Why does the scaffold leave out parameters I have seen elsewhere?

Because it is built from that specific model's published schema rather than a generic template. If a model does not expose an aspect ratio or a motion level, none appears, since sending a parameter the endpoint does not accept is how a request gets rejected. Two models from the same provider often differ here.

Where do the prompting rules come from?

They are the platform's own per-family prompting rules, carried through verbatim rather than rewritten. They are grouped by model family because families share a response to prompt structure — what works on one member usually works on the others.

Why do some models show guidance but no parameters?

Not every model in the catalog publishes a parameter schema. Where one is missing, the prompting guidance and the accepted-input summary still apply, and the model's own page carries whatever else is documented.

Does a longer prompt work better?

It depends on the family, which is why the rules are shown per model rather than as one universal list. Some families reward dense, specific description; others resolve contradictions in a long prompt unpredictably and do better with a short, unambiguous one. Writing for the wrong family is a common reason a prompt appears to be ignored.

AI prompt builder

Most prompt guides are written for a model in general. This one is written from each model's own parameter schema — 83 of 148 models publish one — so the request it builds uses the parameters that endpoint actually accepts, with the enums it actually allows.

Model

What do you want to make?

Request scaffold

Built from this model's own schema — 1 provider variant declared. Parameters it does not expose are deliberately absent; sending one would be rejected.

{
  "model": "happy-horse-1-0-text-to-video",
  "prompt": "<your prompt>"
}

Required (1)

promptstring
Text prompt (max 2500 chars)

Optional (3)

aspect_ratiostring
Aspect ratio
16:9 · 9:16 · 1:1 · 4:3 · 3:4
durationinteger
Seconds (integer enum 3-15)
3 · 4 · 5 · 6 · 7 · 8 · 9 · 10 · +5
resolutionstring
Resolution
720p · 1080p

How this model likes to be prompted

Use vivid, specific descriptions with details about composition, lighting, style, and atmosphere.

What the catalog says it accepts

Durations: 3s, 4s, 5s, 6s, 7s, 8s, 9s, 10s, 11s, 12s, 13s, 14s, 15s
Qualities: 720p, 1080p
Aspect ratios: 16:9, 9:16, 1:1, 4:3, 3:4
Needs an image: No
Returns audio: No

Happy Horse 1.0 Text to Video model page Prompting guides What this will cost

Model finder — pick the model first, by what you can hand it.
Credit calculator — what the request will cost before you send it.
Prompting guides — the longer-form version, per model.

FAQ

Why does the scaffold leave out parameters I have seen elsewhere?: Because it is built from that specific model's published schema rather than a generic template. If a model does not expose an aspect ratio or a motion level, none appears, since sending a parameter the endpoint does not accept is how a request gets rejected. Two models from the same provider often differ here.
Where do the prompting rules come from?: They are the platform's own per-family prompting rules, carried through verbatim rather than rewritten. They are grouped by model family because families share a response to prompt structure — what works on one member usually works on the others.
Why do some models show guidance but no parameters?: Not every model in the catalog publishes a parameter schema. Where one is missing, the prompting guidance and the accepted-input summary still apply, and the model's own page carries whatever else is documented.
Does a longer prompt work better?: It depends on the family, which is why the rules are shown per model rather than as one universal list. Some families reward dense, specific description; others resolve contradictions in a long prompt unpredictably and do better with a short, unambiguous one. Writing for the wrong family is a common reason a prompt appears to be ignored.