Netflix builds a GenAI animation studio

In partnership with

Welcome to VP Land! Studios are building in-house AI pipelines,
open-source developers are turning single images into fully-meshed 3D
worlds, and Google is getting into the AI screenwriting game.

In today’s edition:

Netflix quietly launches a GenAI-native animation studio
One image becomes a full 3D mesh world with objects and sound
DeepMind puts Google into AI screenwriting with Fabula
Mocap meets generative AI video in Blender

Netflix’s INKubator Brings GenAI-Native Animation Pipelines to Shorts and Specials

Netflix is building an internal studio called INKubator (also referred to as INK) that aims to produce short-form animated content using generative AI tools. The unit quietly launched in March under Serrena Iyer, whose credits span DreamWorks, MRC, and A24. Netflix is hiring producers, software engineers, and CG artists for the team.

GenAI-native from the ground up. Job listings describe INKubator as a "next-generation, creative-led, GenAI-native animation studio" aiming for "feature-quality content." The phrasing signals this is not a research lab bolted onto an existing pipeline but a production unit designed around AI-first workflows.
Starting with shorts and specials. Initial output targets animated shorts and specials rather than feature-length films, giving the unit room to develop production methodology before scaling to longer formats.
The hire signals intent. Iyer's background spans traditional animation studios (DreamWorks), independent finance (MRC), and prestige film (A24). That mix suggests INKubator is positioned to bridge AI production capability with the creative standards of theatrical-quality animation.
Netflix has been circling AI production for months. The company experimented with Go-with-the-Flow, an open-source AI animation tool, and CEO Ted Sarandos has framed AI's value as "making films 10% better, not just cheaper." INKubator is the first concrete production unit to emerge from that posture.

Read more for what the job listings reveal about Netflix's GenAI animation pipeline, why the Iyer hire matters, and how INKubator fits the Sarandos "10% better, not just cheaper" strategy.

SPONSOR MESSAGE

Voice dictation that doesn't mangle your syntax.

Most dictation tools choke on technical language. Wispr Flow doesn't. It understands code syntax, framework names, and developer jargon — so you can dictate directly into your IDE and send without fixing.

Use it everywhere: Cursor, VS Code, Warp, Slack, Linear, Notion, your browser. Flow sits at the system level, so there's nothing to install per app. Tap and talk.

Developers use Flow to write documentation 4x faster, give coding agents richer context, and respond to Slack without breaking focus. 89% of messages go out with zero edits. Free on Mac, Windows, and iPhone.

Try Wispr Flow free

Image-Blaster: Open-Source Toolkit Generates 3D Environments, SFX, and Meshes from a Single Image

A World Labs developer released image-blaster, an open-source toolkit that generates 3D environments, sound effects, and meshes from a single image. The pipeline combines Claude Code agents with World Labs Marble and fal.ai APIs (TripoSR, Trellis, Hyper3D Rodin) to produce editable scene files.

Full scene output. The toolkit does not stop at geometry. It produces 3D meshes, sound effects mapped to scene elements, and editable files that can be loaded into standard 3D software.
Agent-driven pipeline. Claude Code agents orchestrate the image-to-3D conversion and SFX generation automatically, reducing the manual steps between source image and finished scene.
Built on a proven model stack. The toolchain pulls from World Labs Marble (the spatial generation platform that launched earlier this year) plus fal.ai-hosted conversion models. World Labs originally demonstrated single-image-to-3D as a research capability; image-blaster packages it as a production-ready pipeline.

Feed in a reference still, get back a 3D scene with audio, ready for further work in Unreal or Blender. Read more here.

DeepMind Releases Fabula, a Gemini-Powered AI Writer Assistant

Google DeepMind launched Fabula, an interactive AI writing tool built specifically for fiction writers, with an emphasis on screenwriters and playwrights. The tool runs on Gemini and uses a two-layer structure: a story plan layer and a script layer, both editable independently.

