Prompt Happy Horse for image-to-video

Last updated: October 12, 2026

Happy Horse animates an approved still into a short clip with sound from a prompt of about twenty words. Pick it when the look is signed off and you want on-model motion, such as an idle or a spell charge, without writing a long shot description.

Let the agent write it

You don't have to write any of this yourself. The Layer agent knows how to prompt each model it runs, from the same guidance this page is built on. Attach the still in chat, name the model and say what should happen, for example: Animate this with Happy Horse 1.1: the mage charges a fireball on his staff, camera locked, 3 seconds, with fire and wind sounds. The agent writes the prompt, shows you its plan, and runs it when you approve. Use the rest of this guide when you want to write or adjust the prompt yourself, and see Write a brief for the agent for how to brief it.

When to use it

Use Happy Horse when you have the frame and need it to move: character idles, ability charges, cloth and fire in the wind, and short shots for trailers and social posts. It keeps the character's design and the frame's style, and generates sound in the same pass.

It loses the frame on fast, explosive action in one take (leaps, spins, slams). Cut that into separate clips, or see Prompt Seedance for image-to-video for a model you can steer with a longer, step-by-step prompt. To generate a clip with no image at all, see Create a video.

Versions on Layer

Happy Horse 1.1

The version this guide's examples were made with. It animates a first frame into a clip of 3 to 15 seconds, at 720p or 1080p, with sound.

Happy Horse 1.0

Takes the same prompt and the same first frame. In our test it rendered the effect more softly: where 1.1 grew the ember into a billowing fireball, 1.0 made a contained glowing orb. It also takes Reference image files and a Source video to edit, which 1.1 does not. Keep one asset set on one version, since the same prompt reads differently on each.

How to write the prompt

The action, then one camera cue, then the sound:

The mage slowly raises his staff and the ember swells into a fireball, his cloak stirring. Locked-off camera. Sound: crackling fire, low wind.

The first frame already fixes who is in the shot, where, and how it looks, so a short prompt only needs to say what moves and what the camera does. That prompt is the one in the clip above.

What works

A short prompt

Aim for about twenty words. We ran the prompt above against a 110-word version on the same frame and seed, which restated the frame ("a young fire mage in a deep red hooded cloak with gold trim stands on a mossy stone ledge at dusk...") and ended with "beautiful, stunning, epic, highly detailed, cinematic masterpiece". Both kept the mage on-model. But every extra detail turned up on screen: swirling leaves, speed streaks, and a cloud of smoke that hid the castle by the end. The short prompt gave the cleaner clip.

Slow, continuous motion

Write motion that stays in the shot: "slowly raises his staff", "his cloak stirring", "sparks drifting up". We ran "the mage leaps high into the air, spins and slams his staff into the ground" on the same frame and seed. The camera followed him into the sky, a ring of fire filled the frame, and the clip ended in a full-frame explosion: the mage was still recognisable, but nothing of the approved composition was left.

The camera cue last

End the motion with one camera cue: locked-off camera, slow dolly-in, tracking shot, aerial. Use one cue per shot.

Concrete words

Leave out hedging adjectives such as "beautiful", "stunning" and "masterpiece", and bare director names. Describe the light and the look instead: "warm amber backlight", "wet stone".

Plain sentences

Write the prompt as ordinary English sentences. Tag lists, JSON and weighted parentheses do worse than the same content written as prose.

Several beats in one clip

In plain prose, a sequence of steps collapses into one motion. For several beats, write a shot list with times: "Shot 1 (wide, 0-3s): ... Shot 2 (close-up, 3-6s): ...", or storyboard cues: "Begin with ..., Cut to ..., End on ...". Give each beat time: two beats fit in 5 to 7 seconds, three in 8 to 10. Describe the character once at the start, and close with a line that keeps their look: "Keep his exact look, red hooded cloak and carved staff, across every shot."

Sound in the prompt

Both versions make sound with the picture. Name what you want to hear after the action ("Sound: crackling fire, low wind") and put any spoken line in quotes.

What to avoid

Restating the frame

Describing the character, the setting and the style again adds nothing the frame does not already set, and the model tends to add whatever extra detail you name. Spend the words on motion.

Negative cues by default

There is no negative prompt for image-to-video, and most "no ..." phrases are wasted words. Add one only for a concrete risk you have seen in a run.

A frame in the wrong shape

The clip keeps the shape of the image; Aspect ratio does not change it. Crop the image to the shape you need before you start. The image must be at least 300 pixels on each side, with its long side no more than 2.5 times its short side.

Settings that matter

First frame holds the image the clip starts from. Duration runs from 3 to 15 seconds; the examples here are 3 seconds. Aspect ratio picks 720p or 1080p; the clip's shape follows the image. Seed, among the advanced settings, makes a run easier to repeat, though the same seed does not always give the same clip. Above the prompt, Camera, Staging, Look and Performance take reference sets for each.

Prompts to start from

Spell charge

The mage slowly raises his staff and the ember swells into a fireball, his cloak stirring. Locked-off camera. Sound: crackling fire, low wind.

Character idle

The mage breathes slowly, his cloak and hood stirring in the breeze, sparks drifting up from the ember. Locked-off camera. Sound: soft wind, crackling ember.

Use it in Layer

  1. Add the approved still to a session's board.

  2. Click View on the image, then Turn into video. Create video opens in the side panel with the image as its First frame.

  3. Click the model name at the top of the panel and choose Happy Horse 1.1. The first frame stays in place.

  4. In the prompt box, write the action, one camera cue and the sound, in about twenty words.

  5. Click Create video. The cost shows on the button before you click.

  6. When the clip lands on the board, click View to open it in the viewer and play it. Then change one thing in the prompt at a time and run it again.

Learn more

Alibaba publishes API references for Happy Horse but no prompt guide of its own, so the prompting advice here comes from the platforms that host the model and from our own runs.

Related