Model
Prompt
Aspect ratio
Duration
Resolution
Public Visibility
iFree generations stay public unless you upgrade.
Flux 3 Free is free at 4 seconds and 480p. Free tasks are processed in order, with members first.
Flux 3 Video
Use a FLUX 3 text-to-video prompt to turn a shot brief into AI video. Define subject, action, camera, light, sound, and pace, then verify the actual selectable model and credits.
Move from prompt to clip with model selection, output controls, credit estimate, and creation history on the same page. Start with one clear shot before attempting a multi-event sequence.
Flux 3 Free
Veo 3.1 Lite
Veo 3.1 Basic
Veo 3.1 Premium
Seedance 2 Mini
Seedance 2
Seedance 2 Fast
HappyHorse 1.0
Seedance v1 Pro
Seedance 1.5 Pro
Kling 3 (Standard)
Kling 2.5
Text to Video generator
Use these FLUX 3 text-to-video prompt structures for ads, social hooks, establishing shots, and story beats. They teach prompt construction without inventing unavailable model features.
Prompt
Wide shot: a black dragon crosses beneath a cold full moon, storm clouds rotate behind it, lightning briefly reveals the wing texture, slow upward camera tilt.
Versatile Text to Video workflows
Give the model an ordered subject, action, setting, camera, lighting, and pace. Clear priorities are more useful than a long pile of style adjectives.
Define a subject doing a visible action in a specific place instead of combining unrelated ideas in one prompt.
Place subject, action, camera, light, and pace in a clear order so the central movement has priority.
Choose the available ratio and duration for a vertical hook, landscape ad, storyboard beat, or exploratory draft.
Keep successful directions and revise one failed action, camera move, or style constraint per generation.
FLUX 3 text-to-video prompt lab
Text-to-video works best when the prompt reads like a compact shot list rather than a mood board. Decide what the viewer should notice first, what visibly changes, and how the camera reveals that change. FLUX 3 Video has announced multimodal, native-audio, and multi-shot capabilities in Early Access, but this independent generator can also use other supported models. Always read the selected provider label before attributing a result to FLUX 3.
Write a concrete subject in a place with a readable starting state: a courier waits beside a rain-soaked tram, or a glass bottle stands under a narrow studio light. This gives the model an opening frame. Abstract goals such as premium, viral, or emotional can guide style later, but they do not tell a FLUX 3 text-to-video workflow what should appear.
State the main action first, then describe one reaction or consequence. The courier looks up as the tram arrives; condensation gathers while a reflection crosses the bottle. If five events compete in a short duration, none has enough screen time. A clear action hierarchy also makes it obvious which sentence to revise after the first AI video test.
Use the camera to expose information, not as a list of fashionable movements. A low tracking shot can reveal speed, a slow push can reveal expression, and a locked wide frame can reveal choreography. Include framing, direction, and pace once. Lighting and lens language should support that visual purpose rather than contradict the requested movement.
Save the prompt, model, ratio, duration, resolution, audio choice, and result before revising. Change only the failed variable: simplify the action, slow the camera, clarify the subject, or remove a conflicting style. This creates reusable FLUX 3 text-to-video prompt knowledge instead of producing unrelated clips that cannot be compared.
When the subject is wrong, move identity details to the opening sentence. When action order fails, cut secondary events and add timing words. When the camera ignores direction, remove lens and style clutter before repeating the path. If audio, continuity, or multi-shot behavior is missing, first confirm that the selected model supports it. An advertised FLUX 3 capability and the controls exposed by a different provider model are not interchangeable evidence.
Text to Video generator
For text to video, describe one filmable shot, select settings that fit the destination, and refine the first result by changing the least successful instruction.
Name the subject, action, location, camera move, lighting, visual style, and intended pace in a coherent scene description.
Choose the displayed model, ratio, duration, resolution, and audio option for a landscape ad, vertical post, or concept test.
Review whether subject, action, and camera were followed. Change one weak instruction at a time instead of replacing the entire prompt.
Questions
Include a visible subject, one main action, setting, camera behavior, lighting, and pace. Add style only after the scene and movement are unambiguous.
The generator can show more than one supported model. Check the selected provider model rather than assuming every task uses FLUX 3 Video.
The available controls change with model support. Use the options shown after model selection as the current source of truth.
The platform estimates credits before submission. Model, duration, resolution, audio, and plan discounts can change the amount shown.
A limited Flux 3 Free benefit or starter credits may be available to eligible accounts. Current limits are shown in the product and are not unlimited.
No. Use the exact provider model shown in the generator. A platform label is not proof of access to the official Black Forest Labs API.
Explore FLUX 3 Video
Move between the canonical FLUX 3 guide, image-to-video, text-to-video, and independent platform pricing without opening duplicate keyword pages.
Write a focused prompt, confirm the actual available model and estimated credit cost, and generate a short draft before expanding the idea.