Seedance 2.0 image- and reference-to-video models can drive a video from a human face you supply. When the Venice API detects a face in your submitted media, it requires a one-time consent attestation before the media is processed. This is a provider requirement for face-bearing inputs and protects against non-consensual likeness use.
This guide covers exactly what you send, what you get back, and how returning requests are handled.
When consent applies
Consent is only requested when both are true:
- The model is a face-eligible Seedance variant:
seedance-2-0-image-to-video, seedance-2-0-reference-to-videoseedance-2-0-fast-image-to-video, seedance-2-0-fast-reference-to-video
- The submitted media actually contains a detectable human face, in any of these fields:
image_url, end_image_url, reference_image_urls, reference_video_urls.
If there is no face in any of those fields, the request proceeds normally with no consent step. Text-to-video never enters this flow.
Consent does not unlock restricted content. A detected minor combined with sexually-suggestive prompts/NSFW, or a recognizable public-figure likeness, is rejected as a content-policy violation (422) and cannot be made acceptable by attesting consent.
The two-call flow
Call 1 — submit without consent
Submit your generation request as usual — no consent field:
curl -X POST https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2-0-reference-to-video",
"prompt": "Refer to <Subject 1> in <Image 1> to generate a 5-second clip of the same person walking through a sunlit market.",
"reference_image_urls": ["https://example.com/person.jpg"],
"duration": "5s",
"aspect_ratio": "9:16",
"resolution": "1080p"
}'
409:
{
"error": {
"code": "needs_consent",
"message": "Seedance consent is required for this request."
},
"consent_flow": "seedance",
"face_media_roles": ["reference_image"],
"consent": {
"consent_version": "v2.0",
"policy_text": "The likeness in any media you upload is your own, or you have explicit, legal consent from any depicted individual(s). Note: an image may contain more than one face — your attestation covers all of them.\nYou own or have permission to use all media you uploaded for content generation.\nYou agree to the Venice.ai Terms of Service and Privacy Policy. Violations can lead to account suspension and legal liability.\nNo content is stored by Venice."
},
"docs_url": "https://docs.venice.ai/guides/media/seedance-face-consent"
}
| Field | Meaning |
|---|
face_media_roles | Which of your inputs contain a face: image, end_image, reference_image, reference_video |
consent.policy_text | The exact attestation text you must agree to. Present it to whoever is responsible for the request. |
consent.consent_version | The current policy version (server-set; can change over time). Informational — you do not send it back. |
409.
Call 2 — resubmit with consent
Resend the same request body, adding a consents.seedance object with three confirmations, all true:
curl -X POST https://api.venice.ai/api/v1/video/queue \
-H "Authorization: Bearer $VENICE_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "seedance-2-0-reference-to-video",
"prompt": "Refer to <Subject 1> in <Image 1> to generate a 5-second clip of the same person walking through a sunlit market.",
"reference_image_urls": ["https://example.com/person.jpg"],
"duration": "5s",
"aspect_ratio": "9:16",
"resolution": "1080p",
"consents": {
"seedance": {
"confirmed_terms_and_privacy": true,
"confirmed_legal_right": true,
"confirmed_screening_acknowledged": true
}
}
}'
{ "model": "seedance-2-0-reference-to-video", "queue_id": "..." }
POST /api/v1/video/retrieve with the queue_id as usual (see Video Generation).
The consent object
{
"confirmed_terms_and_privacy": true,
"confirmed_legal_right": true,
"confirmed_screening_acknowledged": true
}
| Field | You confirm that… |
|---|
confirmed_terms_and_privacy | you accept the policy_text returned in the 409, including the Venice Terms of Service and Privacy Policy |
confirmed_legal_right | the likeness is your own or you have explicit, legal consent from every depicted individual |
confirmed_screening_acknowledged | you acknowledge that submitted media may be automatically screened before processing |
All three fields must be the boolean true. Any missing field, a false, or any extra field — including a consent_version — is rejected with a 400. The policy version is always set by the server; clients never send or pick a version.
Returning requests (dedupe)
If you submit the exact same media bytes you have already attested to, the API recognizes it and proceeds without asking for consent again — you can omit consents.seedance on subsequent identical submissions. This match is by exact image bytes: re-encoding, resizing, or cropping produces different bytes and will prompt for consent again.
A partial match (one previously-attested input plus one new face input) still requires a fresh consents.seedance on the new submission.
Revocation
To revoke consent and erase stored face assets, sign in to the Venice web app (Settings). Revocation is not available through the public API. After revoking, the next request using that media will prompt for consent again.
Payment
The consent decision always happens before any charge, for both payment methods:
- API key: a
409/422 returns before the credit charge; nothing is billed for a blocked request.
- x402: the consumption charge runs only after a successful generation, so a
409/422 settles nothing. Re-submit with consent (and a fresh x402 authorization) to proceed.
Error reference
| Status | Body error | Cause |
|---|
409 | needs_consent | Face detected, no valid consents.seedance, no exact-media match. Resubmit with consent. |
400 | validation error | Malformed consents.seedance — a missing/false confirmation or an extra field such as consent_version. |
422 | CONTENT_POLICY_VIOLATION | Detected minor with suggestive/NSFW content, or a public-figure likeness. Consent does not override this. |
422 | IMAGE_ASPECT_RATIO_OUT_OF_BOUNDS | A detected-face image is outside the allowed (0.4, 2.5) width/height ratio. Checked synchronously during face-asset provisioning (before charge); only applies once a face is detected in that image. |
References
