Skip to main content

A 5-minute presenter from one photo: the consent checklist

By Alan Magliocca · November 18, 2026

Quick answer: In mid-October 2026, ByteDance's BytePlus made OmniHuman 2.0 available by API, generating up to five minutes of talking-person video from one photo and an audio track; reported prices differ between $0.07 and $0.12 per second. Because one photo is all it takes, get written consent from the person in the photo, and the voice, before you make a presenter of anyone.

KEY TAKEAWAYS
  • OmniHuman 2.0 turns one image plus an audio track into a single talking video of 30 seconds to five minutes, offered by API through BytePlus.
  • Reported prices conflict: Webrazzi reports $0.07 per second, while BytePlus's product page lists both $0.07 and $0.12 per second without naming versions.
  • A five-minute video works out to $21 at $0.07 a second or $36 at $0.12; check the BytePlus console before you budget.
  • One photo is easy to find, which is exactly why consent matters: a photo you hold isn't permission to make that person speak.
  • Likeness rules are tightening in California, the EU and elsewhere; check with a lawyer for your own case.

What happened

ByteDance's enterprise AI platform, BytePlus, has made OmniHuman 2.0 available to developers through an API. Webrazzi reported on October 15, 2026, citing information shared by BytePlus, that the model generates a single continuous video of a talking person of up to five minutes from one image and an audio recording. Optional text prompts can direct movement, camera behavior, added objects and scenes with more than one character.

The facts that line up across sources:

  • Inputs: one image, one audio track, and an optional text prompt.
  • Length: 30 to 300 seconds of output, per BytePlus.
  • Resolution: BytePlus's product page says native 720p output.
  • Access: an API through BytePlus.

Price is where the reports disagree. Webrazzi reports $0.07 per second, saying the cost fell by 40% in this version. BytePlus's own product page lists two rates side by side: $0.07 per second on a listing marked "30s to 300s video output" and "40% lower cost," and $0.12 per second on a lip-sync listing. The page doesn't say which rate belongs to which version, and its technical report link still points to version 1.5. Until BytePlus labels them, treat both numbers as possible and check the console before you budget.

What it means for people selling with webinars

A five-minute presenter video from a single photo is long enough for a webinar intro, a module of an evergreen session or a personal follow-up. At either reported price, it costs dollars, not a film crew.

The bigger change is how little it takes. Earlier avatar tools usually needed minutes of recorded footage of the person, which meant that person had to take part. One photo and an audio file can come from anywhere: a speaker page, a LinkedIn profile, a podcast. That's what makes the consent question urgent. The technology no longer forces you to ask. You have to choose to.

Rules are moving the same way. California's SB 1111 adds digital replicas to the state's likeness laws from January 1, 2027, covered in California makes digital replicas a likeness issue. The federal NO FAKES Act is still waiting for a floor vote, covered in get written consent for any voice or face now. In the EU, deepfake content must be clearly disclosed. None of this is legal advice; check with a lawyer for your own case.

What to do this week

  1. List every photo you might be tempted to animate: your own headshots, team photos, guest speakers, past testimonial givers.
  2. Mark which ones you have written, specific consent for. A model release for a photo shoot isn't consent to make that person speak.
  3. Do the same for every voice recording, including podcast clips and webinar audio.
  4. Write a one-page consent form using the checklist below, and send it to anyone you plan to animate.
  5. Decide on your disclosure line before you make the first video.

How BrightStage AI helps

BrightStage AI doesn't use OmniHuman. Its video creator builds a script and slides with AI presenters, and the consent questions below apply to any presenter that looks or sounds like a real person. For what to gather before building anyone's twin, see what to collect from a presenter.

The one-photo consent checklist

Use this before you make a presenter video of anyone, including yourself:

  1. Who is in the photo? Name every person visible, including people in the background.
  2. Who owns the photo? A photographer may hold the rights even if it's your face. Check the license allows AI use.
  3. Who owns the voice? If the audio is someone else's recording, or a cloned voice, the voice owner needs to consent too.
  4. What exactly did they agree to? Get it in writing: which videos, which channels, whether paid ads are included, and for how long.
  5. Can they see the script first? Agree whether they approve each script or a general topic.
  6. How do they withdraw? Write down how they can withdraw consent and how quickly you'll take videos down.
  7. Is payment involved? If a guest or contractor's likeness sells your product, settle compensation in writing.
  8. How will viewers know it's AI? Decide the label: on screen, in the description and in your first spoken line.
  9. Where will it be shown? If EU viewers will see it, plan for the disclosure rules there. If it's California, factor in the 2027 changes.
  10. Where are the files kept? Store the source photo, audio, consent form and every output together, so you can show what was agreed.

If you can't answer all ten for a photo, don't animate it.

What does a five-minute presenter cost?

Here's the cost math using both reported rates, since BytePlus hasn't settled which applies:

  • One 5-minute intro (300 seconds): $21 at $0.07, or $36 at $0.12.
  • Six 5-minute modules for an evergreen session (1,800 seconds): $126 at $0.07, or $216 at $0.12.
  • Ten 90-second personal follow-ups (900 seconds): $63 at $0.07, or $108 at $0.12.

These are generation costs only, before redos, voice recording or editing. Budget at the higher rate until you see the real charge in your account. Remember the 720p output: next to a 1080p or 4K scene it may look softer, which matters if you mix it with other footage.

The rules on AI performers in ads are moving for agencies too. AIVX Labs covers California's labeling rule in AI synthetic performers in ads need a label.

Sources

AIVX LABS ACCOUNT VERIFICATION
Get AIVX Labs verified for $1: RICO, your AI operator, and 35+ AI tools, with 7 days of full access.
Get AIVX Labs Verified for $1 →

Frequently asked questions

How much does OmniHuman 2.0 cost?

Reports conflict. Webrazzi reports $0.07 per second, and BytePlus's product page lists both $0.07 and $0.12 per second without naming which version each applies to. Check the BytePlus console for the rate on your account.

How long can an OmniHuman 2.0 video be?

BytePlus lists 30 to 300 seconds of output, so up to five minutes in one continuous video.

Do I need consent to animate a photo I already own?

Owning a photo isn't the same as permission to make the person in it speak. Get written consent from the person and the voice owner; rules vary by place, so check with a lawyer for your case.

Related reading

Written by Alan Magliocca, founder of BrightStage AI and AIVX Labs (Podavinci LLC). Written with AI assistance and checked against the linked sources. Follow along: LinkedIn · YouTube · TikTok · X · Facebook · Instagram · Skool community