, or a scene whose opening has already been chosen. The source does not guarantee perfect preservation through time. Keep it available while reviewing the output, and compare the features that define the subject instead of relying only on overall color or atmosphere.
Use a source that shows the important outlines and materials, with enough space around the subject for the requested movement. A close crop can force the model to invent parts that become visible later. Check the image for small lettering, ambiguous overlaps, and hidden attachments. These may change as the video develops. Upload material you have permission to use, and remove unrelated personal information before starting. A clean source also makes the final comparison easier.
Describe motion that works with the existing picture. A gentle push toward a still object, a small sway in fabric, or subtle environmental movement can be a focused starting point. Avoid simultaneously changing the subject, setting, lighting, and camera if your priority is resemblance. Name the details that should stay recognizable, then verify them afterward. Written instructions do not lock pixels or ensure that a label, handle, pattern, or facial feature remains identical in every frame.
The standard workflow offers explicit clip lengths from 5 to 20 seconds, with 720p or 1080p output. Choose enough time for the action and enough detail for your next editing step. Duration and resolution affect the estimate, so review it after changing either setting. A longer clip also creates more opportunity for continuity problems to appear. Increasing resolution does not fix an incorrect movement, a changing object, or a soundtrack that conflicts with the scene.
The last second matters when the video will be trimmed, looped, or joined to another shot. Check whether the subject remains recognizable, the camera settles naturally, and the action reaches a useful stopping point. This workflow does not use a supplied ending frame or provide a video continuation tool. If you need an exact cut point, download the clip and make the final trim in your editor. Keep the unedited original so later decisions remain reversible.
Treat each clip as a generated interpretation of the brief.
Look for objects changing shape, duplicate elements, abrupt cuts, and camera moves that reveal inconsistent details. Listen when sound is included, then inspect the file in the tool where you will edit or publish it. Keep an untouched download before making changes.
Preparing a prompt does not start a paid request; generating a new clip has its own cost, which you should review for the selected settings. Follow the current task in the workspace instead of creating another request while waiting. Decide what needs changing before paying for a new attempt.
Write a new scene or upload a clear starting picture. Decide which action and subject details will determine success.
Explain the action, camera direction, pace, and relevant audio. Keep the scene coherent within the chosen length.
Choose an explicit duration and either 720p or 1080p. Check the estimate before starting the request.
Review continuity from opening to ending, inspect the soundtrack, and make final edits in the next tool.
Plans for image and video
Annual plans are billed upfront. The same plan credits refresh every 30 days and expire at the end of that window; unused subscription credits do not roll over.
Explore the workspace and prepare a first request before choosing a paid plan.
For occasional image and video experiments with a predictable monthly ceiling.
Subscriptions renew automatically at the selected monthly or yearly rate until canceled. Cancel future renewals in your account.
For creators who iterate every week and need room for several directions.
Subscriptions renew automatically at the selected monthly or yearly rate until canceled. Cancel future renewals in your account.
For sustained, high-volume creation and access to the highest available output quality.
Subscriptions renew automatically at the selected monthly or yearly rate until canceled. Cancel future renewals in your account.
Prices are in USD. Subscriptions renew automatically; credit packs are one-time purchases.
Compare plansNo. This page uses FLUX 3 for video generation. If you need a still image, use the image workspace and choose an image model appropriate to that task.
Yes. Text-to-video begins with a written scene description. Image-to-video uses one starting picture and instructions for how the shot should develop.
The standard workflow provides 5–20 seconds at 720p or 1080p. Choose the duration explicitly and review the estimate for that combination before generating.
This workflow includes generated audio. Describe relevant sound in the prompt and listen to the returned clip. Replace or edit the soundtrack when exact wording, music, or synchronization is required.
This page creates a new clip from text or one image. It does not provide video editing, continuation, or a multi-keyframe workflow. Use an editor for final assembly and trimming.
Choose one scene, describe how it moves, and decide what should be heard. Review the settings and estimate, then evaluate the finished clip against that brief.
