Luma

Every image you have already made is still editable. Inside: where Extract Layers lives, pulling objects in from your board, editing text directly, and turning one design into a full set of sizes. https://lnkd.in/ddeXVzYk

What if your ads could learn? Luma and Dumbstruck are partnering to close the loop between advertising insight and execution. We’re calling it Creative Intelligence. Dumbstruck’s emotion analytics reveal exactly where an ad loses its audience and why. Luma uses those insights to remake the underperforming moments in the existing footage, without requiring a reshoot. The result: audience feedback becomes optimized, market-ready creative—not a one-time deliverable, but a continuously improving system. Better insights. Better creative. Better use of every media dollar. Learn more about Creative Intelligence from Luma and Dumbstruck: https://lnkd.in/esNaaWM6

FLUX 3 from Black Forest Labs is now available with Luma. It's BFL's first general model for image, video, audio, and action-prediction: creations truer to life, stylistically diverse, and deeply customizable. Video with native audio is live today, with image and fast variants coming soon. Generate with FLUX 3, then keep building in Luma Agents.

Everyone knows the math: ten generations, one usable spot. Producing a quality commercial video with AI has always meant prompting one shot at a time and paying for every miss. Luma Scenes changes that. Powered by Uni-1, Luma's creative intelligence model, Luma Scenes generates every keyframe together, in one context, ensuring temporal consistency from frame to frame. Input your references (brief, assets, prompt) and get back every scene of the spot as a perfectly sequenced, editable storyboard, with your logos, products, and end cards holding consistent across every frame. Refine any scene, time every beat, and approve the full sequence before anything renders. One wrong scene shouldn't cost the whole production. With Luma Scenes it won't. Luma Scenes, powered by Uni-1. Link in the comments.

Seedance 2.5 has landed on Luma. Bring up to 50 references across text, image, video, and audio. Generate multi-shot, high-definition videos up to 30 seconds long. Multiple scenes, camera angles, and pacing shifts in a single generation. Combined with the power of Luma Agents, it’s never been easier to produce your ad, campaign, or masterpiece. Join Luma and be the first to try it: https://lnkd.in/gc6u7Hs7

You direct the look. The motion. The sound. MiniMax H3 is now available in Luma Agents. Generate up to 15 seconds of 2K video with native stereo sound, guided by text, image, video, and audio references. More creative range. One continuous workflow with Luma.

Seedance 2.5 is coming to Luma. Be there when it lands.

Tired of regenerating images to change one detail? So were we. Introducing Layers, now live in Luma Agents. Powered by Uni-1, Layers gives precision editing and control over every element of any image, generated or uploaded. Every image comes apart into independent, editable layers. Objects. Text. Background. Change the one thing that's wrong. The rest stays exactly as designed. The layout you approved on day one is pixel-identical to layout 10, because untouched layers never go back through the model. Every edit is a first edit. Extract layers yourself, or speak with Luma Agents to decompose, edit, and recompose the image back together. Our own creative team used to export every image just to make final edits. They don't anymore. Start in Luma. Finish in Luma. Change one element. Preserve everything else. Link in comments. #LumaAI #Layers #CreativeOps

We're at AMD Advancing AI in San Francisco this week, with our team on the ground through July 23. Luma's mission is to build multimodal AGI that can generate, understand, and operate in the physical world. Getting there takes frontier infrastructure, and we're proud to build alongside our partners AMD and TensorWave Support, who power inference for every major model we ship, from Ray3.2 to Uni-1, inside Luma Agents. Come visit us at booth 420 and meet the team. #AdvancingAI2026 #AMD #AI