Convergent iteration model. Writers refine AI output from broad narrative plans down to scene-level beats, with Fabula maintaining coherence when individual scenes or beats are modified. The system tracks narrative continuity so changes in one scene ripple correctly through dependent scenes.
Co-designed with working writers. DeepMind built Fabula with 42 professional writers during development, a larger user-design cohort than most AI writing tools can claim.
Story plan + linked script. The two-layer approach separates structure from dialogue. Writers can adjust act-level pacing without touching scene text, or revise dialogue while the narrative spine stays stable.

DeepMind's creative AI work has run through Darren Aronofsky's Ancestra at Tribeca and the 1776 series with SAG-AFTRA voice actors. Fabula extends that arc from production into the writer's room.

Read more for how the two-layer story-plan + script architecture works, what the 42-writer co-design cohort shaped, and where Fabula fits in DeepMind's broader creative AI program.

Disney's First Scooter Audio-Animatronic Was Motion-Captured From the Actual Muppet

Disney's Imagineers built a new Audio-Animatronics figure of
Scooter by motion-capturing the actual Muppet performance and translating it
into the figure's movements and expressions. The animatronic debuts in
the Rock 'n' Roller Coaster Starring The Muppets pre-show.

First Scooter animatronic ever built. Disney's Muppet character set hadn't included a physical Scooter figure until this build.
Mocap from the actual Muppet performance. Imagineers tracked Scooter's Muppet performance and translated the data into the animatronic's movement and expression range.
Puppetry data, robotic execution. The pipeline pairs traditional puppetry performance capture with robotic playback rather than scripting movement from scratch.

Read more on how the mocap-to-animatronic pipeline works and what the Rock 'n' Roller Coaster integration looks like.

Three AI Tool Updates: One-Image Video, Aesthetic Image Gen, AI Mouse Control

NVIDIA shipped a new model that turns a single still image and a camera path into a full minute of 720p video on one consumer GPU. The model is called SANA-WM, built on the SANA 2.6B architecture. It follows precise 6-DoF camera moves; NVIDIA's own demo rendered a 60-second clip in 34 seconds on an RTX 5090. The output targets cinematic fly-throughs and spatial walkthroughs. NVIDIA's Lyra 2 platform handles related walkable-world generation; SANA-WM is the rendering speed and camera-control piece.

Krea released a sharper image foundation model focused on aesthetic control and style consistency. The new model, Krea 2, is the follow-up to Krea 1 and supports style references and custom training across artistic styles. It sits alongside Krea's open-sourced realtime video model. Read more on what the new aesthetics tuning changes vs. Krea 1.

Google DeepMind demoed an AI agent that drives the desktop mouse pointer itself, navigating GUIs, clicking buttons, and operating apps autonomously. The work is part of DeepMind's broader agents program and points directly at AI driving post-production and editorial software without a human in the loop. Read more on how the pointer system works and what it could mean for editorial workflows.

Cinematography Database breaks down a pre-vis workflow that runs Rokoko SmartGloves II mocap through Blender into Seedance 2.0 and Kling 3.0 for AI video.

Stories, projects, and links that caught our attention from around the web:

🎬 Ex-Prime Video UK chief Chris Bird launched CineMe AI and HawksHead AI at Cannes, pairing scripts-to-storyboards with audience-prediction.

👩🏻‍💻 Imaginario's StoryLab assembles a rough cut in under a minute using three AI agents driven by a text prompt.

🎬 House of David director Jon Erwin shot "The Old Stories: Moses" in one week using AI, starring Ben Kingsley.

💰 Kuaishou is reportedly evaluating a Kling AI spinoff, with pre-IPO talks for a potential $20B valuation and a possible 2027 Hong Kong listing. Board review is underway but not finalized.

🎥 Door G uses AI in pre-production and finishing, not on the LED wall, keeping the stage reserved for live camera capture.