Haojun Qiu

📢📢📢We introduce Efficient-SID⚡️: training-free single-image diffusion model that generates images by sampling directly from an input image's patch distribution. Our method enables megapixel generation in <1s and scales to gigapixel generation. We also enable stylization, editing, and other applications. The outputs are constrained to follow exactly the patch distribution of the input — something that is very difficult to do with large models! #CVPR2026 Highlight 🌐 haojunqiu.github.io/efficient-SID/ 📄 arxiv.org/abs/2606.04299 [1/6]

@ruihangzhang Really cool work, congrats!!

Introducing 📢📢ProxyPose📢📢 ✨Track 6-DoF motion of any query pixel via video-to-video translation 📰ArXiv: arxiv.org/abs/2607.06555 💻Code: github.com/ruihangzhang97… 🌐Webpage: ruihangzhang97.github.io/proxypose

@_Linjunru Thank you Junru!

Very cool! Congrats @HaojunQiu !

@ruihangzhang Thank you Ruihang!

@HaojunQiu congrats @HaojunQiu!! amazing work!!

📢📢📢We introduce Efficient-SID⚡️: training-free single-image diffusion model that generates images by sampling directly from an input image's patch distribution. Our method enables megapixel generation in <1s and scales to gigapixel generation. We also enable stylization, editing, and other applications. The outputs are constrained to follow exactly the patch distribution of the input — something that is very difficult to do with large models! #CVPR2026 Highlight 🌐 haojunqiu.github.io/efficient-SID/ 📄 arxiv.org/abs/2606.04299 [1/6]

@estheroate Thank your Esther!!

@HaojunQiu congrats, this is really great work!

@DaveLindell Thank you David! Excited to finally share it — thanks for all the guidance on the project!

Really cool project led by @HaojunQiu! We show a patch-based image generation method with closed-form diffusion (i.e., analytical denoising—no neural network). It's *super* efficient and even scales to gigapixel generation! #CVPR2026

📢📢📢 Velox 🚀: Learning Representations of 4D Geometry and Appearance In our #CVPR2026 paper, we introduce a method for learning a native 4D representation, useful for many downstream tasks, such as video-to-4D, 3D tracking, cloth simulation, and others! 🌐: apple.github.io/ml-velox 📝: arxiv.org/abs/2605.04527

@rms80 @kyroskutulakos @DaveLindell Thank you Ryan!

@HaojunQiu @kyroskutulakos @DaveLindell this is cool

CVPR 2026, Denver — Poster #704, June 7th morning session. Come find us! We'll have a live demo at our poster! Huge thanks to my collaborators @kyroskutulakos and @DaveLindell

We can use the same algorithm and introduce guidance from large models or additional generation constraints to show other applications: • Retargeting (content-aware resize) • Symmetrization • Seamless tiling • Style transfer • Text-guided stylization ("Van Gogh", "Monet") [5/6]