1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
const axios = require('axios');
const api_key = "YOUR API-KEY";
const url = "https://api.segmind.com/v1/flux-3-text-to-video";
const data = {
"prompt": "Cinematic medium close-up of a young lighthouse keeper standing on a windswept cliff at golden hour, ocean waves crashing on rocks below. She turns to the camera and says warmly, \"The storm is finally passing.\" Soft golden rim light, shallow depth of field, sound of crashing waves, gusting wind and distant seagulls, subtle film grain.",
"duration": 5,
"resolution": "hd",
"aspect_ratio": "16:9",
"draft": false,
"generate_audio": true
};
(async function() {
try {
const response = await axios.post(url, data, { headers: { 'x-api-key': api_key } });
console.log(response.data);
} catch (error) {
console.error('Error:', error.response.data);
}
})();Attributes
Text describing scene, subject, camera moves, dialogue and sound design. Be specific and cinematic for best results.
Clip length in seconds, 5 to 20 (also accepts auto). Use 5 for quick tests, longer for storytelling.
min : 5,
max : 20
Output detail: hd or fhd. Use hd for fast iteration, fhd for final delivery.
Allowed values:
Frame shape: auto or a fixed ratio. Use 16:9 for cinematic, 9:16 for social verticals.
Allowed values:
Fast low-detail preview rendered at hd. Enable to iterate quickly, disable for final quality.
Generates synchronized speech, effects and ambience. Keep true for sound, false for a silent clip.
To keep track of your credit usage, you can inspect the response headers of each API call. The x-remaining-credits property will indicate the number of remaining credits in your account. Ensure you monitor this value to avoid any disruptions in your API usage.
FLUX 3 Text to Video — Text-to-Video AI Model
What is FLUX 3 Text to Video?
FLUX 3 Text to Video is Black Forest Labs' first video model, turning a single text prompt into a cinematic clip with native, synchronized audio. Instead of stitching sound on in a second pass, FLUX 3 is one multimodal model trained jointly on image, video, and audio, so dialogue, sound effects, and ambience are generated in the same pass as the picture. From plain text you get clips up to 20 seconds at HD or Full HD, 24 fps, across cinematic and social aspect ratios, no reference image or footage required.
Key Features
- Native synchronized audio: multilingual speech with strong lip-sync, plus sound effects and ambient sound.
- Long, single-pass generations up to 20 seconds in HD or Full HD.
- Multi-scene, multi-angle coherence, so shots hold together across cuts.
- Broad stylistic range: live-action realism, animation, motion design, and stylized looks.
- Accurate in-scene text and typography.
- Draft mode for fast previews before you commit to a full render.
Best Use Cases
Use FLUX 3 Text to Video for social ads, product teasers, cinematic B-roll, explainer clips, music-video moments, and dialogue scenes that need talking characters with lip-sync. In testing, a golden-hour coastal prompt with a short spoken line produced a photoreal character, a smooth push-in camera move, and layered wave, wind, and voice audio, reliably across repeated runs.
Prompt Tips and Output Quality
Write the shot like a director: name the subject, camera move, lighting, dialogue in quotes, and the sounds you want. Keep spoken lines short for 5-second clips, and pick 16:9 for cinematic framing or 9:16 for verticals. Turn on audio for talking scenes; enable draft to iterate quickly, then render the keeper at full quality.
FAQs
Does FLUX 3 Text to Video generate sound? Yes. Audio is on by default, including speech, effects, and ambience synced to the picture.
How long can clips be? Up to 20 seconds per generation, from 5 seconds up.
Does it do lip-sync and other languages? Yes, it renders multilingual speech with strong lip-sync.
What resolutions are supported? HD and Full HD, at 24 fps.
Do I need an input image? No. This is pure text-to-video; just describe the scene.
Can I preview cheaply first? Yes, use draft mode for a fast low-detail preview before a full render.
Popular Models
