An AI avatar is a reusable synthetic person your brand generates once and reuses across every campaign. DesignerBox builds one as a nine-image set for 25 credits, and each later scene costs 5 credits to render. The avatar holds because you condition every generation on that set instead of describing the face in words. In the EU, the brand publishing the result carries the labelling duty, not the tool.
You need a presenter for six product launches this quarter. A casting call, a shoot day, and a usage licence per market comes to more than the campaign budget, and the licence expires before the product does.
So you generate a person instead. The first render is excellent. The fourth one has a different nose, and by the tenth your presenter has quietly become someone else.
This covers what an AI avatar actually is as a production unit, why identity drifts and what stops it, which models in the catalogue document holding a face, what a full campaign costs in credits, and the disclosure rules that landed six days ago. It is written for marketing teams and agencies producing presenter-led creative.
Key Takeaways
- The reference set is the asset, not the prompt. An avatar holds because every later generation is conditioned on a fixed set of images of the same person. A face described in words is a new person on every render.
- Nine images, 25 credits, once. DesignerBox generates an avatar as a nine-pose set for 25 credits. Every scene you place that avatar into afterwards is a 5-credit image generation.
- Video is the line where budgets break. Stills are flat-rate. Video bills per second of output, and an 8-second clip on the most expensive model in the catalogue costs more credits than an entire Premium month.
- “Only real people count” is a misreading. The Commission’s Article 50 guidelines say it is enough for a simulated person to resemble someone who “can plausibly exist”, so an invented face clears that particular gate. Whether an anonymous presenter then needs a visible label is a narrower and still unsettled question, covered below.
- The labelling duty sits with you. Article 50(4) of the EU AI Act puts the deepfake disclosure obligation on deployers, which is the brand publishing the ad, not the vendor that rendered it (digital-strategy.ec.europa.eu, August 2026).
- A synthetic presenter making claims is a separate legal question. In the US, the FTC’s Rule on the Use of Consumer Reviews and Testimonials took effect on 21 October 2024 and covers AI-generated reviews (ftc.gov, August 2026). An avatar may present. It may not testify.
- Commercial use starts at Pro. The commercial licence begins on the $35/month tier. Putting avatar output into a paid campaign on a lower plan is the wrong plan, not a grey area.
What is an AI avatar, in production terms
An AI avatar is not a filter and not a one-off portrait. It is a locked identity you can spend against.
The useful definition is operational. An avatar is a set of reference images of one person, consistent in bone structure, skin, hair, and lighting, that you feed into every subsequent generation so the model renders that person rather than inventing one. The set is what you own. The model is only the renderer.
That distinction decides everything downstream. If your presenter lives in a prompt, you have a description, and descriptions resolve differently every time. If your presenter lives in a reference set, you have an asset, and assets are reusable.
DesignerBox generates the set directly: an avatar is a nine-pose character set costing 25 credits, produced from a brief or an uploaded face. The AI avatar generator on designerbox.ai is where that runs.
This is the same mechanic covered in the guide to building an AI brand character, viewed from the other end. That article is about deciding who the character is and how to shoot the reference set. This one is about what the avatar costs to operate once it exists.
Why avatars drift, and what actually stops it
Drift is not a quality problem. It is a conditioning problem.
An image model holds one description of a picture. Identity is one property of that description competing with pose, wardrobe, lighting, and background. Change the scene and you have changed the description, so the face moves with it unless something pins it.
Three things pin it, in descending order of reliability.
A reference set beats a reference image. One photo gives the model a single angle to satisfy. Nine give it a face from multiple directions, which is what stops the jaw and nose resolving differently when the camera moves.
A seed and a fixed prompt skeleton beat free text. Keep the wording that describes the person byte-identical across generations and vary only the scene clause. Rewriting the description each time reintroduces the ambiguity the reference set exists to remove.
Editing beats regenerating. Once you have a frame you approve, edit that frame into the next scene rather than generating a fresh one. An edit starts from your approved pixels. A generation starts from nothing.
