How do we teach vision models to infer visual concepts from just a handful of example images? Check out our new ECCV paper, where we explore learning directly from image sets 👇

Nick Stracke@rmsnorm.bsky.social · last mo.

Most representation learning methods produce generic embeddings that try to preserve everything in an image. We introduce VICIS: a way to use example sets to define a tailored embedding space for what matters. This is useful whenever the desired visual signal is easier to show than to describe. 🧵👇

Diffusion models treat every part of an image equally. → Same number of steps. Same compute. But images aren’t uniform. 🤔 Some regions are easy, others are hard. So why force the model to treat them the same? 🧵

Do we really need pixel generation to model motion? 🤔 We show how directly representing motion in a compact space enables efficient, scalable planning. 10,000× faster than video models, enabling planning and reasoning in open-world and robotics settings. Check it out ⬇️

Nick Stracke@rmsnorm.bsky.social · 4mo ago

Video diffusion models learn motion indirectly through pixels. But motion itself is much lower-dimensional. We introduce 64× temporally compressed motion embeddings that directly capture scene dynamics. This enables efficient planning -> 10,000× faster than video models. 🧵👇

You don't imagine the future by mentally rendering a movie. You trace how things move -- abstractly, sparsely, step by step. We built a model that does exactly this. It predicts motion, not pixels -- and it's 3,000× faster than video world models. Myriad, accepted at @cvprconference.bsky.social

I’m thrilled to share that I’ll present two first-authored papers at #ICCV2025 🌺 in Honolulu together with @mgui7.bsky.social ! 🏝️ (Thread 🧵👇)

🤔 What happens when you poke a scene — and your model has to predict how the world moves in response? We built the Flow Poke Transformer (FPT) to model multi-modal scene dynamics from sparse interactions. It learns to predict the 𝘥𝘪𝘴𝘵𝘳𝘪𝘣𝘶𝘵𝘪𝘰𝘯 of motion itself 🧵👇

🤔When combining Vision-language models (VLMs) with Large language models (LLMs), do VLMs benefit from additional genuine semantics or artificial augmentations of the text for downstream tasks? 🤨Interested? Check out our latest work at #AAAI25: 💻Code and 📝Paper at: github.com/CompVis/DisCLIP 🧵👇

In order to extract features from diffusion models, you have to noise your input and tune the noise level for each downstream task. But isn't there a better way? 🤔 Turns out there is, using our newly proposed feature extraction method CleanDIFT 🧹🚀 Check it out ⬇️

Nick Stracke@rmsnorm.bsky.social · 2y ago

🤔 Why do we extract diffusion features from noisy images? Isn’t that destroying information? Yes, it is - but we found a way to do better. 🚀 Here’s how we unlock better features, no noise, no hassle. 📝 Project Page: compvis.github.io/cleandift 💻 Code: github.com/CompVis/clea... 🧵👇

After many years, our lab finally has a social media presence at @compvis.bsky.social ! 🥳 Give it a follow, we have some amazing research on generative computer vision coming soon!