• VP Land
  • Posts
  • Netflix Sneaks into the AI Model Game

Netflix Sneaks into the AI Model Game

Spline 3D, AI Storyboarding, and more

January 25, 2025

In partnership with

Earlier this week, we covered The Brutalist AI controversy over using Respeecher for some dialogue clean-up (and Midjourney for some mood board generation). I thought the hoopla was dumb.

Well, we go deeper into how dumb I thought it was and the larger issues of mentioning the term ‘AI’ in general and all the baggage it comes with on the latest episode of the podcast, where I’m joined by co-host Addy Ghani.

We also talked about the LA wildfires (and the rise of fake AI footage on social media) and what we saw at CES.

In this issue, we’ve got another mountain of AI updates, from more 3D developments to a surprise paper from Netflix.

Let’s get into it!

Joey

Transforming Images into 3D Scenes with Spell

Spline has unveiled Spell, an AI model that generates 3D worlds from a single image input.

This innovative tool creates consistent, multi-view 3D scenes in minutes, representing a significant advancement in AI-driven graphics for the film and media production industry.

Under The Spell

  • Spell uses a diffusion model to generate 3D worlds across various categories, including people, objects, and environments.

  • The model renders images from multiple angles with high accuracy and can generate controlled camera paths.

  • It simulates physical material properties like reflections and refractions, as well as camera properties such as depth of field.

  • Spell prioritizes physical consistency, simulating real camera interactions with objects rather than using visual interpolation.

  • The AI was trained on a combination of real-world data captured manually and synthetic data rendered using proprietary techniques.

  • Output formats include video, image sequences, and volumetric data (Gaussian Splatting).

SPONSOR MESSAGE

Take on 2025 with $10,000

You’ve got big goals for 2025. And HoneyBook has the tools, resources, and bonus prizes to take you there. Join HoneyBook in January for a chance to win.

FilmAgent: AI Storyboarding and Prep Tool

FilmAgent emerges as a new AI-powered framework for automating film production in virtual 3D spaces. This multi-agent system simulates key crew roles and integrates human workflows to create coherent, physics-compliant videos with strong storytelling capabilities.

Beat Sheet

  • FilmAgent covers major production stages: idea development, scriptwriting, and cinematography.

  • The system employs multiple AI agents to simulate roles like director, screenwriter, actor, and cinematographer.

  • It uses two collaboration strategies: Critique-Correct-Verify and Debate-Judge, to refine scripts and camera settings.

  • Human evaluations rate FilmAgent-produced videos at 3.98 out of 5 on average, outperforming single-agent baselines.

  • Unlike OpenAI's Sora, which struggles with consistency and narrative delivery, FilmAgent produces more coherent stories but requires pre-built 3D spaces.

Go-with-the-Flow: Netflix Gets in the AI Model Game

Eyeline Studios, powered by Netflix, has introduced Go-with-the-Flow, a new technique for controlling motion patterns in video diffusion models.

This innovation allows users to manipulate camera and object movements within a scene and even transfer motion patterns between videos, offering enhanced creative control in video production.

Behind the Scenes

  • Go-with-the-Flow works with both image-to-video and text-to-video models, capable of deriving 3D scenes from motion information alone.

  • The method enables various motion control types, including cut-and-drag animations and first-frame editing.

  • The technique fine-tunes a base model using warped noise instead of pure i.i.d. Gaussian noise, maintaining the same computational cost.

  • Users can adjust motion control strength through "noise degradation" at inference-time.

  • It can transfer motion patterns between videos, including 3D-rendered turntable camera motions and DAVIS dataset motions.

  • The technique allows for advanced camera control applications, such as creating coherent 3D scenes from a single image using monocular depth estimation.

  • Go-with-the-Flow improves temporal consistency in image-to-image translation tasks like relighting and super-resolution, without requiring additional training.

The Latest in AI

Check out Mrwhosetheboss’ studio setup, featuring a virtual production stage using Vu One and Vu Studio at 13:48

👨🏻‍💻 OpenAI's Operator, an AI agent capable of performing web-based tasks autonomously, may reshape how media professionals approach routine online activities.

🏞️ Gaussian splatting could be the key to building photorealistic 3D worlds for future AR/VR holodeck experience.

🥽 Samsung's Project Moohan, an Android XR headset developed in partnership with Google, makes its public debut at Galaxy Unpacked.

🖥️ Asus is launching a 32-inch 6K monitor with similar pixel density to Apple's ProDisplay XDR at a significantly lower price point of $1200.

🦁 Cutting-edge tech brought Disney’s Mufasa to life with stunning realism. Discover how innovative tools shaped the future of VFX magic.

👔 Virtual Production Gigs

Technical Program Manager
Mo-Sys Engineering Ltd
London, UK

Internship - Virtual Production
Orbital Studios
Los Angeles, CA

📆 Upcoming Events

January 28
Silicon Valley Video Summit
Mountain View, CA

February 16 to 20
HPA Tech Retreat 2025
Rancho Mirage, CA

March 7 to 8
Cine Gear NY 2025
New York, NY

March 7 to 15 🆕
2025 SXSW Conference & Festivals
Austin, TX

April 6 to 9
NAB Show Las Vegas
Las Vegas, NV

View the full event calendar and submit your own events here.

Thanks for reading VP Land!

Have a link to share or a story idea? Send it here.

Interested in reaching media industry professionals? Advertise with us.

Reply

or to participate.