Who is FLUX 3 Max Image-to-Video for?

FLUX 3 Max Image-to-Video is for creators who want to develop still visuals into short video scenes. A starting image establishes the opening frame, while the prompt describes how the scene unfolds. On each::labs, the model supports creative clips and visual storytelling with adjustable duration, aspect ratio, and optional audio.

How do starting and ending images work in FLUX 3 Max?

FLUX 3 Max requires a starting image and also accepts an optional ending image to guide the final frame. Without an ending image, the video develops from the starting image and prompt. Users can leave the aspect ratio on auto to follow the first image or select a supported ratio for their output.

Example inputhover

prompt: "Authentic UGC beauty video filmed on a phone propped on a vanity, talking-head shot. She looks straight into the camera and speaks clearly and warmly, her lips, jaw and mouth visibly moving and perfectly lip-synced to every word as she says: "Okay, I'm obsessed with these pearl hair pins. It took me five minutes, and honestly it looks so elegant. Perfect for any event!" She keeps her face toward the camera the whole time she is speaking, with natural expressions, eyebrow raises and a light touch of her hair. Only after she finishes the last word, she briefly turns her head to the side to show the pearls scattered through her long wavy hair, then turns back with a big smile. Natural blinking, the pearls stay in place. Casual, honest social media creator vibe. Sound: her clear natural voice, close to the phone mic, quiet room tone."
duration: 8
image_url
resolution: "1080p"
aspect_ratio: "16:9"
generate_audio: true
reasoning_level: "high"

FLUX 3 Max Image to Video

Image Gen·flux-3·by Black Forest Labs

FLUX 3 Max Image-to-Video turns still images into 5–20-second videos with optional audio and end frames, for creative clips and visual storytelling.

Try it now →

API reference

Runtime (p50): -
Estimated price: $0.0425–$0.0725 per second, depending on resolution

Call the API

prediction.sh

curl -X POST \
  -H "Authorization: Bearer $EACHLABS_API_KEY" \
  -H "Content-Type: application/json" \
  --data '{
    "model": "flux-3-max-image-to-video",
    "input": {
        "prompt": "Authentic UGC beauty video filmed on a phone propped on a vanity, talking-head shot. She looks straight into the camera and speaks clearly and warmly, her lips, jaw and mouth visibly moving and perfectly lip-synced to every word as she says: \"Okay, I'm obsessed with these pearl hair pins. It took me five minutes, and honestly it looks so elegant. Perfect for any event!\" She keeps her face toward the camera the whole time she is speaking, with natural expressions, eyebrow raises and a light touch of her hair. Only after she finishes the last word, she briefly turns her head to the side to show the pearls scattered through her long wavy hair, then turns back with a big smile. Natural blinking, the pearls stay in place. Casual, honest social media creator vibe. Sound: her clear natural voice, close to the phone mic, quiet room tone.",
        "duration": 8,
        "image_url": "https://cdn-us.eachlabs.ai/defaults/d4626a36677c4d63a4e2ec0d48c93488.png",
        "resolution": "1080p",
        "aspect_ratio": "16:9",
        "generate_audio": true,
        "reasoning_level": "high"
    },
    "webhook_url": ""
}' \
  https://api.eachlabs.ai/v1/prediction/

Related models

4 models

Hearthbroken Car VideoUnified AI API for hundreds models. Access video, image, and audio generation from leading providers through one integration with eachlabs.

Grok Imagine Video 1.5 Lite Image-to-Video AI model preview

Grok Imagine Video 1.5 Lite Image-to-VideoxAI

BytePlus OmniHuman 2.0Bytedance

Storm Trend EffectUnified AI API for hundreds models. Access video, image, and audio generation from leading providers through one integration with eachlabs.

* FAQ

About FLUX 3 Max Image to Video

01 / 03

What is FLUX 3 Max Image-to-Video?

FLUX 3 Max Image-to-Video generates a video from a starting image and a text prompt describing the action. It accepts PNG, JPEG, or WebP images and can generate clips from 5 to 20 seconds. Creators can choose 720p or 1080p output and decide whether to include generated audio.