Fixing it in post since 1495

Top Stories - Overview

It's been a good week if you're someone who makes things as Blackmagic shipped the final release of DaVinci Resolve 21 with eight new AI tools that address real daily friction, Stability AI dropped a four-model audio family with open weights and clean commercial licensing, and Clear Angle Studios and Union VFX went public with a relightable Gaussian splat pipeline that moves the technology from demo to production.

Meanwhile, Lionsgate deepened its Runway relationship with an equity stake, the Art Directors Guild fired back at Martin Scorsese over his Black Forest Labs partnership, and Amazon's high-profile GenAI animation push collapsed within 48 hours of its own announcement.

1. DaVinci Resolve 21 Ships Final Release With Eight New AI Tools and Major Workflow Expansion
Newsshooter | CineD

What Happened: Blackmagic Design has shipped the final release of DaVinci Resolve 21, adding eight new AI-powered tools alongside 300-plus new features across the platform. Headliners include IntelliSearch for media pool searching by face, object, or dialogue keyword; CineFocus for simulated rack focus on existing footage; a Speech Generator that can clone a voice from a 10-second clip; and a Face Age Transformer. A new Photo page brings the full node-based color pipeline to still images for the first time.

Why Is This Important? Several of these tools go straight at problems editors and colorists deal with every day - finding a specific face in a large media pool, creating a rack focus effect on locked-off footage, maintaining a consistent voice when a pickup line doesn't match. CineFocus alone could save a finishing session. The free version of Resolve retains meaningful AI features, which means this isn't an upgrade conversation only for Studio license holders. If you haven't moved to 21 yet, the feature list makes a strong case.

Tags: #AITools #PostProduction #ColorGrading #PostWorkflow #DaVinciResolve

2. Stability AI Drops Stable Audio 3.0: Open Weights, Six-Minute Tracks, Licensed for Commercial Use
Digital Music News | The Decoder

What Happened: Stability AI launched Stable Audio 3.0, a family of four models generating tracks up to six minutes and twenty seconds - up from 47 seconds in the previous open release. Three of the four models ship with open weights, meaning they're free to download and run locally on consumer hardware. Importantly, all four are trained exclusively on licensed data. The Large model, the highest-quality variant, is API and enterprise only, with a commercial license requirement for companies above $1 million in annual revenue.

Why Is This Important? For sound designers and composers working on spots or short-form content, two things matter here: the track length finally clears a practical minimum for music beds and underscore work, and the licensed training data means you're not carrying copyright risk into a client deliverable. The open-weight Small model running fully on-device is also worth noting - offline generation without clip length limits is a meaningful workflow option for anyone working on projects where cloud tools aren't practical or allowed.

Pro Tip: The Large model's enterprise licensing terms are worth reading...

Tags: #AIAudio #AIMusic #SoundDesign #PostWorkflow #StabilityAI

3. Ideogram 4.0: First Open-Weight Image Model With Layout Control and Best-in-Class Text Rendering
Variety | The Decoder

What Happened: Ideogram released version 4.0 on June 3, its first open-weight model - a 9.3 billion-parameter Diffusion Transformer that generates at native 2K resolution and accepts structured JSON prompts with bounding-box coordinates and color palette specs. It ranks first among open-weight models on Design Arena and claims the highest in-image text rendering accuracy of any model at its parameter size, scoring 0.97 on the X-Omni English OCR benchmark. Weights are downloadable and run on a single 24GB GPU.

Why Is This Important? The text rendering benchmark is the headline for motion designers and title artists - if it holds up in production, it significantly reduces the cleanup pass that most generative image tools still require for any work involving legible type. The JSON layout control is the other angle worth testing: bounding-box placement and palette conditioning turns a one-off generation into something closer to a repeatable template, which has real value for anyone building comp mockups or style frames at volume. Non-commercial use is free; commercial work requires a paid plan or API license.

Tags: #AITools #GenerativeAI #MotionGraphics #ImageGeneration #Ideogram

4. Clear Angle Studios and Union VFX Launch Relightable 4D Gaussian Splat Pipeline for Feature Film Crowds
postPerspective | fxguide

What Happened: Clear Angle Studios has launched a modular Volumetric Capture Rig using 40 synchronized machine-learning cameras in a half-dome configuration, designed for dynamic 4D Gaussian splat capture. In partnership with London's Union VFX, the companies developed a relightable crowd pipeline for an upcoming feature film - capturing performers in groups and converting them into lightweight splat assets that can be relit in post to match any scene. Clear Angle has previously done capture work for Disney, Universal, and Netflix.

Why Is This Important? Relighting is the problem that has kept Gaussian splatting out of serious production use - captured lighting is baked in, and a splat that can't be matched to a plate is a splat you can't use. Union's solution to that changes the equation. For VFX supervisors and the people designing how shots actually get built, the crowd application is the proof-of-concept that matters: if you can capture a group of performers, relight them in post, vary costumes, and integrate them into a CG scene at a fraction of the geometry reconstruction cost, that's a workflow conversation worth having early in pre-production.

Tags: #VFX #PostProduction #GenerativeAI #GaussianSplatting #UnionVFX

5. SXSW London Panel: AI for Indie Filmmakers Spotlights Autodesk Flow Studio for CG Character Work
postPerspective

What Happened: At SXSW London's "Hollywood VFX, Indie Budget" panel on June 11, filmmakers discussed practical AI tools for lower-budget productions. A central focus was Autodesk Flow Studio - formerly Wonder Dynamics - which uses AI to replace live-action performers with CG characters while preserving performance and camera movement. Panelists framed the tool as designed to work alongside traditional VFX workflows, not replace artists, noting it supports both live-action and animation workflows.

