ClipMode Animate Explained: MiniMax H3

What is Animate?

Animate turns stills of your character into a short AI video. ClipMode fills in likeness from identity photos, then MiniMax H3 generates motion, camera, and a soundtrack from your prompt.

You can open it from a model’s Create page, or tap Animate this on a still in the gallery to jump straight in with that image as a reference.

What is MiniMax H3?

MiniMax H3 is the video engine behind Animate. It is built for reference-to-video: you give it photos of a face, a prompt for motion, and it keeps that person recognizable across the clip.

H3 always produces audio. Sound on lets the model invent ambience or speech from your prompt. Sound off forces a silent track — use it when you will add your own music later.

How to use Animate

Using Animate only takes a few steps.

• 1

  Open Animate

  From a trained model, open Create and tap Animate (marked NEW). Flux and Classic both use the same MiniMax H3 engine.

• 2

  Check identity photos

  ClipMode pre-fills a starter set from your training and recent stills. Tap + to swap or upload — up to 9 photos. Clear, face-forward shots keep likeness consistent.

• 3

  Write a motion prompt

  Describe what happens, not just how they look. Camera, action, and audio work well: “She turns toward camera and smiles. Soft café ambience.”

• 4

  Add styles (optional)

  Open + LoRA / Style for curated MiniMax H3 looks — motion, camera, combat, detail. You can stack up to three. Skip them if a plain prompt is enough.

• 5

  Set duration, ratio, and sound

  Pick 5, 10, or 15 seconds, an aspect ratio, and Sound on or off. Longer clips cost more credits.

• 6

  Generate

  Tap Generate video. Clips can take a few minutes — keep the tab open, or leave and find the result in My Creations.

Identity photos

Animate treats the photo strip as likeness examples, not a storyboard. Front-facing, well-lit faces beat busy full-body shots. Mix a couple of angles if you have them; don’t dump nine near-duplicates.

You can pull from training images, recent generations, or uploads. If a clip drifts off-face, swap in a clearer still and generate again — the engine is only as consistent as the references.

Styles and LoRAs

H3 styles are optional LoRAs: small add-ons for motion, camera, or look. The gallery includes things like smoother body motion, combat, whispering close-ups, aerial flyovers, and eye/skin detail.

Keep Better Motion at low strength (around 0.2–0.4). Don’t stack three heavy styles unless you know they cooperate — they can fight each other. Paid accounts can also import MiniMax H3 LoRAs from Civitai or Hugging Face.

Duration, ratio, and credits

Clips are 5, 10, or 15 seconds. Portrait (9:16) is the default for social; landscape and square are in the ratio row. A 5-second clip starts at 20 credits; longer durations cost more.

Credit prices are for the current 480p Animate runs.

Why use Animate?

• Same face as your stills — no separate video training step
• Queue a clip from a gallery photo with Animate this
• Optional H3 styles for camera, motion, and look
• Built-in soundtrack, or silence if you will edit audio later

Final thoughts

Animate removes the usual video blank page: you already have a character. Give H3 a few clear identity photos, a motion prompt, and generate.

If you haven’t tried it yet, open a model and tap Animate. One short clip is enough to see whether the workflow fits.