A talking-head tool puts consent on the uploader by contract

The shot

D-ID's Creative Reality Studio animates a still photograph so that it appears to speak a script, with lip movement synced to text, audio, or a recorded voice. The company, per its own about page, was founded in 2017 and states its studio and API have 'powered more than 200 million avatar videos' for a developer community it puts at over 280,000; this note treats the product documentation as a living record, current as retrieved on 16 September 2026, rather than pinning it to one dated event.

What the documents show

The product page states the studio turns 'text, audio, or still images into high-quality videos with lifelike digital avatars,' accepting a photo up to 10MB in JPEG or PNG form, or a face generated from a text prompt, and caps output at five minutes with standard avatars rendered up to 1280×1280 and premium avatars at 1080p on paid plans. The company's own end-user license agreement is explicit about consent at the point of upload: users must certify they 'own or have the necessary rights and permissions' to any submitted image, and the agreement separately prohibits content that 'impersonates another person' or includes another person's personal information without 'first obtained the express consent' of that person. These are the company's own contractual terms, a documented policy, not an independently audited enforcement record of how consistently they are applied.

The workflow

As documented, a user uploads or generates a face image, supplies a script as text or an audio file, and the studio renders a video with synced lip movement, all inside the size, duration, and resolution limits the product page states. The license agreement layers a second step onto that workflow: the user attesting to rights and consent before the image is submitted at all, a gate the documents place on the person uploading rather than on the system verifying identity itself.

What the tool does not change

The agreement's consent and impersonation clauses do not describe any technical check confirming a submitted photo's subject actually agreed; the documents place that responsibility contractually on the user, not as a claim about an automated safeguard. Political use, the agreement adds, is 'subject to our individual internal review and approval process,' meaning human review, not the model, is the stated check for that category. Nothing here describes the studio verifying a script's accuracy or a presenter's suitability for a given audience; that remains a production decision.

• Does the platform verify a photo subject's consent technically, or does it rely on a contractual attestation, as the license states here?
• Does the intended output fall within the documented duration and resolution limits for the plan in use?
• What review, per the cited terms, applies to political or sensitive uses before publication?

D-ID's own documentation frames Creative Reality Studio as a bounded product, with stated size, duration, and resolution limits, and a license that puts consent and impersonation obligations on the uploader. That is a documented policy position, not evidence of how it plays out in any specific case not named in these materials.