We as SOAD have just released new music for the first time in 15 years. The time to do this is now, as together, the four of us have something extremely important to say as a unified voice. Read our full statement at SystemOfADown.bandcamp.com. #ProtectTheLand #GenocidalHumanoidz

Always a great fun when Two Minute Papers covers your work! w/ Yoni Kasten Dolev Ofri-Amar @oliverwang81

I'm going to use this tweeter account to post feature visualizations of various ViT models. The first one, ViT base patch32 trained on ImageNet1k. Showing channel 980 (class "volcano").

Excited to share that our paper was accepted to CVPR 2022!

Text2LIVE: Text-Driven Layered Image and Video Editing abs: arxiv.org/abs/2204.02491 project page: text2live.github.io

Billion params text-to-image models are amazing! But...not designed for editing real-world images/videos. Text2LIVE (ECCV oral) trains on 1 example and allows for various semantic, localized editing! text2live.github.io Omer Bar Tal Dolev Ofri-Amar Rafail Fridman Yoni Kasten 1/3

SceneScape: Text-Driven Consistent Scene Generation abs: arxiv.org/abs/2302.01133 project page: scenescape.github.io text-driven perpetual view generation -- synthesizing long videos of arbitrary scenes solely from an input text describing the scene and camera poses

Neural Congealing: Aligning Images to a Joint Semantic Atlas abs: arxiv.org/abs/2302.03956 project page: neural-congealing.github.io

Plug-and-Play accepted to #CVPR2023 :-) Congrats PnP team: Narek Tumanyan Michal Geyer OneViTaDay!

[1/5] Always wondered what people see when looking at a Rorschach test? SpaText - our recent #CVPR2023 paper from @MetaAI may give you a sneak peek! TL;DR: We extend text-to-image models with region-specific textual controllability. Project Page: omriavrahami.com/spatext/

Superised to find that Google Deepmind released a benchmark for Tracking Any Point in 3D (TAPVid-3D)! 😍 tapvid3d.github.io I believe human perceptions for low level motions is not just those 2D flows. We understand 3D dynamics and motions!

Excited to introduce our new paper, Generative Omnimatte: Learning to Decompose Video into Layers, with the amazing team at Google DeepMind! Our method decomposes a video into complete layers, including objects and their associated effects (e.g., shadows, reflections).

🔍 Unveiling new insights into Vision-Language Models (VLMs)! In collaboration with OneViTaDay & Tali Dekel, we analyzed LLaVA-1.5-7B & InternVL2-76B to uncover how these models process visual data. 🧵 vision-of-vlm.github.io