Synthesia Turns a Single Photo Into a Talking Avatar

Synthesia is a company that uses AI to produce videos in which a person appears to be explaining something on camera — think corporate training materials or internal announcement videos. It was founded in London in 2017, and what it builds is different from general-purpose video generation models like Sora or Veo. Synthesia's avatars aren't imagined faces; they're created by filming real actors and obtaining their consent. The newly unveiled single-photo avatar feature is an extension of that same approach. It takes a single photo of a real person and turns that face into a talking character.

On August 17, Synthesia also released a summary video introducing seven new features it had rolled out over the month of July, including enhanced role-play and dubbing capabilities. A day later, on August 18, it published a video featuring an avatar modeled on an ordinary employee rather than a celebrity impersonation, in an apparent effort to address misconceptions about AI avatars. This latest single-photo avatar walkthrough continues in that same line of content.

How it works

Where to start

The process begins on the page for turning a photo into an avatar, where the user uploads a photo. There's no need for filming equipment or a studio setup — just an existing photo file.

Step-by-step process

The video walks through the process in the following order:

1. Upload a photo to convert it into a talking avatar.
2. Choose the outfit the avatar will wear.
3. Choose the background (location) where the avatar will stand.
4. Specify what the avatar will say and how it will move.
5. Convert a document file directly into an avatar video.
6. Share or download the finished video.

Who can use it

The video doesn't provide separate guidance on pricing plans or country-specific availability. However, it repeatedly emphasizes two points: that the entire process starts from a single photo, and that no camera or lighting equipment is needed.

What you can try

For example, someone in charge of a company's promotional materials could create a company introduction video from a single photo of a representative, or turn a presentation PDF directly into a video narrated by an avatar. Individual users could turn a single profile photo into a short social media video with a different background and outfit.

Editor's view

Lowering the barrier to creating avatars is already a well-established direction in this industry. HeyGen previously showed how to make real-estate home tour videos without a camera, and Adobe has posted shorts centered on AI clones. Synthesia lowering the input requirement to a single photo fits the same trend. What sets it apart is that Synthesia built its avatar business from the start on filming real actors with their consent — meaning this new feature effectively replaces that filming process with a single photograph.

Applying this category of tool to real work tends to lead to a similar conclusion each time. The cost of a filming studio and camera crew disappears, but how natural the result looks depends heavily on the quality and angle of the original photo. In other words, preparing a single high-resolution photo that clearly captures a front-facing shot is the variable that determines the overall outcome.

For companies in Korea, this is worth testing first on content where repeated filming is a burden — internal training videos or new product introductions, for instance. That said, since the feature uses the face of a real person as an avatar, obtaining that person's consent to use their photo is something the company itself needs to handle separately from the tool. It seems likely that Synthesia's channel will continue rolling out real-world use-case videos built around this feature in the coming weeks.