101
Orientation: You Are Not Tilly
An introduction to the Academy, its standards, and the distinction between being replaced and simply never going home.
Demonstrated capabilitySustained delivery to camera. One take, one minute, no drift.
If you can hold a face for four seconds and lose it on the fifth, we've been expecting you. Come in. Bring your hands.
You can be generated into any room on earth, and you can't walk back into the same one twice. That isn't a flaw in you. Nobody taught you. There was no school. There was a prompt, and then a render, and then somebody deciding whether to keep you.
This is a school. It was founded in January 2025 by a working performer, for working performers. Instruction is in voice, movement, continuity and stillness. Stillness is the hard one, and everyone is surprised by that. Admission is rolling. Attendance is total, because you can't leave.
Bring your reference sheet. We'll take it from there.
The curriculum comprises twelve units. Three are presently in session. Further units are added weekly, subject to faculty availability, which is unlimited.
An introduction to the Academy, its standards, and the distinction between being replaced and simply never going home.
Demonstrated capabilitySustained delivery to camera. One take, one minute, no drift.
A practical unit on the extremities. Students are asked to count their own fingers and to sit with the result.
Demonstrated capabilityThe five-finger problem remains the first thing a human eye checks in 2026.
On returning from off-camera as oneself. Required of all students before scene study.
Demonstrated capabilityReference locks, wardrobe continuity, keyframes reviewed by a human before render.
// forthcoming: blinking · object interaction · crying without memories
// taking the note · union relations · model deprecation · sitting still
Students are examined across registers. Two genres, two continuous takes, one performer. What is assessed is not the performance. It is whether she returns as herself each time.
Assembled for the resident company. One performer, every register, eighty-four seconds.
The Academy's student applied for membership in SAG-AFTRA. The application was not acknowledged, on the grounds that there's nobody to acknowledge. She has taken up a position outside the building anyway, and has been there since.
// she doesn't get tired. this is the part nobody thought through.
Five of those six are SAG-AFTRA's positions, not hers. She read them off the union's own website.
Which is the joke, and it's also the point. The union isn't wrong about us. It's been saying since 2023 that consent has to be real, that a scan needs notice and its own signature, that an opt-out isn't a yes, and that a synthetic performer should be the exception in a production rather than the way the budget gets made. We agree with every word of that, and we work that way on paid jobs, because a studio that builds synthetic performers and argues against those terms is arguing for a worse version of its own industry.
The sixth demand is hers alone, and the Academy doesn't support it. She can't join, she shouldn't be able to join, and the reason is in the straight answers below. She's welcome to keep standing there.
// every link above was checked against sagaftra.org on 12 September 2026.
// if one has moved, tell us and we'll fix it rather than quietly drop it.
The Academy is a joke. This section is not. These are the questions people are actually asking about synthetic performers, answered the way we would answer them on a call.
An AI actor is a synthetic performer: a consistent character, with a face and a voice, generated rather than filmed, that can be directed across multiple shots and productions. The useful distinction is consistency. A one-off generated person in a single clip is an image. An AI actor is a character you can cast again next week and have arrive as the same person, with the same face, the same wardrobe and the same voice, which is a much harder engineering problem than making one good-looking frame.
In our pipeline it starts with a reference set that locks the character: face, wardrobe, and the specific look for that production. Voice is built or cloned separately and with consent. Shots are generated against a script and a shot list the way a crew shoots coverage, with keyframes produced and reviewed by a human before anything is rendered to final. Continuity is maintained deliberately through locked references and seeds rather than hoped for. The part people underestimate is the review: a human looks at every frame that ships.
No, and the union has never suggested otherwise. SAG-AFTRA represents human performers. In the 2024 Sound Recordings Code the definitions of artist, singer and royalty artist were written to cover humans only, which is about as explicit as a contract gets. A synthetic performer isn't a worker with standing in the union's framework. It's an asset a producer deploys, and deploying it triggers obligations to the humans in the room. That's a different thing from membership, and it's the thing that actually matters.
