IA Lab

AI experiments, AI+Art crossovers, and curated external articles about artificial intelligence and creative technology.

0 57 0 Comment on Seedance 2.5 — ByteDance Says the Stitching Era Is Over: One Unbroken 30-Second AI Video Shot, 50 References, and Edits That Don’t Nuke the Whole Clip

Seedance 2.5 — ByteDance Says the Stitching Era Is Over: One Unbroken 30-Second AI Video Shot, 50 References, and Edits That Don’t Nuke the Whole Clip

Every AI video model so far has been living a small lie. You ask for a 20-second…

0 84 0 Comment on ARDY — NVIDIA Just Gave Human Motion AI a Joystick: Real-Time, Promptable Character Animation You Steer While It Generates

ARDY — NVIDIA Just Gave Human Motion AI a Joystick: Real-Time, Promptable Character Animation You Steer While It Generates

For years, text-to-motion AI has been a beautiful lie for anyone building interactive worlds. You type a…

0 43 0 Comment on LichtFeld Studio — The Free Splat Studio That Trains, Edits and Renders Locally Just Went All-Vulkan, and Handed the Keys to an AI Agent

LichtFeld Studio — The Free Splat Studio That Trains, Edits and Renders Locally Just Went All-Vulkan, and Handed the Keys to an AI Agent

We have covered a dozen ways to get a Gaussian splat into your DCC of choice —…

0 56 0 Comment on Qwen-Image-Edit Just Made Character Consistency Free — And Handed 3D Artists the Missing First Step to Image-to-3D

Qwen-Image-Edit Just Made Character Consistency Free — And Handed 3D Artists the Missing First Step to Image-to-3D

For years, the single hardest problem in AI character art wasn’t making a pretty image — it…

0 77 0 Comment on VLM3 — Meta Just Argued You Don’t Need a Special 3D Model: A Plain 4B Chat Model Learns Depth, Correspondence and Camera Pose Natively

VLM3 — Meta Just Argued You Don’t Need a Special 3D Model: A Plain 4B Chat Model Learns Depth, Correspondence and Camera Pose Natively

Hook For two years the whole 3D-AI industry has been building bigger, weirder, more specialized machines —…

0 75 0 Comment on GSOPs — Gaussian Splats Just Walked On Stage: Lake Heckaman Made TouchDesigner Dissolve, Relight and Audio-React Real Captures in Real Time

GSOPs — Gaussian Splats Just Walked On Stage: Lake Heckaman Made TouchDesigner Dissolve, Relight and Audio-React Real Captures in Real Time

Hook For three years we’ve watched Gaussian splats colonize every offline tool on Earth — Blender, Cinema…

0 60 0 Comment on Marble Just Opened Its Doors — Fei-Fei Li’s World Model Will Now Speak Any Place Into Existence and Hand You the Splats

Marble Just Opened Its Doors — Fei-Fei Li’s World Model Will Now Speak Any Place Into Existence and Hand You the Splats

For two years we’ve covered every world model that dared to challenge it — Genie, Lyra, HunyuanWorld,…

0 45 0 Comment on Microsoft Lens — A 3.8B Image Model That Beats FLUX on a Fifth of the Compute, Because It Read Better Captions

Microsoft Lens — A 3.8B Image Model That Beats FLUX on a Fifth of the Compute, Because It Read Better Captions

Everyone in image-gen has been chasing the same dragon: more parameters, more GPUs, more compute. Microsoft Research…

0 40 0 Comment on 404—GEN — A Swarm of Competing Miners Will Build Your Gaussian Splat From a Text Prompt, and Drop It Into Unity for Free

404—GEN — A Swarm of Competing Miners Will Build Your Gaussian Splat From a Text Prompt, and Drop It Into Unity for Free

Type “a cyberpunk Shiba Inu bartender, neon visor, mixing drinks,” wait roughly ten seconds, and a textured,…

0 39 0 Comment on LTX Trainer — Lightricks Just Turned 13 Ways to Train a Video Model Into One YAML File, and Handed the Keyboard to Claude Code

LTX Trainer — Lightricks Just Turned 13 Ways to Train a Video Model Into One YAML File, and Handed the Keyboard to Claude Code

For two years the deal with AI video has been simple and a little insulting: you get…