API 레퍼런스 · Media
Video · edit people
Read the people in a frame of your clip, left to right: what each wears and where their head is.
Free to call. Made for Replace Pro (eroq-replace-pro): send one frame of the clip (its first, where everyone stands) and how many people to read. For each person, left to right: outfit, their clothing and accessories (never their face, body or who they might be) — put it in that person's picture prompt (POST /v1/images/generations) to keep the clip's clothes; and placement, where their head is (x its centre, top the crown, head its height, as fractions of the frame) — send the list as placements to POST /v1/videos/edits so each picture lands at that spot and size. placement is null when a head could not be boxed: send no placements then.
Answers 503 people_unavailable when the frame cannot be read.
요청 본문
One frame of the clip, as a data:image/jpeg|webp|png;base64,… URI (under 3 MB).
How many people to read, left to right — 1 to 6. Default 1.
curl https://eroq.ai/v1/videos/edits/people \
-H "Authorization: Bearer $EROQ_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"frame": "data:image/webp;base64,UklGR…",
"people": 2
}'응답
{
"object": "video.edit.people",
"people": [
{
"outfit": "orange and blue striped short-sleeved shirt open over a white t-shirt, light blue jeans, silver chains, white sunglasses",
"placement": { "x": 0.35, "top": 0.13, "head": 0.18 }
},
{
"outfit": "orange knitted short-sleeved polo, black leather trousers, silver watch, dark sunglasses",
"placement": { "x": 0.67, "top": 0.1, "head": 0.18 }
}
]
}