Alibaba video model with built-in sound
Happy Horse 1.0 AI Video Generator
Happy Horse makes picture and sound in one pass, so you get finished clips: a felt balloon lifting off a quilt, a line spoken on a rainy platform, or a three-shot forge scene from one prompt.
Happy Horse is Alibaba's text and image to video model. On Pixel Dojo it makes clips of 3 to 15 s at up to 1080p and 24 fps, in 16:9, 9:16, 1:1, 4:3 or 3:4, and every clip comes with a stereo soundtrack. We made every example on this page with it, mostly on version 1.1, the newer engine at a lower rate per second, and each one comes with its exact prompt so you can start from it.
20 credits ($1.25) per 5 s clip at 720p. Included in every plan, from $10/month.
Already a member? Sign in
By Brian Lovett · Updated
Happy Horse 1.1 vs 1.0 vs WAN 2.7: specs and pricing
Both Happy Horse versions make 24 fps clips with stereo sound, 3 to 15 seconds long. Here is how they line up with WAN 2.7, Alibaba's other video model, on a 5 second dialogue clip at 1080p.
| Spec | Happy Horse 1.1 | Happy Horse 1.0 | WAN 2.7 |
|---|---|---|---|
| Rate per second | 3 credits at 720p, 4 at 1080p | 4 credits at 720p, 6 at 1080p | 2 credits at 720p, 3 at 1080p |
| Output | 1920 by 1080, 24 fps, stereo sound | 1920 by 1080, 24 fps, stereo sound | 1920 by 1080, 30 fps, stereo sound |
| Clip lengths | 3 to 15 s | 3 to 15 s | 2 to 15 s |
| Credits for a 5 s 1080p clip | 20 | 30 | 15 |
What you can create with Happy Horse
We made these clips on Pixel Dojo with Happy Horse 1.1, sound on: a stop motion felt short, sparks flying off an anvil, a flamenco performance in vertical 9:16 and a sunrise flight over rice terraces. Tap Use this prompt on any tile, then set Model to Happy Horse 1.1 and the length and resolution shown on the tile to make your own version.
Stop motion charm from a single text prompt. A knitted balloon in red and mustard stripes tips upright and floats off a patchwork quilt past felt sheep and button trees, every stitch of yarn texture in view. Start from this prompt for animated shorts and storybook openers.
Stop-motion felt animation, gentle rising crane shot: a small knitted hot air balloon in red and mustard stripes lifts off a patchwork quilt landscape of stitched green fields, woolly felt sheep and button trees, its tiny burner puffing a cotton-wool flame, visible yarn texture and a slightly stepped handmade frame rate, warm window light, tilt-shift miniature look. Sound: a soft burner whoosh, a gentle breeze, a distant sheep bleat, a light plucked ukulele.
Heat and sparks in extreme close-up. A leather glove grips the tongs as the hammer lands on a glowing orange horseshoe and throws a burst of white sparks across the frame. Start from this prompt for craft, maker and action openers.
Extreme close-up, handheld, locked on the anvil: a heavy hammer strikes a glowing orange horseshoe held in iron tongs and a fan of white-hot sparks sprays toward the lens, a roaring coal forge and drifting smoke behind, a scarred leather glove gripping the tongs. 50mm lens, deep amber and charcoal grade, embers floating, fine film grain. Sound: ringing hammer strikes on hot steel, the roar of the forge, crackling sparks.
A vertical performance piece for social. Under a single hard spotlight the dancer raises her arms, then spins into a full emerald skirt flare as dust rises, scored from the Sound line in the prompt. Start from this prompt for dance, fashion and music clips.
Static wide shot, low angle: a flamenco dancer in a long-sleeved black bodice and a flared emerald green ruffled skirt stamps her heels on a dusty wooden stage and whips the skirt into a full spinning arc, dust rising through a single hard white spotlight, pure black background, dark hair in a tight bun with a red flower. 35mm lens, high contrast, deep emerald and black grade. Sound: sharp heel stamps in rhythm, crisp hand claps, a fast nylon-string guitar.
Sweeping aerial travel footage at sunrise. The camera glides low over water-filled rice terraces as golden light rakes across the paddies and mist drifts toward a red-roofed farmhouse on the ridge. Start from this prompt for travel and nature openers.
Aerial drone shot gliding low over emerald rice terraces carved into a steep valley at sunrise, then rising to reveal mist rolling between the ridges and a lone red-roofed farmhouse, golden light raking across the water in each paddy. 24mm lens, crisp detail, warm gold and jade grade. Sound: wind rush, distant birdsong, trickling water.
Where Happy Horse shines: dialogue, sound and multi-shot scenes
Two more clips we made on Pixel Dojo: the rainy platform line on version 1.0, and a three-shot forge scene on 1.1 from one prompt with numbered shots. Here is what they show and how to get results like them.
Happy Horse 1.0 brings drama to dialogue. She delivers the line in a tight close-up through rain and steam, with a wet black bob and a warm film grade as the camera pushes in.
