Zhixiang Wang
136 posts
Zhixiang Wang
@lightchaserx
Researcher. Recreating the physical world. | Ex PhD @UTokyo_News_en, @Snap, @NTUsg, @ NTUtw
Tokyo-to, Japan Katılım Ocak 2021
1.9K Takip Edilen622 Takipçiler
Check our latest work on generative rendering
Alaya Studio@alayastd
𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗪𝗼𝗿𝗹𝗱 𝗥𝗲𝗻𝗱𝗲𝗿𝗲𝗿 A new toolkit, dataset, and baseline for world-scale rendering: - collect G-buffers from AAA games - scale data for rendering complex world scenes - improve rendering performance - enable game effect editing
English
Zhixiang Wang retweetledi
𝗜𝗻𝘁𝗿𝗼𝗱𝘂𝗰𝗶𝗻𝗴 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗪𝗼𝗿𝗹𝗱 𝗥𝗲𝗻𝗱𝗲𝗿𝗲𝗿
A new toolkit, dataset, and baseline for world-scale rendering:
- collect G-buffers from AAA games
- scale data for rendering complex world scenes
- improve rendering performance
- enable game effect editing
English
Zhixiang Wang retweetledi
WildWorld
A large-scale dataset for dynamic world modeling with 108M+ frames from Monster Hunter: Wilds, featuring 450+ actions and explicit state annotations including skeletons, camera poses, and depth maps.
English
Zhixiang Wang retweetledi
PackForcing
Short Video Training Suffices for Long Video Sampling and Long Context Inference
paper: huggingface.co/papers/2603.25…
English
Zhixiang Wang retweetledi
Stream-DiffVSR: A low-latency, streamable video super-resolution model;
- distills Stable Diffusion Upscaler to just 4 steps;
- processes 720p ~3fps.
jamichss.github.io/stream-diffvsr…
English
Deeply useful philosophy and practice for my research and beyond.
Jia-Bin Huang@jbhuang0604
Junior students who have just started doing research? Check out the (75 and counting) awesome tips! github.com/jbhuang0604/aw…
English
Zhixiang Wang retweetledi
Junior students who have just started doing research?
Check out the (75 and counting) awesome tips!
github.com/jbhuang0604/aw…
English
We will (potentially) host an IJCV special issue based on this workshop. Come to discuss with us if you to contribute. And check the updates at responsible-imaging.github.io
English
🚗 Join us for the Responsible Imaging Workshop @ ICCV 2025!
🗓 Oct 20 afternoon | 📍 Hawaii Conv. Center, 305A
Featuring exciting talks and discussions by Bill Freeman, Kyros Kutulakos, Amir Zamir, Laura Waller and @MarkSheinin Don’t miss it!
#ICCV2025
English
The deadline for non-processing track is approaching (August 19)! Please consider to share your latest innovations and perspectives through our @ICCVConference workshop! #ICCV2025
Jian Wang@_JianWang_
📣Call for Submissions - Responsible Imaging @ICCVConference! Join us at #ICCV2025 to push the boundaries of responsible imaging systems. 📅 Deadline: 7/3 Proceedings, 8/19 Non-Proceedings 🥇 Best poster/paper awards, over $1,000 in prizes 🌐 More info: responsible-imaging.github.io
English
@a1k0n Same here! Just realized this myself and came across your tweets while searching online.
English
GRPO looks more like an importance sampling or evolutionary strategy (akin to cross-entropy method or CMA-ES) than reinforcement learning to me.
English
Zhixiang Wang retweetledi
[Eng] Thanks to many kind colleagues for nominating me to run for ICML Board. If elected, I hope to extend my experience to bring ICML & Asian ML community closer together, and to foster even more ICML-NeurIPS dislogue. I'd appreciate your vote & help in spreading the word.
English
Zhixiang Wang retweetledi
Zhixiang Wang retweetledi
Sekai: A Video Dataset towards World Exploration
A high-quality 5k hrs of egocentric worldwide video + audio dataset for world exploration, created from Youtube with high-quality annotations
English
We are excited to announce the 2nd Workshop on AVGenL: Audio-Visual Generation & Learning at #ICCV2025! This year, in addition to an outstanding lineup of speakers, we’ll be featuring industrial demo sessions. Stay tuned—more details are coming soon!
goo.su/p8cj
English
Zhixiang Wang retweetledi
📣Call for Submissions - Responsible Imaging @ICCVConference!
Join us at #ICCV2025 to push the boundaries of responsible imaging systems.
📅 Deadline: 7/3 Proceedings, 8/19 Non-Proceedings
🥇 Best poster/paper awards, over $1,000 in prizes
🌐 More info: responsible-imaging.github.io
English
Zhixiang Wang retweetledi
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
🔗microsoft.github.io/renderformer
English
@ducha_aiki @JerredChen Nice! Blur contains information about motion (although still ambiguous). Based on the observation, we had leveraged a special blur to correct rolling shutter distortion in CVPR’22: lightchaserx.github.io/ngs/ and for frame interpolation
English
Image as an IMU: Estimating Camera Motion from a Single Motion-Blurred Image
@JerredChen , Ronald Clark
tl;dr:predict flow from blurred image -> solve for velocity, use as IMU information.
arxiv.org/abs/2503.17358
English