Netflix's Use of AI at Eyeline VFX Studio Revealed by Mohsen Mousavi at View Conference

Netflix-owned Eyeline studio uses generative AI to replace locations, reconstruct crowds, and alter lighting after filming, while developing tools that could even let directors reposition the camera after an actor's performance is captured. Mohsen Mousavi, deputy head of Eyeline's virtual department, said: "We are not pointing our path toward shots and films. We are not that."

The Netflix-owned studio used composite production techniques on "Brazil 70: The Third Star" and "The American Experiment," while developing tools to change lighting, locations and camera angles.

Netflix-owned studio Eyeline uses generative AI to replace locations, recreate crowds and change lighting after filming, while developing tools that can even allow directors to reposition the camera after an actor's performance has been recorded.

Moussavi, whose credits include the final season of Game of Thrones, argued that producing a convincing portrayal of artificial intelligence is only the beginning. He suggested that production models produce roughly "60, 70 percent," but "getting to 100 percent is the most important thing." He was describing the final difficult stretch between promising output and film that meets the consistency, clarity, and creative control needs of professional post-production rather than a formal performance measure. Popular in Variety

The eye-line approach starts before the cameras start rolling. The generative art section explores those designs alongside traditional artists, while the "sparse set" workflows allow performers to work on partial physical sets under relatively flat lighting. The environment and lighting can then be recreated and the original performances of the actors preserved. "This is an extension of the traditional art sector, not a replacement," Mousavi said.

An associated volume recording system allows camera perspective and lighting to be reviewed after filming. At the demonstration, Eyeline showed footage shot in one location and transformed into another, including environments created without relying on green screens. Emphasizing the importance of preserving the actors' identity and performance, Mousavi said: "This gives us the freedom of light and the freedom of the camera after filming the performance."

The studio has already applied parts of the approach to Netflix productions. In "Brazil 70: The Third Star," a drama about the national team's 1970 World Cup triumph, Eyeline helped create a massive, emotionally responsive stadium crowd to recreate the tournament. Rather than relying solely on AI, the process combined filmed elements, sprites, conventional CG, and generative techniques. Mousavi said that it would be "very, very difficult on that budget" to achieve these results through traditional VFX methods.

For Netflix's five-part historical documentary The American Experiment, he integrated backdrops, generator dust and other digitally generated elements with action footage. Eyeliner also showed techniques for altering the appearance of a performer while maintaining the original role, as well as a short investigation into post-shoot camera and lighting setups.

Among the technologies still under development are controllable production fluids. Current generation simulations are unreliable, Mousavi admitted. "They are not decisive, and they are very difficult to control," he said. Eyeline requires artists to adjust the speed, viscosity, and turbulence of water with keyframes and add foam or spray without changing the underlying simulation. "Often in visual effects, what you really want to do is you want to break fluid physics to achieve a certain creative effect," he explained.

Other demonstrations included reconstructing a 3D Gaussian image from winter to autumn and re-illuminating human performances with the help of artificial intelligence. A system uses 40 synchronized RED V-Raptor cameras to collect training data. Eyeline has also developed more portable photography equipment. Mousavi also presented research on creating a navigable 4D scene from a single camera recording. Such experiments, he emphasized, are distinct from techniques already employed in completed demonstrations.

Eyeline systems rely heavily on custom work rather than off-the-shelf generators. "Most of what we do is based on our own research and models," he emphasized. The studio also uses external contributions and specially recorded training data, although Mousavi did not identify the providers of its base model.

The presentation comes as Netflix separately expands its AI ambitions, including the purchase of Ben Affleck's filmmaking technology company InterPositive. The streamer said in July that approximately 300 titles are already using some form of generative artificial intelligence.

For Mousavi, technical ability alone is not enough. "We want to make sure it's not a black box," he added, stressing that cinematographers and production designers should be able to shape the results. High dynamic range, 4K rendering and temporal compatibility remain significant challenges, even as Eyeline advances its solutions. "The goal is to make it completely cohesive, regardless of whether the final image originates from the camera, traditional VFX or generative technology," he concluded.