Hui Ren

PhD student @ UIUC

[Hui Ren @rhfeiyang]

🎨Getting ready for #NeurIPS2025! ✈️☀️

I’m excited to present our work "Opt-In Art: Learning Art Styles Only from Few Examples" at Creative AI Track in San Diego.

If you are interested in generative ai, definitely stop by and connect!

See you there!

[Joanna @materzynska]

1/ 🎨✨ If an AI has never seen art before, can it develop artistic capabilities?
Our team from MIT, Northeastern and ShanghaiTech University analyzes this question in our Art-Free Diffusion paper.

[Hui Ren @rhfeiyang]

Curious about how the industry is thinking about DLLMs

[Chumeng Liang @lowerbad]

Continuous diffusion dominates image & video generation, but people used to believe that it inherently lags behind its discrete counterparts in language modeling.

Today, we challenge this belief with LangFlow: the first continuous diffusion language model that rivals—and even beats—discrete diffusion. (1/7)

Blog: https://t.co/EtZRSx9MQv
GitHub: https://t.co/NgWUDDAXd6
Arxiv: https://t.co/2WfaQL7IZZ

[Jia-Bin Huang @jbhuang0604]

Excited to return to my alma mater, UIUC, next week!

Beyond a research talk, I’m preparing a session on Effective Visual Communication, distilling lessons from content creators and my experiments in making YouTube videos.

Still working on it, any tips or resources are welcome!

[Hui Ren @rhfeiyang]

@jbhuang0604 Really looking forward to it!

[Hui Ren @rhfeiyang]

@yash2kant @igilitschenski @UofTCompSci @Snapchat Congratulations!!!

[Hui Ren @rhfeiyang]

Today 16:30, Creative AI Session, Upper Level Room 29A-D
Find me with @materzynska @rohitgandikota @giannis_daras @davidbau @abtorralba
See you there!

[Hui Ren @rhfeiyang]

🎨Getting ready for #NeurIPS2025! ✈️☀️

I’m excited to present our work "Opt-In Art: Learning Art Styles Only from Few Examples" at Creative AI Track in San Diego.

If you are interested in generative ai, definitely stop by and connect!

See you there!

[Hui Ren @rhfeiyang]

@pratyusha_PS Congratulations!!!

[Rohit Gandikota @rohitgandikota]

If you train a diffusion model with NO art samples - they can still learn to mimic art !?

Come checkout this @NeurIPSConf poster by @rhfeiyang and @materzynska where they present how it is possible for artists to "opt-in" rather than having to "opt-out" of these models

[Hui Ren @rhfeiyang]

I am actively looking for Summer 2026 Research Internship in topics of Video Generation, Vision-language Models and Unified Multimodal Models.

🙏 Referrals or pointers appreciated. Happy to share more details!

[Hui Ren @rhfeiyang]

🎨Getting ready for #NeurIPS2025! ✈️☀️

I’m excited to present our work "Opt-In Art: Learning Art Styles Only from Few Examples" at Creative AI Track in San Diego.

If you are interested in generative ai, definitely stop by and connect!

See you there!

[Joanna @materzynska]

1/ 🎨✨ If an AI has never seen art before, can it develop artistic capabilities?
Our team from MIT, Northeastern and ShanghaiTech University analyzes this question in our Art-Free Diffusion paper.

[Hui Ren @rhfeiyang]

👉 Catch our session here: https://t.co/fYBcX3onxK

🧠 Try our Demo: https://t.co/n8EvLPokKe

[Hui Ren @rhfeiyang]

Proud to collaborate with @materzynska @rohitgandikota @giannis_daras @davidbau @abtorralba

[Yi Ma @YiMaTweets]

When we Chinese were building the Great wall, we were firm believers in the "scaling law": If we were to build a wall just long enough, it would be sufficient to defend the enemy from the north. However magnificent an engineering marvel that was, it is just a wall after all, not a full defense system. So it did not work. We see this as common sense by now. But we do not seem to recognize the same goes for today's large pre-train models... However large amount of knowledge it stores, it is merely a fragment of an intelligent system. Strictly speaking, knowledge does not generalize, only the ability to improve knowledge does, which we call Intelligence.

[AK @_akhaliq]

discuss: https://t.co/NOW3piwWF2

[AK @_akhaliq]

Feature4X

Bridging Any Monocular Video to 4D Agentic AI with Versatile Gaussian Feature Fields

[Joanna @materzynska]

🚀 Join the First Workshop on Mechanistic Interpretability for Vision (MIV) at #CVPR2025! 🔍

✨ Got ideas? Submit your papers here: https://t.co/mTQnP9IzmR
🎤 Excited for the talks? Check out the speaker lineup: https://t.co/UOw7nDfB8L

[Hui Ren @rhfeiyang]

@yasiu071 @materzynska The token length limit is 512, (SD is 77, for reference)

[Hui Ren @rhfeiyang]

@yasiu071 @materzynska The prompts like "yeasts and dermatophytes" may be rare, which leads to low quality because it is not in the same distribution as training. Long and detailed captions would be better.😀

[Hui Ren @rhfeiyang]

Check out our demo for the Art-Free Diffusion model and the Art Adapters! 🎨✨

https://t.co/3YNvyLE1oZ

[Joanna @materzynska]

Our project demo is LIVE! 🎨✨ Check out the Art-Free Diffusion model and the Art Adapters build on top of it!
https://t.co/IezFvQjDy1

[Jia-Bin Huang @jbhuang0604]

Live view of the AI/ML grad school application process

[Hui Ren @rhfeiyang]

@xd817370701035 @xiaolonw Shame on your bias