📢𝐐𝐮𝐢𝐜𝐤𝐒𝐩𝐥𝐚𝐭: Fast 3D Surface Reconstruction via Learned Gaussian Initialization
Yueh-Cheng Liu learns 2DGS initialization, densification, and optimization priors from ScanNet++ => fast & accurate reconstruction!
Project: liu115.github.io/quicksplat
w/ Lukas Höllein and @niessner.bsky.social
📢📢 𝐀𝐯𝐚𝐭𝟑𝐫 📢📢
Avat3r creates high-quality 3D head avatars from just a few input images in a single forward pass with a new dynamic 3DGS reconstruction model.
youtu.be/P3zNVx15gYs
tobias-kirschstein.github.io/avat3r
Great work by Tobias Kirschstein with J. Romero, A. Sevastopolsky, S. Saito.
(1/n)
📢📢 𝐍𝐞𝐑𝐒𝐞𝐦𝐛𝐥𝐞 𝐯𝟐 𝐃𝐚𝐭𝐚𝐬𝐞𝐭 𝐑𝐞𝐥𝐞𝐚𝐬𝐞 📢📢
Head captures of 7.1MP from 16 cameras at 73fps:
* More recordings (425 people)
* Better color calibration
* Convenient download scripts
github.com/tobias-kirsc...
📢 ScanNet++ v2 Benchmark Release! 🏆
Test your state-of-the-art models on:
🔹 Novel View Synthesis 📸➡️🖼️
🔹 3D Semantic & Instance Segmentation 🤖🔍🕶️
Shoutout to @awhiteguitar.bsky.social & Yueh-Cheng Liu for their incredible work👏
🚀Check it out: kaldir.vc.in.tum.de/scannetpp/
Can GANs compete in 2025? In 'The GAN is dead; long live the GAN! A Modern GAN Baseline', we show that a minimalist GAN w/o any tricks can match the performance of EDM with half the size and one-step generation - github.com/brownvc/r3gan - work of Nick Huang, @skylion.bsky.social, Volodymyr Kuleshov
Researcher: "We let the data speak for itself."
Earlier that day:
Excited to announce ScanNet++ v2!🎉
@awhiteguitar.bsky.social & Yueh-Cheng Liu have been working tirelessly to bring:
🔹1006 high-fidelity 3D scans
🔹+ DSLR & iPhone captures
🔹+ rich semantics
Elevating 3D scene understanding to the next level!🚀
w/ @niessner.bsky.social
kaldir.vc.in.tum.de/scannetpp
#NeurIPS was a blast - geting to know great people and meeting old friends, lots of LLM, multi-modal, synthetic data and diffusion.
Whats a better way to wrap up such a week than watching an NHL game.
(1/2)
📢📢𝐆𝐀𝐅: 𝐆𝐚𝐮𝐬𝐬𝐢𝐚𝐧 𝐀𝐯𝐚𝐭𝐚𝐫 𝐑𝐞𝐜𝐨𝐧𝐬𝐭𝐫𝐮𝐜𝐭𝐢𝐨𝐧 𝐟𝐫𝐨𝐦 𝐌𝐨𝐧𝐨𝐜𝐮𝐥𝐚𝐫 𝐕𝐢𝐝𝐞𝐨𝐬 𝐯𝐢𝐚 𝐌𝐮𝐥𝐭𝐢-𝐯𝐢𝐞𝐰 𝐃𝐢𝐟𝐟𝐮𝐬𝐢𝐨𝐧📢📢
We reconstruct animatable Gaussian head avatars from monocular videos captured by commodity devices such as smartphones.
📢MeshArt: Generating Articulated Meshes with Structure-guided Transformers
Daoyi Gao generates articulated meshes with a hierarchical transformer, modeling articulation-aware structures that guide mesh synthesis.
w/ Yawar Siddiqui, Lei Li
Project Page: daoyig.github.io/Mesh_Art/
🚨 Exciting news! If you're at #NeurIPS, come check out our poster tomorrow morning at 11 AM during the first session! 🎉
We're presenting "Neural Experts: Mixture of Experts for Implicit Neural Representations (INRs)" – here's what it's all about: 🧵👇
Great collab with @adai.bsky.social, Norman Müller and @niessner.bsky.social
- Probabilistic diffusion formulation is well suited for ambiguous cases in single image reconstruction task
- Exploit the intermediate shape representation for efficient surface alignment loss for joint training
- Represent objects using disentangled shape encoding results in high-quality shapes
Learnings:
- Formulate single image scene reconstruction as conditional diffusion processes
- Learn a generative scene prior by denoising all objects simultaneously leads to more coherent arrangements
Super happy to present our #NeurIPS paper 𝐂𝐨𝐡𝐞𝐫𝐞𝐧𝐭 𝟑𝐃 𝐒𝐜𝐞𝐧𝐞 𝐃𝐢𝐟𝐟𝐮𝐬𝐢𝐨𝐧 𝐅𝐫𝐨𝐦 𝐚 𝐒𝐢𝐧𝐠𝐥𝐞 𝐑𝐆𝐁 𝐈𝐦𝐚𝐠𝐞 in Vancouver. Come to our poster #2804 on Wednesday 11am - 2pm in East Exhibit Hall, if you want to learn more about 3D Scene Diffusion!
See you tomorrow!
Project Page: manuel-dahnert.com/research/sce...
📢SceneFactor: Generating & editing 3D indoor scenes from text!
Alexey Bokhovkin presents a factored latent diffusion to generate high-level semantics + geometric refinement, enabling for controllable, large-scale scene synthesis
w/ Quan Meng, Shubham Tulsiani
alexeybokhovkin.github.io/scenefactor
@ajdavison.bsky.social could go on this list :)
I dont know too much about the differences, e.g. the carries they use, but Nomad was good
Nomad eSim worked fine and was really convenient. It's a little bit more expensive but if you don't plan to livestream the entire conference, it'll be enough data.
Exciting news! 🎉
Our podcast now has a BlueSky account! Follow us for updates, behind-the-scenes, new episodes and more at @talking-papers.bsky.social .
See you there! #Podcast #BlueSky
go.bsky.app/MfnrAdL
Let me know if you want to be added, or have any suggestions!
Are you a social media manager running an account for an organization or high-profile individual? Or do you own a website and want a simpler username?
On Bluesky, you can verify your account by setting your website as your username.
We just refreshed our how-to guide:
For everyone who wants to update their list of social media badges on their website: FontAwesome already has the proper icon :)
fontawesome.com/icons/bluesky
Y
Excited to announce the second iteration of our TUM AI Lecture series!
We'll be having a lineup of top-notch speakers across areas in AI from both academia and industry. Talks will be streamed live and and remain available online - check out prev talks:
niessner.github.io/TUM-AI-Lectu...
I've created a startepack on Generative Modeling: go.bsky.app/Hd9ykTw
Future Bsky homes for our #computervision conferences!
@cvprconference.bsky.social
@iccv.bsky.social
@eccv.bsky.social
@wacvconference.bsky.social
Stay tuned!
@karpathy.bsky.social could potentially be added ;)
Yeah here's my geometry processing starter pack. I hand-picked researchers and practitioners either in geometry processing or not, and some people are definitely missing so,
go.bsky.app/Dqw66qn
Converted my Graphics Research list to a starter pack (not sure what's the difference though). Let me know who we are missing here :)
Here goes! go.bsky.app/ApQNTt2