Slow push in to a medium close-up on a rainy night train platform: a woman in her thirties with a wet black bob and a camel trench coat stands in drifting steam under a flickering sodium lamp, rain falling behind her, she looks at someone just off camera and says, "Don't wait for me. I'm not coming back." 85mm lens, shallow depth of field, teal and amber grade, fine film grain. Sound: rain drumming on the platform roof, a train hissing, a distant departure chime. No subtitles, no on-screen text.
Three shots from one prompt, 10 s at 1080p on 1.1. It cuts from the hammer strike to the farrier lit by the forge to the steam of the quench in the order we wrote them, with a soundtrack scored from the Sound line in the prompt.
Three-shot sequence in a dim stone smithy. Shot 1: extreme close-up on the anvil, a heavy hammer strikes a glowing orange horseshoe held in iron tongs and a fan of white-hot sparks sprays toward the lens. Shot 2: medium shot, a bearded farrier in his fifties wearing a scarred leather apron looks up from the anvil, his face lit orange by a roaring coal forge, smoke curling past him. Shot 3: close-up, the tongs plunge the horseshoe into a wooden bucket of water and a thick cloud of steam erupts. 50mm lens, deep amber and charcoal grade, embers floating, fine film grain. Sound: ringing hammer strikes on hot steel, the roar of the forge, a sharp hiss of steam.
- Every clip arrives with a stereo soundtrack made with the picture. Add a Sound line to steer it: rain on a roof, hammer strikes on hot steel, a fast nylon-string guitar.
- Put dialogue in quotes and end with "No subtitles, no on-screen text." Both platform clips deliver the full line on camera in a clean, caption-free frame.
- Number your shots for a scene. Our three-shot forge prompt cut exactly where we wrote Shot 1, Shot 2 and Shot 3, all inside 10 s.
- Name where the camera should stop. "Slow push in to a medium close-up" is exactly the framing 1.1 gave us on the platform scene.
- Material words carry the look. Yarn texture, felt sheep, button trees and a stepped handmade frame rate are what give the balloon clip its stop motion feel.
- At 1080p you get 1920 by 1080 in 16:9, 1080 by 1920 in 9:16 and 1440 by 1440 in 1:1, all at 24 fps, ready for film, vertical social or a square feed.
Happy Horse 1.1 and WAN 2.7 on the same spoken line
We gave Happy Horse 1.1 and WAN 2.7, Alibaba's other video model, the same dialogue prompt at 5 s, 1080p and 16:9. Both speak the line on camera with sound under it. Happy Horse 1.1 holds the camera move exactly as written at a filmic 24 fps, and WAN 2.7 is the lower price per clip.
Same prompt: Slow push in to a medium close-up on a rainy night train platform: a woman in her thirties with a wet black bob and a camel trench coat stands in drifting steam under a flickering sodium lamp, rain falling behind her, she looks at someone just off camera and says, "Don't wait for me. I'm not coming back." 85mm lens, shallow depth of field, teal and amber grade, fine film grain. Sound: rain drumming on the platform roof, a train hissing, a distant departure chime. No subtitles, no on-screen text.
Happy Horse 1.1. The camera pushes in and settles on a medium close-up exactly as written, and she delivers the full line through the rain and steam.
Slow push in to a medium close-up on a rainy night train platform: a woman in her thirties with a wet black bob and a camel trench coat stands in drifting steam under a flickering sodium lamp, rain falling behind her, she looks at someone just off camera and says, "Don't wait for me. I'm not coming back." 85mm lens, shallow depth of field, teal and amber grade, fine film grain. Sound: rain drumming on the platform roof, a train hissing, a distant departure chime. No subtitles, no on-screen text.
WAN 2.7 at 30 fps. It frames her over the shoulder of a figure in the foreground as she speaks the line under yellow lamplight.
Slow push in to a medium close-up on a rainy night train platform: a woman in her thirties with a wet black bob and a camel trench coat stands in drifting steam under a flickering sodium lamp, rain falling behind her, she looks at someone just off camera and says, "Don't wait for me. I'm not coming back." 85mm lens, shallow depth of field, teal and amber grade, fine film grain. Sound: rain drumming on the platform roof, a train hissing, a distant departure chime. No subtitles, no on-screen text.
| Happy Horse 1.0 | WAN 2.7 | |
|---|---|---|
| Camera | Slow push in that settles on a medium close-up, as written | Over-the-shoulder push in past a figure in the foreground |
| Frame rate | 24 fps | 30 fps |
| Credits | 20 | 15 |
Happy Horse settings for text and image to video
The Happy Horse tool has five main controls and a More options panel. Here is what each one does and how we set it for the clips above.
- ModelHappy Horse 1.0 (the default) or Happy Horse 1.1
- Choose 1.1 to match the showcase. It is the newer engine at a lower rate: our 5 s dialogue clip at 1080p was 20 credits on 1.1 and 30 on 1.0, and both spoke the line on camera.
- ModeText to Video, Image to Video or Reference to Video
- In Image to Video your still becomes the first frame, so describe only the motion and start from a clean image, since everything in it carries into the clip. Reference to Video takes up to 9 reference images, and the prompting guide shows how to use them.