Google is the one provider in the catalogue that publishes a number for held identity. Nano Banana Pro documents up to 5 character reference images (ai.google.dev, August 2026), and Nano Banana Pro on designerbox.ai is the default choice when the face has to survive a scene change. Kontext Multi is the tool for holding an approved frame through successive edits, which is the second half of the job.
What a presenter campaign costs in credits
Stills are predictable. Video is not, and the gap between them is wider than most plans assume.
| Operation | Credits | What you get |
|---|---|---|
| Create avatar | 25 | The nine-pose reference set, once |
| Generate image | 5 | One new scene with that avatar |
| Edit image | 5 | One approved frame moved into a new scene |
| Video | Per second of output | Priced by model and duration |
Run that against a real brief. A presenter, six launches, four stills each.
The avatar costs 25 credits once. Twenty-four stills at 5 credits each is 120. Total 145 credits, inside the 500 that Basic includes at $15/month, with room for the retries that presenter work always needs.
Now add video. Video is billed as credits per second multiplied by duration, which makes duration the budget lever, not clip count. An 8-second clip on the most expensive video model in the catalogue costs 6,400 credits, which is more than the 2,500 a Premium month includes. One clip. That is the number to plan around, and it is why presenter video gets storyboarded as stills first and animated only once the frame is approved.
Two gates matter here. AI video needs Premium at $75/month. The commercial licence needs Pro at $35/month. Check both against the pricing page before a campaign, because discovering the video gate mid-flight costs a week.
Who has to disclose a synthetic presenter
This is the part that changed recently, and the part most guidance still gets backwards.
The EU AI Act’s transparency obligations have applied since 2 August 2026, and the Commission’s guidelines on those obligations set out how the deepfake test works (digital-strategy.ec.europa.eu, August 2026). Three things in there decide how you operate an avatar.
“Existing persons” does not exclude your invented one. The Act defines a deepfake as content resembling “existing persons, objects, places, entities or events”, which reads as though a fully invented face escapes. It does not. The Commission’s guidance states that it is enough for simulated persons to resemble someone or something that exists, “can plausibly exist or could have plausibly existed in reality”. A photorealistic presenter who happens to be nobody plausibly exists, so that gate is met.
But the criteria are cumulative, and one of them is contextual. Clearing the plausibility gate is not the same as owing a label. The test also asks whether the content would falsely appear to a person to be authentic or truthful, which depends on context and framing rather than on resolution. Photorealism makes that likelier without deciding it. Clearly fantastical or physically impossible content sits outside the definition entirely.
So treat an anonymous presenter as genuinely unsettled. The worked examples in the guidance deal with recognisable people, and an ordinary anonymous model in a product ad is not among them. There is no enforcement decision on that case yet. The asymmetry is what should drive the choice: a label costs you almost nothing, and the penalties attached to Article 50 reach 15 million euro or 3% of global annual turnover. Label by default and treat the exceptions as a decision your counsel signs off, not one a marketer makes.
The duty lands on the deployer either way. Article 50(4) puts deepfake disclosure on deployers rather than providers, and disclosure must reach the viewer on first exposure in a clear and distinguishable way. The deployer is the brand running the ad. A vendor’s machine-readable marking does not discharge your obligation, because the two duties sit in different places in the Act.
Platform labels run in parallel and on their own logic. Meta applies an “AI info” label across Facebook, Instagram, and Threads when it detects C2PA or IPTC provenance markers, or when the uploader discloses (transparency.meta.com, August 2026). That label is automatic and is not a substitute for the disclosure the Act asks of you.
The US exposure is a different shape. There is no federal synthetic-media labelling rule for ads, but the FTC’s Rule on the Use of Consumer Reviews and Testimonials took effect on 21 October 2024 and expressly reaches AI-generated reviews (ftc.gov, August 2026). The practical line: an avatar can introduce a product, demonstrate it, and read approved copy. An avatar cannot deliver a customer testimonial, because there is no customer. The same reasoning is worked through for garment imagery in the guide to labelling AI-generated fashion images.
None of this is legal advice. Both rules are recent and your counsel should see the actual creative.
When an avatar is the wrong tool
Three cases where a real person is the answer.
Testimonials and reviews. Covered above. A synthetic face attesting to a real experience is a fabricated endorsement regardless of how it is labelled.
