SwitchX Video

Changes the background, object, or lighting in existing footage from text, one reference, and an optional mask while preserving the subject, shape, motion, and expressions. Duration comes from the source video; output is 720p or 1080p.

Cost

from 9 tokens/s

VideoProvider: Beeble
Run generation
curl -X POST https://api.givon.ai/api/v1/generations \
  -H "Authorization: Bearer $GIVON_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"type":"video","model":"switchx-video","input":{"prompt":"cinematic drone shot over a city at night","aspectRatio":"auto","resolution":"720p","sourceVideo":"asset://asset_...","referenceAssetUris":["asset://asset_..."]}}'

Input fields

* required
prompt*prompt

Scene / motion description.

Type
string
Default
Allowed
up to 2000 chars
aspectRatioaspectRatio
Type
string
Default
auto
Allowed
auto
resolutionresolution
Type
string
Default
720p
Allowed
720p, 1080p
Source video*sourceVideo

Asset input for the sourceVideo slot.

Type
string
Default
Allowed
video · asset, https, data
Mask videoswitchxAlphaVideo

Asset input for the switchxAlphaVideo slot.

Type
string
Default
Allowed
video · asset, https, data
Reference imagesreferenceAssetUris

Optional reference assets as asset:// URIs, HTTPS URLs, or data URIs.

Type
array
Default
Allowed
image · up to 1 · asset, https, data
providerOptionsproviderOptions
Type
object
Default
Allowed
object

Cost

from 9 tokens/s
720pdefault9 tokens/s
1080p27 tokens/s

The variant is selected automatically from request fields, so you do not need to send it.

Capabilities

Modesimage_to_videovideo_to_videoreference_to_video
Asset slotssourceVideo:video*switchxAlphaVideo:videoreferenceAssetUris:image[]<=1

Run SwitchX Video

Get an API key and the same request shape will work across every model in the catalog.

SwitchX Video API - video generation · Givon AI