Coding agents can now operate graphics tools. How will they change the way we design graphics algorithms and conduct graphics research?
Join the discussion at the 1st Workshop on Agentic Graphics on Dec 1 afternoon at #SIGGRAPHAsia2026!
🔗 https://t.co/0HhJfgsWkP
After spending a lot of time processing the events of this past week, I came to the conclusion that one can respect several aspects of a person’s character, such as their grit and determination, generosity, and positive vision for the future, while simultaneously disapproving of some of their choices and decisions. I try not to hold a singular perspective.
I want to share two anecdotes of my personal interactions with Jensen, one over email and another in person:
1. I was a co-author on a publication that won the “best paper award” at SIGGRAPH a few years back (I’m not trying to brag, I promise!). Excited by its potential, I sent out an email internally at NVIDIA explaining the research and outlining the long-term vision. I also made a 10 minute video, that I spent hours refining, that went more in-depth into the work.
That very day, Jensen was the first person to respond to the email. He called it “game changing” whilst understanding that the research wasn’t quite ready for prime time. He asked how we might go about industrializing this work, and tagged relevant individuals who might be able to help. This spawned a flurry of responses and collaborations that last to this day.
I’ve since been given the opportunity to take this work from “research to production”. Those I report to show faith in me daily, even though they understand full-well the large amount of effort involved in making such a transition with a nascent technology.
Jensen’s ability to connect with another person's research vision is unique. He has also fostered a very special culture within the company that allows people to do “their life’s work”. I do not take it for granted.
2. I got the opportunity to chat briefly with Jensen at an internal NVIDIA event at SIGGRAPH. The conference had been struggling with fundraising and Jensen had backed it as the primary sponsor that year.
Given my own background in computer graphics, I passionately described to him how graphics researchers, time and again, make fundamental breakthroughs in algorithms & computing, and that it was great he was supporting the community. I mentioned Gaussian splats as a recent example of impactful research published at SIGGRAPH. Even though splats were relatively new at the time, he already had an understanding of the technology and its potential. After our conservation, he declared to everyone in the room, “Let’s make SIGGRAPH great again”!
Even though NVIDIA has transitioned to being an “AI company” over the past decade, Jensen has not forgotten its roots in computer graphics. He helped the community when it was at its lowest.
Today, I am launching @ninesrobotics with @tianx2li.
We build autonomous robots for AI data centers.
Currently, AI only optimizes itself in software. By building the automation layer for server rooms, we are creating the world’s first fully software-defined, physically adaptive AI factory - the essential bridge that allows agentic AI to autonomously manage, manipulate, and optimize its own physical hardware stack.
NeurIPS 2026 Spotlight! 🌟 Physics simulators like MuJoCo MJX are changing robotics. But for studying the sensorimotor learning of speech, there's no equivalent. We built diffTLM, a differentiable vocal tract simulator, to help close that gap. (1/6)
If video models scale like LLMs, they'll soon generate a perfect video plan for any embodied task. Do we need anything else? Introducing Mirror, which learns to follow these plans just by practicing, and even crafts an iron pickaxe in Minecraft w/o rewards or action labels! (1/n)
Can diffusion models generate high-quality images without CFG?
Yes! We introduce ✨Correction Training for Diffusion Models✨, which significantly improves sample quality without relying on classifier-free guidance.
Project page: https://t.co/Ic8foxITJ0
🧵 (1/n)
After several wonderful years at Cornell, I will be moving to Michigan next year (partly driven by family reasons). Excited for new adventures in U Michigan and Ann Arbor!
If you are a PhD applicant looking to work with me please apply to U Michigan!
✨Thinking with Blender~
Meet VIGA: a multimodal agent that autonomously codes 3D/4D blender scenes from any image, with no human, no training!
@berkeley_ai#LLMs#Blender#Agent 🧵1/6
I'm recruiting multiple PhD students this cycle to join me at Harvard University and the Kempner Institute! My interests span vision and intelligence, including 3D/4D, active perception, memory, representation learning, and anything you're excited to explore!
Deadline: Dec 15th.
📢Thrilled to share that I'll be joining Harvard and the Kempner Institute as an Assistant Professor starting Fall 2026!
I'll be recruiting students this year for the Fall 2026 admissions cycle. Hope you apply!
What if a Transformer could render?
Not text → image.
But mesh → image — with global illumination.
No rasterizers. No ray-tracers. Just a Transformer without per-scene training.
RenderFormer does exactly that.
#SIGGRAPH2025
🔗https://t.co/vk6cA4FgOE
We've released our paper "Generating 3D-Consistent Videos from Unposed Internet Photos"! Video models like Luma generate pretty videos, but sometimes struggle with 3D consistency. We can do better by scaling them with 3D-aware objectives. 1/N
page: https://t.co/Hgu8uo3tvu
We'll present "Neural Control Variates with Automatic Integration" at the Monte Carlo for PDE session in Mile High 4 at 11:25 tomorrow. Come to our talk to learn more about creating control variates function with arbitrary neural networks!
#SIGGRAPH2024
How do we create realistic models of dressed humans directly from visual data?
We introduce PhysAvatar, a framework that estimates the shape, appearance, and physical parameters of dressed human avatars from multi-view videos.
Page: https://t.co/d87GBIsse2
(1/6)
#SIGGRAPHAsia2023 Combining Neural Fields with Monte Carlo methods can result in faster and more accurate PDE solvers.
Come to see our talk tomorrow at 15:45pm. C4.8 (level 4). w @stevenygd, Xi Deng, @chrismdesa, @BharathHarihar3, Steve Marschner