AI Fashion Video API

A short fashion video from a prompt alone, from a start image you animate — with an optional end image the clip lands on — or from up to three reference pictures (a person, a garment, a place) put into a scene you describe. Billing is per second.

Get an API key Claude MCP

POST/v1/generateworkflow: freeform_video-v1

One endpoint serves every workflow — this page documents freeform_video-v1, the current version. Credits are debited at submission (2.2 cr/s standard, 4.4 cr/s pro, 23 cr/s ultra); a failed generation is refunded automatically. Earlier versions of this endpoint, if any, keep being served under their own names.

Parameter	Value for this workflow
workflow	"freeform_video-v1" required
prompt	Describe the creation or the change — English works best.
start_image	Start frame — one public HTTPS URL.
fields.audio	Audio. The sound of the clip — music, voices, ambience. Leave the default for the natural sounds of the scene.Defaults to “No music, natural environment sounds.”.
duration	5 · 10 · 15 (seconds)
end_image	Optional closing frame — one public HTTPS URL.
references	Optional, one to three public HTTPS URLs — a person, a garment, a place — put into a scene the prompt describes (name them Image 1, Image 2, Image 3). Replaces the start and end frames; never sent with them.
tier	"standard" (2.2 cr/s), "pro" (4.4 cr/s) or "ultra" (23 cr/s) — 1080p on our top video model
webhook_url	Optional — we POST the outcome here when the generation settles. Without it, poll the URL returned in the response until the status settles.

Example request

curl -X POST https://thenewblack.ai/api/v1/generate \
  -H "Authorization: Bearer tnb_live_XXXXXXXXXXXX" \
  -H "Content-Type: application/json" \
  -d '{
    "workflow": "freeform_video-v1",
    "prompt": "The model walks toward the camera, fabric flowing",
    "start_image": "https://example.com/start-frame.jpg",
    "fields": {
      "audio": "No music, natural environment sounds."
    },
    "duration": "5",
    "tier": "standard"
  }'

// 202 Accepted
{
  "generation_id": "3f1c9a6e-...",
  "status": "running",
  "poll": "/api/v1/generations/3f1c9a6e-..."
}

Example response

Poll /v1/generations/{id} (or receive the webhook) until the generation settles:

// 200 OK
{
  "generation_id": "3f1c9a6e-...",
  "workflow": "freeform_video",
  "status": "succeeded",
  "result": {
    "type": "video",
    "url": "https://.../result.mp4",
    "thumbnail": "https://.../thumb.webp"
  }
}

Results are stored for 48 hours — copy them to your own storage. Authentication, errors and webhooks are documented in the full API reference.

Frequently asked questions

What does the video endpoint need?

A prompt and a duration from the allowed list. A start image is optional: with one, the clip animates it and keeps its aspect ratio; without one, you choose the format.

Can I use pictures as references instead of a start frame?

Yes — send references, one to three URLs, and name them in the prompt as Image 1, Image 2, Image 3: the video puts that person, that garment, that place into a new scene, in the format you choose. References replace start_image and end_image; a call with both is refused.

How is video billed?

Per second of output — see the per-second standard and pro prices in the catalogue.

How much does it cost?

2.2 credits per second on the standard tier, 4.4 per second on pro, 23 per second on ultra. Credits come with plans and packs; a failed generation is always refunded.

Other endpoints: Virtual Try-On API · AI Fashion Models API · Fashion Design API · Reimagine Model API · Swap Model API · AI Stylist API · all endpoints