Why Is This Important? Autodesk Flow Studio has been around long enough to have a production track record, and the indie budget context is the right one for it - this is the tool that lets a small team do character replacement work that would otherwise require a much larger VFX house. For editors and VFX supervisors on mid-range productions, if you're not already using it, you might want to take a look. The "artist-in-control" language from the panel also reflects where people are landing on AI tools right now, as it's becoming less about whether to use them, more about who stays in the driver's seat when you do.

Tags: #VFX #FilmProduction #PostWorkflow #AITools #AutodeskFlow

6. Lionsgate Takes Equity Stake in Runway, Will Mine IP for AI-Generated Shorts
Variety | TheWrap

What Happened: Lionsgate has taken an equity stake in Runway and announced a joint development program to produce AI-generated short-form episodic series using existing Lionsgate IP. The companies will also launch a new initiative to co-develop original IP using generative AI tools, with a slate of projects planned. Lionsgate first partnered with Runway in 2024 but the original deal ran into limitations around model capabilities and rights questions. No release timeline or specific franchises have been named.

Why Is This Important? This isn't just a licensing deal anymore - Lionsgate now has a financial stake in Runway's valuation, currently sitting at $5.3 billion. That alignment changes the nature of their relationship and signals that at least one major studio has decided generative AI video is worth owning a piece of, not just renting access to. For those of you working in or around studio production pipelines, this worth watching because studios investing equity (i.e. real dollars) in AI platforms is a different kind of commitment than a simple technology partnership.

Tags: #FilmProduction #VideoGeneration #BusinessFunding #Runway #Lionsgate

7. Art Directors Guild Slams Scorsese for "Turning His Back on Human Artists" in AI Partnership
Variety | The Hollywood Reporter

What Happened: The Art Directors Guild, IATSE Local 800, issued a public statement on June 9 condemning Martin Scorsese's advisor role with AI startup Black Forest Labs. The guild accused Scorsese of "turning his back on the human artists who throughout his career have helped him create his most memorable works," arguing that his endorsement of FLUX for storyboarding and pre-vis "circumvents the input" of art directors, production designers, illustrators, scenic artists, and set designers. Scorsese, for his part, said the tool helps him communicate his vision to his creative team more clearly and efficiently.

Why Is This Important? The ADG statement names the specific crafts it sees as threatened - art directors, illustrators, storyboard artists, production designers and that specificity is what gives it weight. This isn't a general union position on AI; it's a direct call-out aimed at the person many in those crafts consider a patron of their work. For anyone in pre-production or visual development, this is the clearest articulation yet of where the labor fault line sits on generative image tools.

Tags: #FilmProduction #GenerativeAI #AITools #UnionLabor #BlackForestLabs

8. Amazon's GenAI Animation Push and the "Punky Duck" Collapse
Deadline | IndieWire

What Happened: On May 27, Amazon MGM Studios and AWS launched the GenAI Creators' Fund and greenlit three AI-animated series for Prime Video, powered by an in-house platform called Project Nara. Within 48 hours, the highest-profile greenlit director, Jorge R. Gutierrez - known for "The Book of Life" and "Maya and the Three" - scrapped his project "Punky Duck" after intense backlash from peers and fans over his embrace of AI tools, having previously spoken out against AI in creative work.

Why Is This Important? The full arc is the story - a major studio makes a loud bet on AI animation, names a respected director as its marquee partner, and that director reverses course in two days under public pressure. It's a story about how fast the cultural temperature around AI can turn, even when the people involved have genuine creative intent. For studios and production companies weighing how to publicly position their own AI work, the Punky Duck episode is an instructive case study.

Tags: #FilmProduction #VideoGeneration #GenerativeAI #Animation #Amazon

General AI News

  • Anthropic Releases Claude Fable 5, First Public Mythos-Class Model - The most capable Anthropic model yet went public on June 9 with hard safety guardrails blocking high-risk outputs in cybersecurity, biology, and chemistry; priced at 2x Claude Opus 4.8. CNBC

  • Apple Rebuilds Siri From Scratch With Google Gemini at WWDC 2026 - Apple confirmed a ground-up Siri rebuild on a 1.2 trillion-parameter Gemini model at WWDC on June 8, reportedly worth $1 billion per year; the redesigned assistant is multimodal, context-aware, and ships with iOS 27 in the Fall. TechTimes

  • Utopai Studios Launches PAI 2.0: Agent-First Cinematic Storytelling Platform - PAI 2.0 guides users through planning, generating, and refining long-form video with narrative continuity across shots; the platform has reached $11M ARR since its April debut. Variety

  • OpenAI Launches Dreaming V3: ChatGPT Gets Autonomous Memory Architecture - A new background synthesis system replaces ChatGPT's saved-memories list, automatically building context from a user's full conversation history; rolling out to Plus and Pro users in the US first. iClarified

  • Midjourney V8.1 Becomes the Default Model - As of June 10, V8.1 is now the default for all Midjourney users, bringing native 2K output, 4-5x faster generation than previous versions, and improved prompt adherence, no opt-in required. Midjourney Docs

  • Runway Gen-4 Launches With Native Audio and 60-Second Continuous Generation - Runway's Gen-4 model adds synchronized audio generation alongside video, extended clip length, and improved social-format output; available across subscription tiers. TheWrap

  • Adobe Stock Now Accepts AI-Generated Video Submissions - Adobe has published dedicated generative AI video submission guidelines for contributors, a policy shift from prior years when only AI still images were accepted. Adobe Stock Contributor Help

Keep Reading