Creative Workflow Roundup: Runway 4K Upscaling, Veo 2 Physics, ChatGPT LUTs, and Generative AI Breakthroughs in 2025
A Note for Subscribers: If you’ve been curious about character animation workflows, my new video tutorial is live where I show you how to create character animations using Runway’s Act-One tool. This step-by-step guide walks you through crafting a 10-second animated film featuring multiple characters, each with unique voices generated using Eleven Labs. To bring it all together, I’ve also incorporated b-roll elements created with OpenAI’s Sora. Watch the full tutorial here.
Welcome to the first Creative Workflow Roundup of 2025! Here’s what I’m covering this week: Runway’s brand-new 4K upscaling feature is making production-ready outputs easier than ever, while experiments with Google’s Veo 2 are showing off physics so realistic it’s hard to believe they’re AI-generated. Danish artist Martin Nebelong is redefining 3D rendering workflows by combining PS5 models with Video-to-Video AI tools like Sora, and ChatGPT is finding surprising new use cases in color grading with custom LUT generation. Meanwhile, ByteDance has released impressive research like AnyDressing and LatentSync, even as TikTok’s fate in the U.S. hangs in the balance.
Runway Gen-3 Alpha: 4K Upscaling Brings AI Video Closer to Production Quality
The News: Runway has rolled out a major update to its Gen-3 Alpha video generation platform: the ability to upscale outputs to 4K resolution. This new feature, available now for all paid plans, simplifies workflows by making it possible to create high-resolution, production-ready outputs within the platform. The update is another step forward for Runway, which continues to push the boundaries of text-to-video AI tools. This addition could make Runway more competitive in an increasingly crowded field of AI video tools.
Lab Notes: I tested the new 4K feature, and while it does deliver on resolution, it also brings some trade-offs. Here’s one of my exports from Runway: