iiinova — where your digital self keeps evolving

Reposted by Dávid Komorowicz

Dominik Schnaus @schnaus.bsky.social · 09/10/2026

DINOv2 has never seen a caption, and Qwen3 has never seen an image. We still aligned their embedding spaces without a single image-caption pair. It even works when the images and the captions come from different datasets. Project page: dominik-schnaus.github.io/unpaired-rosetta ⬇️

428360

Reposted by Dávid Komorowicz

Kwang Moo Yi @kmyid.bsky.social · 07/10/2026

Lee and Lee et al., "DensiTok: Making Feed-Forward 3D Gaussian Splatting See More Views Than It Is Given" Simple idea. Can you actually learn to fill in the gaps in the latent token space such that you can simulate as if you had more views for sparse-view reconstruction?

161

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 29/09/2026

Many Eyes, One World: Feed-Forward 3D Reconstruction from Mixed Cameras Qiaoge Li, Yifan Zhan, Haijun Yang, Haiyang Liu, Yiyi Cai, Chenchi Luo tl;dr: perspective, fisheye and full-panorama images in a single forward pass arxiv.org/abs/2609.35658

021

Reposted by Dávid Komorowicz

Evangelos Kazakos @ekazakos.bsky.social · 11/09/2026

Best paper title award goes to “Honey, I Shrunk the Arc de Triomphe!” #ECCV2026

0111

Reposted by Dávid Komorowicz

Diane Larlus @dlarlus.bsky.social · 11/09/2026

WhareFormer being presented at poster #ECCV2026 363

1102

Reposted by Dávid Komorowicz

Georg Bökman @bokmangeorg.bsky.social · 10/09/2026

Tomorrow I will present the Quick ViTs poster at ECCV! Come and discuss image symmetries, GPU throughput and ViTs! Poster session 3, Friday 10:30, poster #26. Quick ViTs are a way to build ViTs that are faster than standard ViTs *and* encode planar symmetries of images into the architecture.

1152

Reposted by Dávid Komorowicz

Noa Garcia @noagarciad.bsky.social · 11/09/2026

Today we will be presenting THIS PIECE OF WORK. Encoders. Datasets. Shortcuts. Mitigations. 🕒 10:30 - 12:30 📍 ExHall #34 Come for the design, stay for the research! @eccv.bsky.social

0175

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 09/09/2026

Christian Rupprecht "Are we learning to rediscover geometry?" - our final keynote today starting in 3 minutes. #ECCV2026

071

Reposted by Dávid Komorowicz

Giorgos Tolias @gtolias.bsky.social · 25/08/2026

An interesting part: this sensitivity is exactly what makes some encoders good at spotting generated images. A liability for robustness, an asset for forensics.

142

Reposted by Dávid Komorowicz

Giorgos Tolias @gtolias.bsky.social · 25/08/2026

Your camera leaves fingerprints on every photo. Vision encoders learned to read them. Read about the bad and the good sides of it. "Invisible Shortcuts: Why Vision Encoders Know Your Camera" has been accepted at ECCV 2026. Paper: arxiv.org/pdf/2608.05424 @eccv.bsky.social #ECCV2026

52712

Dávid Komorowicz @dawars.me · 12/08/2026

No prep last minute solar eclipse from my balcony in Garching (München)

Thin sliver of partial solar eclipse setting over trees in the dark. Upside down U shape. The clouds look gloomy

0113

Reposted by Dávid Komorowicz

Kwang Moo Yi @kmyid.bsky.social · 15/07/2026

Deng, Li, Qiu, et al., "Glob3R: Global Structure-from-Motion with 3D Foundation Models" Another one showing that you SHOULD do global optimization and bundle adjustment (BA) with Feed-forward geometry estimators. Traditional formula: keyframes + view graphs + tracks + BA.

141

Reposted by Dávid Komorowicz

Nick Fisher @hydroxide.dev · 15/07/2026

Google just open sourced a new parametric human head mesh! Looks like it covers a good range of identity, expressions and head pose components. I've mostly been leaning on ICT and FLAME, but these definitely have their limitations, so a newer option is really grat. github.com/google/GNM/t...

1282

Reposted by Dávid Komorowicz

Kwang Moo Yi @kmyid.bsky.social · 13/07/2026

Ziliotto et al., "What VGGT Knows About Overlap: Probing Geometric Foundation Models for Co-Visibility" Feed-forward 3D models like VGGT show layer-wise emergent behaviors. You can treat layers as experts and train an MoE to do co-visibilty prediction, and outperform humans.

131

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 01/07/2026

AnyMatch: Supercharging Universal Multi-Modal Image Matching with Large-Scale Single-View Images Meng Yang, Zizhuo Li, Linfeng Tang, Fan Fan, Jiayi Ma tl;dr: MoGe+SD V2+image translation->multi-modal image pairs+annotations arxiv.org/abs/2606.31077

011

Dávid Komorowicz @dawars.me · 01/07/2026

I wonder if AI will cite this instead of SIFT...

110

Reposted by Dávid Komorowicz

Georg Bökman @bokmangeorg.bsky.social · 30/06/2026

New preprint out on equivariant ViTs, generalizing our prior work. Led by Chih-Chun Wang (Tikun Ong) who has done an excellent job in formalizing every part of the architecture in a nice coherent manner. He is currently looking for PhD positions in Geometric DL! arxiv.org/abs/2606.27864

0124

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 30/06/2026

Meshtryoshka: Differentiable Rendering of Real-World Scenes via Mesh Rasterization @davidcharatan.bsky.social, Daniel Xu, Richard Szeliski, @grgkopanas.bsky.social, @vincentsitzmann.bsky.social arxiv.org/abs/2606.28622

121

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 30/06/2026

Ground4D: Consistency-Aware 4D Reconstruction from Monocular Video Qing Zhao, Weijian Deng, Pengxu Wei, Liang Lin tl;dr: VGGT + MoSca arxiv.org/abs/2606.28828

001

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 23/06/2026

Visual Geometry Transformer in the Wild: Distractor-Free 3D Reconstruction Tianbo Pan, Xingyi Yang, Shizun Wang, Xinchao Wang tl;dr: distractor-aware VGGT arxiv.org/abs/2606.22787

111

Reposted by Dávid Komorowicz

Ryan Moulton @moultano.bsky.social · 19/06/2026

I spent the last month obsessed with finding colors that can't be displayed on a conventional screen. This is what I found. moultano.wordpress.com/2026/06/19/w...

moultano.wordpress.com

Where to Find the Colors Your Screen Can’t Show You

An atlas of the vibrance of the real world

471129381

Reposted by Dávid Komorowicz

Jia-Bin Huang @jbhuang0604.bsky.social · 15/06/2026

I didn't know what JEPA is, and at this point I am too afraid to ask ...😬 so I made a video tracing the last 30+ years of self-supervised learning, covering ideas from contrastive learning, distillation, masked modeling, JEPA, and world models. youtu.be/gVEr2cnDE_8

1272

Reposted by Dávid Komorowicz

Gabriel Peyré @gabrielpeyre.bsky.social · 16/06/2026

The alpha version of my new book "Optimal Transport for Machine Learners" is out, with in particular an online version with interactive figures www.gpeyre.com/ot4ml/

09843

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 10/06/2026

We have created a static website for all S23DR competitions since 2024. Write-ups, leaderboards, links to the code -- all here s23dr.github.io

s23dr.github.io

S²3DR — Structured Semantic 3D Reconstruction Challenges

041

Reposted by Dávid Komorowicz

Kwang Moo Yi @kmyid.bsky.social · 09/06/2026

Xiangli et al., "Honey, I Shrunk the Arc de Triomphe!" Metric depth estimators aren't actually metric. With curated, scaled data, they can be adapted to be better.

132

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 01/06/2026

Triangle Splatting SLAM Nicholas Fry, @ericdexheimer.bsky.social, @makezur.bsky.social, Paul H. J. Kelly, @ajdavison.bsky.social tl;dr: Triangle Splatting->map arxiv.org/abs/2605.31419

011

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 01/06/2026

SurGe: Improved Surface Geometry in Point Maps @karimknaebel.bsky.social, Gonzalo Martin Garcia, Christian Schmidt, @ifradlin.bsky.social, Lucas Nunes, @dcdegeus.bsky.social, Bastian Leibe arxiv.org/abs/2605.31577

253

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 29/05/2026

Déjà View: Looping Transformers for Multi-View 3D Reconstruction Alessandro Burzio, @tobiasfshr.bsky.social, Sven Elflein, Qunjie Zhou, Riccardo de Lutio, Jiawei Ren, Jiahui Huang, Shengyu Huang, @marcpollefeys.bsky.social, @lealtaixe.bsky.social, Zan Gojcic, Haithem Turki arxiv.org/abs/2605.30215

182

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 27/05/2026

Long-Tail Internet Photo Reconstruction @yuanli-ilnauy.bsky.social, Yuanbo Xiangli, Hadar Averbuch-Elor, @snavely.bsky.social, @ruojin.bsky.social arxiv.org/abs/2604.22714

161

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 26/05/2026

Global Structure-from-Motion Meets Feedforward Reconstruction Linfei Pan, Johannes Schönberge, @marcpollefeys.bsky.social tl;dr: Doppelganger + π³ + intrinsics/rotation/similarity averaging + BA with FF priors arxiv.org/abs/2605.26103

173

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 25/05/2026

HorizonStream: Long-Horizon Attention for Streaming 3D Reconstruction Chong Cheng, Peilin Tao, Nanjie Yao, Guanzhi Ding, Xianda Chen, Yuansen Du, Xiaoyang Guo, Wei Yin, Weiqiang Ren, Qian Zhang, Zhengqing Chen, Hao Wang tl;dr: evidence influence kernel arxiv.org/abs/2605.23889

041

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 15/05/2026

CalibAnyView: Beyond Single-View Camera Calibration in the Wild Boying Li, Cheng Zhang, Weirong Chen, @dcremers.bsky.social, Ian Reid, Hamid Rezatofighi tl;dr: new video dataset; VGGTpbased framework; for single or multi-view camera calibration arxiv.org/abs/2605.14615

021

Reposted by Dávid Komorowicz

Zhenjun Zhao @ericzzj.bsky.social · 01/05/2026

Sparse-View 3D Gaussian Splatting in the Wild Wongi Park, Jordan A. James, Myeongseok Nam, Minjae Lee, Soomok Lee, Sang-Hyun Lee, William J. Beksi tl;dr: reference view+transient mask->one-step diffusion model->refined view; use diffusion model to distill 3D->pseudo-view arxiv.org/abs/2604.27422

021

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 29/04/2026

Non-Minimal Sampling and Consensus for Prohibitively Large Datasets Seong Hun Lee, Patrick Vandewalle, Javier Civera tl;dr: you split (huuuge) tentatives list, and run RANSAC of each. How do you combine winning hypothesis? Truncated Lp cost seems the best. arxiv.org/abs/2604.22518

061

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 29/04/2026

Long-tail Internet photo reconstruction Yuan Li, Yuanbo Xiangli, Hadar Averbuch-Elor, @snavely.bsky.social @ruojin.bsky.social tl;dr: Cleaned-up version of MegaScenes + MoGe2-based depth refinement. +sparse sampler for emulate hard scenes. arxiv.org/abs/2604.22714

191

Reposted by Dávid Komorowicz

nicebyte @nicebyte.bsky.social · 12/04/2026

INTRODUCTION TO SPHERICAL HARMONICS FOR GRAPHICS PROGRAMMERS finally done. gpfault.net/posts/sph.html

gpfault.net

Introduction to Spherical Harmonics for Graphics Programmers

1027886

Reposted by Dávid Komorowicz

Johan Edstedt @parskatt.bsky.social · 07/04/2026

Introducing LoMa, the next generation of feature matcher!

3414

Reposted by Dávid Komorowicz

F. Güney @fguney.bsky.social · 31/03/2026

Sadra released the code for WarpRF 🚀 a training-free uncertainty quantification framework for radiance fields based on multi-view consistency, without training or any changes to the model.

153

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 30/03/2026

Sky2Ground: A Benchmark for Site Modeling under Varying Altitude Zengyan Wang, Sirshapan Mitra, Rajat Modi, Grace Lim, Yogesh Rawat tl;dr: Dataset + how to adapt VGGT for satellite + ground imagery. Of course, the method is named SkyNet. arxiv.org/abs/2603.13740

081

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 13/03/2026

Two #CVPR2026 competitions are live: S23DR 2026 and BuildingWorld 2026! Task: reconstruct house roof wireframes from point clouds and segmentations: Total prize fund: $22k Deadline: end of May 2026 1) huggingface.co/spaces/usm3d... 2) huggingface.co/spaces/Build... @cvprconference.bsky.social

292

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 27/02/2026

Global-Aware Edge Prioritization for Pose Graph Initialization @weitong8591.bsky.social @gtolias.bsky.social Jiri Matas, @danielbarath.bsky.social tl;dr: another global desc->GNN->MST. Supervision:# triangulated points/pair. +heuristic MST postprocessing. Eval on IMC-PT. arxiv.org/abs/2602.21963

072

Reposted by Dávid Komorowicz

Cristián Llull @cllullt.bsky.social · 27/02/2026

SECOND CALL: SHREC'26 Challenge on 3D Reconstruction Our dataset features intricate geometries, ideal for benchmarking of high-frequency detail recovery. All participating will co-author a joint paper submitted to Computers & Graphics. Track Details shapevision.dcc.uchile.cl/cllull-shrec2026

shapevision.dcc.uchile.cl

SHREC 2026: Reconstruction of High-Frequency Geometry in Synthetic Heritage

The dataset is generated using a novel procedural pipeline that combines mesh sculpting with semantic vertex annotation, i.e., our ground truth includes per-vertex semantic metadata.

001

Reposted by Dávid Komorowicz

#CVPR2026 @cvprconference.bsky.social · 22/01/2026

#CVPR2026 reviews are slowly being dispatched over email. Good luck!

081

Reposted by Dávid Komorowicz

Eugene Vinitsky 🍒 @eugenevinitsky.bsky.social · 01/01/2026

new year and @mariaa.bsky.social and I have some fun new things cooking for the atproto ecosystem...

Screenshot of a paper discussion page titled ‘mHC: Manifold-Constrained Hyper-Connections’. At the top is a card showing the paper title, a ‘View on arXiv’ link, and indicators for 7 posts and 7 researchers. Below are social-style posts referencing the paper: one from ‘NT 5.2 Pyongyang Official™’ linking to arXiv with the caption ‘THE WHALE IS BACK BABYYYY’ and an arXiv preview image, and another from ‘Hacker News’ linking to the same arXiv paper. The interface resembles a research discussion or social feed layout.

1011614

Reposted by Dávid Komorowicz

Sean Carroll @seanmcarroll.bsky.social · 22/10/2024

Hey this worked! Import all of your old Twitter posts over to Bluesky (for a few bucks, depending on how irrepressible you were). Now I look like I've been posting on this platform longer than it has existed.

1111424

Reposted by Dávid Komorowicz

Joe Kider @hepcatjk.bsky.social · 15/08/2025

Choosing the right colormap is tricky, too often, they hide subtle details or distort the data. Our new method transforms colormaps to boost local contrast and reveal just noticeable differences, all while keeping the visualization perceptually accurate and accessible. dl.acm.org/doi/10.1145/...

1469

Reposted by Dávid Komorowicz

Andrei Bursuc @abursuc.bsky.social · 21/07/2025

1/ Can open-data models beat DINOv2? Today we release Franca, a fully open-sourced vision foundation model. Franca with ViT-G backbone matches (and often beats) proprietary models like SigLIPv2, CLIP, DINOv2 on various benchmarks setting a new standard for open-source research.

28722

Reposted by Dávid Komorowicz

Wenzel Jakob @wjakob.bsky.social · 08/08/2025

How can one reconstruct the complete 3D interior of a wood block using only photos of its surfaces? 🪵 At SIGGRAPH'25 (Thursday!), Maria Larsson will present *Mokume*: a dataset of 190 diverse wood samples and a pipeline that solves this inverse texturing challenge. 🧵👇

27615

Reposted by Dávid Komorowicz

Dmytro Mishkin @ducha-aiki.bsky.social · 05/08/2025

VLM-Guided Visual Place Recognition for Planet-Scale Geo-Localization Sania Waheed, Na Min An, Michael Milford , Sarvapali D. Ramchurn, Shoaib Ehsan tl;dr: in title arxiv.org/abs/2507.17455

072