HappyHorse 1.1

For HappyHorse 1.1, splitting the prompt into labeled parts for subject, action, setting, camera, lighting, and mood keeps the instructions from blending into each other. When you have little to fill in, a single short paragraph may work better.

Split the prompt into labeled parts

Coming soon

Giving six parts their own headings keeps action, lighting, and camera instructions from blending into each other, even though the prompt ends up about as long as the same scene written as prose.

Paste the template below into the video node's prompt, keep part headings such as ## Subject as they are, and replace the bracketed text with details that fit your scene.

## Subject
[Subject and wardrobe]

## Action
[What happens, including the direction of movement]

## Setting
[Location, time of day, weather, props]

## Camera
[Camera movement, lens, depth of field, framing]

## Lighting
[Key light direction, color temperature, contrast]

## Mood
[Two or three mood words]
PartWhat to write
## SubjectThe subject and wardrobe
## ActionWhat happens, including the direction of movement
## SettingLocation, time of day, weather, props
## CameraCamera movement, lens, depth of field, framing
## LightingKey light direction, color temperature, contrast
## MoodTwo or three mood words

You can check the supported inputs, length, ratio, and resolution in Video models.

What to watch for when writing

Leaving parts empty or repeating the same instruction in several parts can make the result worse.

  • This structure works best when most parts have specific details to fill in. Empty parts can lower the quality of the result.
  • If only a few parts have anything to fill in, a single paragraph of around 20 words may work better.
  • Write each instruction in one part only. Asking for a static camera in both Action and Camera does not strengthen the instruction; the two can cancel each other out.
  • Mood is for atmosphere, not for raising quality. Adding words such as cinematic or masterpiece can actually lower quality.