
Denys Iliash retweeted
Rule #1 of 3DGS capture: don't touch anything! Open a door and it messes up your whole scene. ๐ช
But what if you do?
๐๐ฐ๐ข๐ญ๐๐ก๐๐ฉ๐ฅ๐๐ญ (SIGGRAPH Asia 2026) turns what you touched into switches you can flip ๐
switchsplat.github.io
Excited to share GATOR ๐! From a few casual photos (clutter, occlusion, estimated cameras), we reconstruct complete, textured 3D objects placed back in their scene. A generative model builds the asset, then an agent refines it. research.nvidia.com/labs/lprโฆ
Casual photos in, simulation-ready 3D objects out! Introducing ๐GATOR๐, generative + agentic 3D object reconstruction. Here, humanoids interact with GATOR-reconstructed furniture [1/n]
Website: research.nvidia.com/labs/lprโฆ
Paper: arxiv.org/abs/2610.11215
Denys Iliash retweeted
Actually there was some more to explore! ๐ @NeurIPSConf #Neurips2026 See you at the conference!
Happy to see these papers accepted #Neurips2026 ๐ ๐ ๐
Please come and find us at the conference!
More details & links in ๐งต
Denys Iliash retweeted
Happy to see these papers accepted #Neurips2026 ๐ ๐ ๐
Please come and find us at the conference!
More details & links in ๐งต
Denys Iliash retweeted
3D models can look right and still fall apart.
Introducing Functionalization: graph completion adds missing structure and connectors to rectify motion and make models work.
SIGGRAPH Asia 2026
Paper, code, data available at: mingrui-zhao.github.io/Functโฆ
Denys Iliash retweeted
1/8 Good morning Malmรถ ๐ธ๐ช! #ECCV2026
Have you wondered how far we are from building interactable 3D worlds? Come join our FAI workshop and hear the worldโs leading researchersโ opinions!
Today, 9 Sept ยท 14:00โ17:35 CEST
Malmรถ Arena ยท Palissad East
Talk details follow!
Join us for the Functionality, Articulation & Interaction workshop at #ECCV2026 tomorrow! Our incredible lineup:
- Andrea Vedaldi (@Oxford_VGG)
- Wei-Chiu Ma (@weichiuma)
- Evangelos Kalogerakis (@EvangelosKalog1)
- Angela Dai (@angelaqdai)
- Mikaela Angelina Uy (@mikacuy)
Hosted by: @jiang_hanxiao, @RuiqisNotes, @richardzhangsfu
๐ diliash.github.io/fai_eccv_wโฆ
@eccvconf
3
159
Denys Iliash retweeted
1/5 From one physical digital twin to 1,000+ interactive users โ on a single GPU. ๐
Introducing Boba ๐ง, our #ECCV2026 system for high-throughput batched physical digital twins.
Come try it live!
๐
Sep 11, 12:30โ17:00
๐ Demo 4
๐งต๐
๐ How can we create interactive Physical Digital Twins from videos?
Thrilled to share our latest work: PhysTwin! ๐ Using inverse physics optimization, we generate photo-realistic, physically accurate, and real-time interactive virtual replicas. ๐ฅ
๐jianghanxiao.github.io/phystโฆ
Denys Iliash retweeted
Super proud of our work.
You can read more about it here:
๐ serchirag.github.io/roms-imlโฆ
๐ arxiv.org/pdf/2607.19332
Diffusion/flow matching is a go-to generative AI paradigm, but it's slow, sample inefficient and complicated. Do we really need all this complexity?
We design a small and simple one-step model, ROMS-IMLE, which attains an FID of 2.56 on ImageNet.
Joint w/ @researchirag
1/6
Denys Iliash retweeted
I will be giving two workshop talks at #ECCV2026 (stay tuned) and opening a third on Functionality, Articulation, and Interaction (diliash.github.io/fai_eccv_wโฆ). Please also check out our oral paper on PADFormer (arxiv.org/abs/2608.04210) on Friday.
