ReelBolt now has avatars. You can make an avatar of a real person, from their own pictures, short videos and voice recordings, and show them in short generated clips. The person has to agree first, and you record that agreement in ReelBolt.
This post covers what an avatar can do, why consent is built into every step, and the limits we chose to keep.
What you can do now
An avatar belongs to one project. You make it in the project’s Avatars tab, next to Voices, from files you have already uploaded: at least one clear, front-facing picture of the person, and if you like, a short video of them speaking to the camera and short recordings of their voice. You can also link one of the project’s voices to the avatar.
Once the avatar is ready, a workflow can show that person in a short clip. On the app page you pick the person from an Avatar field, then press Run. Before the run starts you see an estimate of the credits it will use.
Why consent comes first
A face and a voice belong to a person. So an avatar works like a cloned voice in ReelBolt: it exists only after the person has said yes.
- You confirm that you are the person, or that you have their written permission. You can attach the signed permission as a file of the project.
- You confirm that the person is 18 or older. Avatars of children are not allowed.
- You read the consent statement in full and tick it. It says the person agreed to their face, body and voice being used to make AI videos in this project, and to their files being sent to the AI services the project uses for that.
- You choose how long to keep the avatar: one year by default, two years at most. You get an email a week before that date.
Consent is not a one-time box. The avatar is checked again just before each clip is made. If consent was withdrawn, the keep-until date has passed, or one of its files changed, the next clip is not made and you are not charged for it.
The person can change their mind at any time. Withdrawing consent makes the avatar unusable everywhere in the project at once, and it is deleted 30 days later. Deleting it removes the avatar and its files straight away. The person shown does not need a ReelBolt account to ask: they can write to us, as our Privacy Policy explains.
Some uses are off limits whatever the paperwork says: an avatar of someone who did not agree, or of a public figure unless the avatar is of you. The full rules are in our Acceptable Use Policy.
One more safeguard: the assistant can tell you which avatars are ready and why one is not, but it cannot make, change, withdraw or delete an avatar. Consent is only ever given by you, in the Avatars tab.
Two kinds of clip
Two ready-made recipes show an avatar.
Spokesperson clip shows the person in a short generated clip that you describe in a prompt. It does not promise exact words, even when the avatar has voice recordings.
Talking avatar (budget) is the cheaper option, and the one to use when the person should say a script. It writes a few sentences from your brief, speaks them in the voice linked to the avatar, makes one still picture of the person, and animates that picture so they say the words. The clip lasts as long as the speech: at least 5 and at most 15 seconds. The still is kept with the avatar and reused for later clips while the avatar is unchanged.
What gets sent, and what it costs
To keep the person consistent, the avatar’s pictures, videos and recordings are sent to the video service with every clip. That needs two separate permissions. The first is the person’s consent, recorded when you made the avatar. The second is a switch on the step, Allow sending your media to the video service, which you turn on yourself. The assistant can never turn it on for you.
Sending pictures and video costs extra with some video services. The estimate you see before a run includes it. For a talking clip, the cost is one still picture plus the seconds of video.
Marked as AI-generated
People watching a clip should know it was made with AI. Every clip of a person is marked as AI-generated in the video file itself, and it shows a small AI-generated label in the picture. You can turn the visible label off for an avatar, but only if you tell your viewers in another clear way. The mark in the file stays either way, and our rules do not allow removing or hiding it.
What avatars do not do
- No stock avatars. Every avatar is a real person who agreed. There is no library of ready-made presenters.
- No long presenter videos. Avatars appear in short clips. There is no recipe for a presenter who talks through a whole video.
- No lip-sync of existing footage. Dubs still keep the speaker’s original mouth movement. ReelBolt does not re-sync lips in footage you already have.
- Limited sharing across projects. An avatar can be used only in the project it was made in, unless an Owner or Admin saves it to the workspace avatar library; then members of that workspace can use it in the workspace's other projects. It never leaves the workspace.
If an AI presenter is the core of your video, a tool built for that may suit you better. Our ReelBolt vs HeyGen and ReelBolt vs Descript comparisons go through the differences.
Plans and availability
Avatars come with Pro, Team and Enterprise. They are not on Creator or the free trial. The pricing page has the plan details.
Avatars are still being rolled out and may not be switched on in your workspace yet. Making clips may reach your workspace later than the Avatars tab. The avatars you prepare now will work with it.
How to start
The Avatars guide walks through the files you need, the consent you give, how to make a clip, and how to withdraw or delete an avatar. To compare avatar clips with the other things ReelBolt makes, see Choose a recipe.