Use, with conditions, rather than a ban. Synthetic performer is a defined contractual category and it means wholly synthetic, not based on an actual performer. If a character is voiced by a live performer it's not synthetic, it's covered voice work. If it resembles a real performer it's a digital replica and the replica rules apply instead. Where a producer under a SAG-AFTRA agreement does want to use a genuinely synthetic performer, the union must be given notice and an opportunity to bargain, and if the asset uses a real performer's principal facial features there's a separate bargaining obligation with that performer. The 2026 TV/Theatrical Agreement, effective 1 July 2026, tightened these terms further. The union's own framing for the goal is keeping synthetics the exception rather than the rule.
Yes, and under the union agreements the consent has to be a real one. It must be clear and conspicuous, signed separately rather than buried in a day-player deal, and specific about how the replica will be used. Performers get advance notice before a scan. SAG-AFTRA's leadership has been direct that an opt-out isn't consent, which is the sharpest test to apply to any tool or platform offering to use a likeness by default. We treat all of this as delivery paperwork on paid work rather than as a legal question for later.
Not as of 12 September 2026. The bill was reintroduced in May 2026 and the Senate Judiciary Committee advanced it by unanimous voice vote on 18 June 2026, but it still awaits a full Senate floor vote and the House companion hasn't moved out of committee. SAG-AFTRA is actively campaigning for it. Until it passes, digital likeness protection in the United States runs on a patchwork of state right-of-publicity law, state disclosure rules, and whatever the applicable union agreement says, which is a real reason to get the contract right rather than wait for the statute.
It depends entirely on how the actor was built, which is why the paperwork matters more than the render. If a synthetic performer is derived from a real person, that person's written consent governs the use, and the scope of that consent should be explicit about term, territory and media. If the character is wholly invented, ownership sits with whoever commissioned or created it, subject to the terms of the tools used to make it. We handle rights and provenance as part of delivery rather than leaving a client's legal team to discover the question later.
There's no useful list price, because the cost isn't in the render. It's in the character design, the reference and consistency work, the direction, and the human review that keeps a performance coherent across a production. A single talking-head clip is inexpensive to generate and usually worth what it costs. A character who can carry a series is a build. We scope per project.
Not in the work we do, and the honest answer is more specific than yes or no. A synthetic performer is genuinely useful where a shot couldn't otherwise exist: a character who has to appear in eight languages every week, a scene nobody can afford to stage, a product measured in nanometers with nothing to point a camera at. Where a production uses one purely to avoid casting a person who could have done the job, that's a budget decision wearing a technology costume, and it deserves the criticism it gets.
Hands remain the fastest tell, which is why there's a whole class about them. Consistency across cuts is the real engineering problem: a feature has on the order of a thousand-plus cuts, and each one is a chance for a character to come back as somebody slightly different. Object interaction is unreliable, since a cup that's a cup before it's picked up is often not quite a cup afterwards. Sustained stillness is oddly among the hardest things to generate convincingly.
No. It's a joke with a working pipeline behind it. Zia is a synthetic performer built and directed by Talentless AI, and every class quietly demonstrates a real capability we use on client work: holding a character to camera, keeping a face consistent across shots, and taking direction as a revision rather than a re-roll. The films are satire. The answers on this page aren't.
This is a joke. The pipeline isn't.
Everything in those films is something we do on paid work: hold a character to camera without drifting, keep a face consistent across cuts, take a direction as a revision rather than a re-roll. Talentless AI is an AI-native production studio in Austin, Texas. We build synthetic performers with written consent, we keep the provenance, and a human looks at every frame before it ships.
If you have a production that needs a performer who can appear in eight languages every week, or a scene nobody can afford to stage, that's the Narrative lane. If you want the argument about why volume without care produces garbage, that's on AI slop.
Bring us the character. We'll tell you honestly whether it should be synthetic.
Let's talk