SAG approved an AI film that digitizes real actors, trains custom models on their likenesses, and captures performances using Runway Act-One — all while protecting actor rights through strict usage contracts.
Speaking with us at AI on the Lot for our Inside the AI Studio series, Kavan the Kid from Phantom X reveals how his team navigated union approval, cast established actors like Breckin Meyer, and built a production pipeline that could reshape how films get made.
Watch the full interview here:
Their breakthrough film Echo Hunter represents a significant milestone in AI filmmaking — not just for its technical achievements, but for establishing a framework that protects actor interests while embracing AI tools.
The SAG Breakthrough: Setting Industry Precedent
Getting union approval wasn't about convincing skeptics — it required meticulous documentation of every aspect of the AI process.
We had to be very thorough in our documentation of exactly what we were doing, how the actors would be represented, the fact that when we digitize their likeness, it's for that film and only that film, cannot be used for anything else.
The approval process centered on contractual limitations and transparent usage policies. Each actor's digital likeness gets locked to the specific project, with no rights to repurpose the AI models elsewhere.
Key protection elements include:
Detailed documentation of AI processes and actor representation
Film-specific licensing that prevents model reuse
Clear boundaries on how digital likenesses can be applied
Traditional compensation structures for the human performances
They were surprisingly a little bit more open to it than what we were thinking.
Production Pipeline: Real Actors Meet AI Models
Phantom X developed a hybrid approach that captures authentic human performance while leveraging AI for visual consistency.
The technical workflow combines:
Arcana Labs for training custom actor models
Sound stage capture for principal cast in Los Angeles
Zoom direction sessions for remote actors like Gedeon Burkhard in Germany
Runway Act-One for facial performance transfer
Live audio recording to maintain vocal authenticity
I think what actors bring to the table when they're performing is such nuances that sometimes AI might not come up with or we can’t think to prompt because it's so unique as an individual.
The team deliberately avoided AI-generated voices, recognizing that authentic vocal performances significantly elevate the final product compared to synthetic alternatives.
Cast and Creative Collaboration
The project attracted established talent including Breckin Meyer (Clueless, Robot Chicken), Taylor John Smith (Warfare), and Danielle Bisutti (God of War voice acting).
Production approach prioritized:
Traditional direction methods adapted for AI capture requirements
Facial performance focus optimized for Runway Act-One processing
Collaborative composer workflow with Miguel Johnson creating nine original tracks
Iterative music creation with real-time feedback loops during editing
I would be going through it and be like, I need tension at like 36 seconds, and then I need it to drop at like 45. And I'd send him emails like at 2 in the morning, “Hey, man, I need 35 seconds, it goes up. And then, at 42 seconds, it goes back down.” Within 10 minutes, this guy would hit me back.
Technical Limitations and Future Vision
Current AI tools create specific constraints that influence creative decisions. The team found that mixing real plates with AI backgrounds can create visual inconsistencies that break immersion.
Until we get to, and I have no doubt it will, when the tools get to where it's fully, like, you can't tell the difference, I think it's better to do the characters in an AI sense so everything still stays uniform.
Desired next-generation capabilities include:
Full-body performance capture beyond facial animation
Enhanced skin detail rendering in AI-generated characters
Improved camera control systems for AI-generated scenes
Higher quality facial performances matching tools like Veo 3
I keep making the joke and, and Runway, I love you, but like, please make Act-Two.
The Final Frame: Scaling AI Film Production
This three-person team proves that small studios can achieve feature-quality AI filmmaking when combining the right tools with traditional filmmaking expertise.
Without AI, I think, it'd actually be impossible. Our backgrounds are all very different. We all have these very distinct backgrounds that have brought us together.
The Phantom X model suggests a future where AI amplifies human creativity rather than replacing it, with union-approved frameworks protecting performer rights while enabling new forms of storytelling.