- Resolution720p (the default) or 1080p
- Use 1080p for finished pieces like the balloon, flamenco and forge scenes above. 720p is the lower price per second and suits short social clips like the anvil sparks and the terrace flight, at 9 credits for 3 s on 1.1.
- Aspect ratio16:9, 9:16, 1:1, 4:3 or 3:4
- Frame for where the clip will live. Our 9:16 flamenco came back 1080 by 1920 for vertical social, and 1:1 at 1080p comes back 1440 by 1440. In Image to Video the clip follows the shape of your image.
- Duration3, 5, 10 or 15 s
- Give a camera move or a big action 5 s to play out, like the balloon lift off and the flamenco spin. For a scene, give each numbered shot about 3 s: our three-shot forge scene fits in 10 s.
- Things to avoid (More options)Your own text
- This is the negative prompt, and the tool pre-fills a default avoid list. The clips on this page ran without one, so clear it to match them exactly. For dialogue, write no subtitles into the main prompt, as we did.
- Seed (More options)Random or a fixed number
- Lock a seed when you love a clip and want to change only a word or two. Every clip here ran on a random seed.
How to make a Happy Horse video with sound
- 1
Set Model to Happy Horse 1.1
The tool opens on 1.0. Switch to 1.1 for the lower rate per second, then pick your resolution, aspect ratio and length.
- 2
Write it like a shot list
Camera first, then the subject, one action, light, lens, and a Sound line. Happy Horse makes sound for every clip, so tell it what to hear. Every prompt on this page follows that order, so copy one and swap in your subject.
- 3
Give your character a line
Put the words in quotes and add "No subtitles, no on-screen text." Say where the camera should stop, like "slow push in to a medium close-up."
- 4
Cut a scene with numbered shots
Write Shot 1, Shot 2 and Shot 3, each with its own framing and action, and set 10 s or more. Our forge scene cut at every shot we wrote, for 40 credits at 1080p.
Happy Horse pricing on Pixel Dojo
Happy Horse is included in every Pixel Dojo plan, from $10/month, and each clip uses credits from your plan. The tool shows the exact price for your settings before you generate.
Happy Horse prices by the second. Version 1.1 is 3 credits a second at 720p and 4 at 1080p. Version 1.0 is 4 credits a second at 720p and 6 at 1080p.
A 3 second clip at 720p on 1.1 is 9 credits, a 5 second clip at 1080p is 20 credits, and the 10 second three-shot forge scene on this page is 40 credits.
Happy Horse 1.0 questions
What is Happy Horse AI?
Happy Horse is Alibaba's AI video generator. It turns a text prompt or a still image into a clip of 3 to 15 s at up to 1080p, with a stereo soundtrack made in the same pass. On Pixel Dojo you can run version 1.0 or the newer 1.1 from one tool, or through the Pixel Dojo API.
What is Happy Horse best at?
Video with sound and rich texture. It adds a soundtrack to every clip, speaks lines you put in quotes on camera, cuts numbered shots right where you write them, and renders tactile looks like stop motion felt and yarn, sparks off an anvil and dust in a spotlight.
What is the difference between Happy Horse 1.0 and 1.1?
Version 1.1 is the newer engine at a lower rate per second: 3 credits at 720p and 4 at 1080p, against 4 and 6 on 1.0. On our 5 s dialogue prompt at 1080p, 1.1 was 20 credits and 1.0 was 30, and both spoke the line on camera.
Does Happy Horse make sound and dialogue?
Yes. Every clip comes with a stereo soundtrack made in the same pass as the picture. Put a spoken line in quotes and your character says it on camera, like the rainy platform scene on this page, and add a Sound line for rain, footsteps or music.
How long can a Happy Horse video be, and at what resolution?
Clips run 3 to 15 s at 720p or 1080p and 24 fps, in 16:9, 9:16, 1:1, 4:3 or 3:4. A 1080p clip comes back 1920 by 1080 in 16:9, 1080 by 1920 in 9:16 and 1440 by 1440 in 1:1.
Happy Horse vs WAN 2.7: which should I use?
Both are Alibaba video models with sound, and both speak a quoted line on camera. On our shared dialogue prompt, Happy Horse 1.1 held the slow push in to a medium close-up exactly as written, at a filmic 24 fps, for 20 credits. WAN 2.7 runs at 30 fps and costs 15 credits for the same clip. Pick Happy Horse when the camera move and a cinematic look matter, and WAN 2.7 when you want the lowest price per clip.
How much does Happy Horse cost?
Happy Horse is included in every Pixel Dojo plan, from $10/month. On version 1.1 a 3 s clip at 720p is 9 credits and a 5 s clip at 1080p is 20 credits, and the tool shows the price for your settings before you generate. The public API runs it with the same per-second pricing; send version 1.1 for the lower rate.
Make your first Happy Horse 1.0 clip
Write a prompt above or start from any example on this page. One subscription covers every model on Pixel Dojo.
Start with Happy Horse 1.0