Runway Releases Gen-4 Video Model, Blows Up Reddit, Other Socials. Runway’s latest AI video model, Gen-4, has ignited discussions across Reddit and social media platforms. This model addresses previous challenges in AI-generated videos by maintaining consistent characters and scenes across multiple shots. Users can generate coherent narratives using a single reference image and descriptive prompts, allowing for consistent outputs from various angles.
Runway’s dramatic improvements this week follow on the heels of OpenAI’s ChatGPT Image Generator. Curious Refuge says the upgraded ChatGPT image generator may be the best in the business right now. Dall-E was integrated into OpenAI’s ChatGPT some time ago. It is a very idiosyncratic image generator, with a distinct style that made everything seem generically digital. This upgrade changed all of that. It handles cinematic styles, and the words are no longer gibberish, making it much more useful for design. Available to free and paid users alike, this made me think OpenAI’s goal is to enable me to prompt a finished video, with music, dialog, and sound effects.
Higgsfield Launches to Bring Cinematic Camera Moves to AI Video. Alex Mashrabov, former head of AI at Snap, has launched Higgsfield AI to give generative video tools the language of cinema. Unlike many rivals focused on style and fidelity, Higgsfield brings intent to AI-generated visuals by enabling complex camera techniques—like crash zooms, dolly shots, and body-mounted rigs—via simple text prompts. The platform builds on Mashrabov’s viral app Diffuse, pivoting from short-form gags to structured visual storytelling for TikTok and YouTube Shorts. With presets mimicking pro cinematography and endorsements from creators like Jason Zada and John Gaeta, Higgsfield could reshape AI-native filmmaking.
Pika’s “Selfie With Your Younger Self” Pika introduced a new feature that lets users create realistic videos in which they appear to interact with a younger version of themselves. The AI stitches together a recent video with a childhood photo, resulting in a stylized, emotionally charged clip. This viral feature has sparked both delight and discomfort across social platforms. While it’s free, premium subscribers enjoy faster render times. Pika continues to push the boundaries of AI video with tools that blend nostalgia, realism, and personalization.
Udio Announces The Launch of New “Styles” and “v1.5 Allegro” Features. Udio has introduced Styles, a new feature that lets users generate original music inspired by the sonic DNA of any uploaded track. Whether it’s a rough demo or finished piece, Styles recreates the instrumentation, tone, and vibe of the source, enabling cohesive sound across albums, ads, films, or podcasts. To prevent misuse, uploads must be user-owned or licensed. Styles debuts alongside version 1.5 “Allegro,” a major update that cuts generation time by up to 30% without quality loss. The new tools mark a leap forward for AI-assisted music production and creative control.
Prodia Claims Title as World’s Fastest AI Image Generator. Atlanta-based startup Prodia has launched what it says is the world’s fastest AI image generation platform, capable of producing visuals—logos, artwork, product images—in under 300 milliseconds. Benchmarked by Artificial Analysis, the platform dramatically outpaces larger competitors, thanks to custom GPU optimization and a ground-up rebuild focused on speed. Prodia’s plug-and-play API lets developers integrate real-time image generation into apps, games, and marketing tools without compromising performance. The milestone marks a major win for Atlanta’s growing tech scene and sets a new standard for real-time AI infrastructure in creative and commercial applications.
Tim from Theoretically Media sent this with a note. “Just wanted to pop over my latest AI Short film to you. It’s been a minute since I got a chance to flex out…stunningly, almost 7 months since Dead Sea. So I was long overdue! This one was made with Veo-2, but with some extra lip sync spank with Hedra!”
I could not do my job without Caleb Ward’s weekly Curious Refuge roundup of Cinematic AI tools. Without the work of Caleb and his staff, there would be no way to keep up. And it is free.
This column is the script for the news portion of the author’s weekly podcast co-hosted by former Paramount futurist and co-founder of Red Camera, and Rony Abovitz, founder of Magic Leap, Mako Robotics, and Synthbee AI. This week our guest is Patty Rangel, Producer, Impact Investor, and Futurist. We can be found on Spotify, iTunes, and YouTube.
What We’re Reading
