🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

TikTok presents Depth Anything Unleashing the Power of Large-Scale Unlabeled Data paper page: huggingface.co/papers/2401.10… demo: huggingface.co/spaces/LiheYou… Depth Anything is trained on 1.5M labeled images and 62M+ unlabeled images jointly, providing the most capable Monocular Depth

Introducing Lumiere 📽️ The new video diffusion model we've been working on Google AI * Text-to-Video * Image-to-Video * Stylized Generation * Inpainting * Cinemagraphs and more 🎨 W/ amazing team incl. Hila Chefer omer tov Inbar Mosseri Tali Dekel Deqing Sun Oliver Wang

Seeing the project we’ve been working so hard on getting well received by the community is so fulfilling! Omer Bar Tal Hila Chefer Charles Herrmann Roni Paiss Shiran Zada Ariel Ephrat Junhwa Hur Yuanzhen Li, Tomer Michaeli Oliver Wang Deqing Sun Tali Dekel Inbar Mosseri

Introducing✨Still-Moving✨—our work from Google DeepMind that lets you apply *any* image customization method to video models🎥 Personalization (DreamBooth)🐶stylization (StyleDrop) 🎨 ControlNet🖼️—ALL in one method! Plus… you can control the amount of generated motion🏃‍♀️ 🧵👇

We are excited to introduce "VidPanos: Generative Panoramic Videos from Casual Panning Videos" VidPanos converts phone-captured panning videos into (fully playing) video panoramas, instead of the usual (static) image panoramas. Website: vidpanos.github.io Paper:

It's time to share something exciting that we've been working on at Google. ReCapture is a method to generate new versions of a user-provided video with customized camera trajectories. It's basically post-capture cinematography for all. Project links are below. 🧵

VideoJAM is our new framework for improved motion generation from AI at Meta We show that video generators struggle with motion because the training objective favors appearance over dynamics. VideoJAM directly adresses this **without any extra data or scaling** 👇🧵

Excited to share our #SIGGRAPH2025 paper: #TokenVerse 🎉 A new way to personalize text-to-image models by combining objects, styles, poses, lighting — pulling each concept from a different image and blending them into a single generation.

Is it just a dream? #Veo3

Adding objects is coming soon. Stay tuned!👀