VP Land
Posts
What the Academy Thinks About AI

What the Academy Thinks About AI

Adobe Adds More AI Models, Veo 2 Tricks

April 25, 2025

Powered by

Happy April 25th - hope you brought a light jacket.

Shout-out to this person on Reddit who has ranked all 295 running moments from Tom Cruise’s filmography.

The best commentary: “Collateral not only features the best sitting performance of Tom Cruise’s career, it also features his most aggressive running.”

And another reminder - if you like the newsletter (and podcast and YouTube channel), consider becoming a member.

Joey

AI Gets Oscar Nod

The Academy of Motion Picture Arts and Sciences has officially addressed the role of generative AI in filmmaking with new rules for the 98th Oscars.

The key takeaway: AI tools "neither help nor harm" a film's chances at nomination, but human creative authorship remains paramount in the Academy's judgment. This marks the first time the organization has formally acknowledged AI's growing presence in Hollywood production.

From the Rulebook: "With regard to Generative Artificial Intelligence and other digital tools used in the making of the film, the tools neither help nor harm the chances of achieving a nomination. The Academy and each branch will judge the achievement, taking into account the degree to which a human was at the heart of the creative authorship when choosing which movie to award.”
Industry Context: This stance arrives less than two years after AI protections became central to the 2023 WGA and SAG-AFTRA strikes.
Recent Precedents: The ruling follows controversies around films like The Brutalist, which used AI voice cloning to perfect Adrien Brody's Hungarian accent—a film that ultimately won Brody the Best Actor Oscar despite initial controversy.
Implementation: The Academy considered but ultimately decided against requiring filmmakers to disclose AI usage, opting instead for a judgment-based approach that evaluates the degree of human creative input.

And in a ‘wait, this wasn’t a requirement already?’ change, Academy members must now confirm they've watched all nominated films in a category before voting.

SPONSOR MESSAGE

Atomos Sumo: The Ultimate Monitor-Recorder

Experience the ultimate on-set solution with the Atomos Sumo 19SE – now available at an incredible value.

This production powerhouse delivers professional-grade monitoring, recording, and switching capabilities that would typically cost thousands more, all in one robust package that will transform your workflow without breaking the bank.

Premium Display Quality: Features a daylight-viewable 19-inch 1920x1080 touchscreen with stunning 1200-nit brightness and 10-bit color depth, covering the DCI-P3 color gamut for perfect visibility even in direct sunlight.
Professional Recording Capabilities: Capture content in multiple high-end formats, including ProRes RAW (up to 6Kp30), ProRes 422, and DNxHR up to 4K60p, with multi-channel recording for complete production flexibility.
Complete Connectivity: Equipped with four 12G-SDI inputs, HDMI 2.0 connectivity, signal cross-conversion, and professional audio support including dual XLR inputs with +48V phantom power.
Advanced Production Tools: Access essential monitoring features including waveform, vectorscope, false color, focus peaking, anamorphic de-squeeze, and comprehensive LUT support for both SDR and HDR workflows – tools typically found only in much more expensive systems.
Trusted Across The Industry: Used on hundreds of productions, from the latest season of The Last of Us to capturing hundreds of camera feeds for MrBeast videos.

Check out the Sumo - Only 1995 USD/EUR

Firefly Opens Doors to Third-Party AI Models

Adobe has dramatically expanded its Firefly AI platform, now integrating third-party AI models alongside its own commercially safe offerings.

The update brings unprecedented creative flexibility through multi-model access, expanded video generation capabilities, and improved image creation—all while maintaining Adobe's commitment to transparent AI use.

Adobe announced at MAX London that Firefly now incorporates partner models from Google Cloud (Imagen3 and Veo2), OpenAI, and Black Forest Labs (Flux 1.1 Pro), with more partnerships planned with fal.ai, Runway, Pika, Luma, and Ideogram
The Firefly Video Model, previously in beta, is now generally available, enabling creators to generate video content from text prompts and images with enhanced creative control
New Firefly Image Model 4 and Image Model 4 Ultra deliver improved realism, detail and complexity for still image generation
The platform introduces Firefly Boards (previously Project Concept), providing an AI-first collaborative surface for moodboarding, storyboarding, and concept exploration
Adobe maintains its commitment to transparency by automatically attaching Content Credentials to all AI-generated content, clearly indicating whether content was created with an Adobe Firefly model or a partner model

Adobe's vision extends beyond the Firefly app, with plans to integrate third-party AI models directly into applications like Premiere Pro. At NAB 2024, Adobe previewed how video editors could leverage models from OpenAI, Runway, and Pika Labs within Premiere Pro's familiar workflow.

As Meagan Keane explained in her VP Land interview, this approach will allow editors to access specialized AI capabilities without leaving their preferred editing environment, potentially transforming how professionals approach video production tasks like generating B-roll or extending shots.

Veo 2 Motion Tips

Creating compelling motion in Veo 2 videos requires strategic prompt engineering. Follow these key techniques to significantly improve your AI-generated video quality.

Use descriptive motion language with specific action verbs like "striding confidently" instead of generic terms like "walking" to guide the AI in creating more natural movement
Specify camera movements such as "tracking shot," "slow pan," or "gradual zoom" to control how the scene unfolds visually and create more dynamic content
Include environmental interactions by describing how subjects interact with their surroundings (e.g., "leaves rustling," "fabric responding to wind") for more grounded, realistic motion
Add timing instructions like "quick, sudden movements" or "slow, deliberate gestures" to control the pace and rhythm of the generated video
Consider aspect ratio when crafting prompts - use 16:9 for cinematic landscapes and 9:16 for vertical content like waterfalls or tall buildings

John Knoll’s motion control system bridges classic filmmaking techniques with modern technology, showing how old-school methods still have a vital role today.

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

🕺 Seaweed, ByteDance's video generation model, stands out for creating realistic videos with diverse actions and emotions

🎥 Edelkrone's Krone X integrates DollyPLUS, JibPLUS/JibONE, and HeadPLUS into a 6+1 axis system for synchronized motion control without programming

🎬 GameChanger's virtual production tech ditches green screens and physical sets, enabling speedy, budget-friendly TV show creation with big budget-level visuals

🔊 Runway ML CEO Cristóbal Valenzuela demos Gen-4's audio-driven interfaces for immersive world-building, offering filmmakers AI-powered tools to craft dynamic, responsive environments in real time

🤑 Essay on how the new economy rewards AI-powered solo creators over large teams, reshaping content creation hierarchies

Addy and Joey dive into Netflix's Ted Sarandos' AI strategy for filmmaking, audience perceptions of AI-generated stories, and Ryan Coogler's groundbreaking deal for copyright ownership of Sinners.

Read the show notes or watch the full episode.

Watch/Listen & Subscribe

Spotify | Apple Podcast | YouTube