Regulated claims. Health, financial, and medical categories combine a synthetic presenter with a claim that needs a qualified human behind it. The disclosure burden compounds instead of resolving.
Founder and team presence. Audiences buy the actual person. Replacing a founder with a rendered one trades the only asset that could not be copied.
An avatar earns its place where the presenter is a role rather than an identity: product demonstration, format-filling social creative, and the twenty variants a paid test needs before anyone knows which one works.
Getting one into production
Five steps, in order.
- Decide the role, then the face. Write who this person is to the brand before generating anything. A reusable presenter needs a defined role, not an attractive render. The brand spokesperson persona on designerbox.ai is a worked example of the specification.
- Generate the set and lock it. Nine poses, 25 credits, one session. Approve or regenerate the whole set. Do not mix poses from two attempts, because that is two people.
- Fix the prompt skeleton. Keep the identity wording identical forever. Vary only the scene.
- Build scenes by editing, not regenerating. Move approved frames into new backgrounds and wardrobes at 5 credits each.
- Make it repeatable. Save the sequence so the next campaign reruns it instead of rediscovering it. The brand-locked spokesperson workflow is the version of this that survives a team handover.
Teams running this at volume drive it from a chat client instead of the canvas. The MCP server exposes 43 tools across 8 groups, avatar creation among them, which is what makes an avatar set something a script can produce on a schedule. The setup is covered in the guide to AI creative agents.
FAQ
How much does an AI avatar cost?
In DesignerBox, an avatar is a nine-pose set for 25 credits, and every scene you place it into afterwards is a 5-credit image generation. A presenter with twenty-four stills across six launches costs about 145 credits, which fits inside the 500 credits Basic includes at $15/month. Video is priced per second of output and costs substantially more.
Can I use an AI avatar in paid advertising?
Yes, with two conditions. The commercial licence starts at the Pro tier ($35/month), so the plan has to support it. And if the avatar is photorealistic and you are advertising in the EU, the disclosure obligation in Article 50(4) applies to you as the deployer, not to the tool that rendered it.
Do I have to label an AI avatar if the person does not exist?
Probably, and the safe answer is to label. The Act’s deepfake definition mentions “existing persons”, which many readers take as an exemption for invented faces. It is not one: the Commission’s guidance says it is enough for a simulated person to resemble someone who “can plausibly exist or could have plausibly existed in reality” (digital-strategy.ec.europa.eu, August 2026). The criteria are cumulative though, and the remaining test of whether content would falsely appear authentic is contextual. For an anonymous presenter in a product ad the question is genuinely open, with no enforcement decision yet, so label by default and let counsel approve any exception.
Which model holds a face most reliably?
Nano Banana Pro is the one that publishes a figure, documenting up to 5 character reference images (ai.google.dev, August 2026). For carrying an already-approved frame through a series of edits rather than generating fresh scenes, Kontext Multi is the better fit. Both are in the catalogue on one subscription.
Why does my AI avatar look different in every image?
Because the face is living in the prompt instead of in a reference set. A text description resolves differently on every generation. Lock a nine-image set, keep the identity wording byte-identical, vary only the scene clause, and build new scenes by editing an approved frame rather than regenerating from scratch.
Can an AI avatar give a customer testimonial?
No. The FTC’s Rule on the Use of Consumer Reviews and Testimonials took effect on 21 October 2024 and covers AI-generated reviews (ftc.gov, August 2026). A synthetic person has no experience to report, so a testimonial delivered by one is fabricated. An avatar can demonstrate a product and read approved brand copy.
Do I need a video plan to use an avatar?
Not for stills. Image generation and editing run from Basic upward. AI video requires Premium at $75/month, and video bills per second of output, so a talking presenter is a materially different budget from a set of presenter stills.
Avatar mechanics and credit costs verified against the DesignerBox product brief. EU AI Act Article 50 obligations, Commission guidance, Meta labelling policy, and FTC testimonial rules verified from primary sources as of August 2026. Model reference-image counts verified from provider documentation. This is not legal advice; both the EU and US rules cited are recent and your counsel should review actual creative. Individual results vary.