Denys Iliash retweeted
๐ข Excited to share our ๐๐ถ๐ณ๐ฐ๐จ๐ณ๐ข๐ฑ๐ฉ๐ช๐ค๐ด ๐๐๐๐ 2026 paper:
๐๐ฑ๐๐ฎ๐ป๐ฐ๐ฒ๐ ๐ถ๐ป ๐ก๐ฒ๐๐ฟ๐ฎ๐น ๐ฏ๐ ๐ ๐ฒ๐๐ต ๐ง๐ฒ๐
๐๐๐ฟ๐ถ๐ป๐ด: ๐ ๐ฆ๐๐ฟ๐๐ฒ๐
We organize the rapidly growing field of Neural 3D Mesh Texturing, covering how learning-based methods generate, transfer, complete, and refine texture and material appearance on 3D meshes.
We cover:
๐งฑ ๐๐ฐ๐ถ๐ฏ๐ฅ๐ข๐ต๐ช๐ฐ๐ฏ๐ด: Mesh geometry, UV mapping, differentiable rendering, PBR materials, and neural generative models.
๐งญ ๐๐ฆ๐ต๐ฉ๐ฐ๐ฅ ๐ง๐ข๐ฎ๐ช๐ญ๐ช๐ฆ๐ด: GAN-based methods, CLIP-guided and diffusion-based optimization, iterative painting, synchronized multi-view generation, and feed-forward approaches.
๐ ๐๐ท๐ข๐ญ๐ถ๐ข๐ต๐ช๐ฐ๐ฏ ๐ข๐ฏ๐ฅ ๐ฐ๐ถ๐ต๐ญ๐ฐ๐ฐ๐ฌ: Datasets, evaluation metrics, applications, limitations, and future research directions.
... and much more.
We also created a project page with a growing list of neural mesh texturing papers, making it easier to explore recent work in this fast-moving field.
๐ฐ Paper: arxiv.org/abs/2606.00137
๐ Project Page: sairajk.github.io/neural-mesโฆ
Suggestions, corrections, and missing papers are very welcome!
Joint work with my supervisors, @richardzhangsfu and @arash_mham.
#ComputerGraphics #3DVision #GenerativeAI #NeuralRendering #MeshTexturing #TextureSynthesis #DiffusionModels #GANs #VLM #3DAssets #PBR #DigitalContentCreation #GameDev #VFX #XR #DigitalHumans #Eurographics #EurographicsSTAR #CGF #Rendering
Working on perception for embodied AI? Interactive objects/scenes? Physical grounding? Submit to our #ECCV2026 workshop!
๐ diliash.github.io/fai_eccv_wโฆ
1/ ๐ข Announcing the FAI Workshop @ ECCV 2026 in Malmรถ ๐ธ๐ช
3D generative AI makes objects that look stunning โ but can they move, obey physics, and actually be interacted with?
That gap is what FAI is about. ๐งต
#ECCV2026 #ComputerVision #3DVision #GenerativeAI #Robotics
Denys Iliash retweeted
What senses does embodied AI need beyond vision?
Join us at Sense of Space Workshop @CVPR โ a full-day workshop on multi-sensory modeling for embodied intelligence.
๐
June 3rd
๐ Mile High 2C, Denver
๐ผ Posters: Hall A #275โ282
Featuring invited talks from leading researchers across robotics, vision, sensing, and embodied AI. @NimaFazeli7 @notmahi @mapo1 @wojmatusik Christian Theobalt @mangahomanga @Boyiliee @yswhynot @LingjieLiu1 @pliang279 @Ismini_L
Website: sense-of-space.github.io/
#CVPR2026 #EmbodiedAI #Robotics #MultimodalAI #TactileSensing #SpatialIntelligence
Powered by the amazing team!
@RaoFu79761158 @LiGuankfd2 @alex_kai2020
Kun He @tomhodan Ergys Ristani Jessica Yin @AntheaYLi @yining_hong Devin Murphy Ray Song
@xyz2maureen @ericyi0124 Qi Ye @YunzhuLiYZ
@haoshu_fang @ruoshi_liu Vatsal Mehta @LuoYiyue
@MengyuLearner
Introducing ๐๐ฟ๐๐ถ๐๐ฒ๐ฟ๐๐ฒ dataset: 5.4K articulated objects for embodied AI ๐ค
โข 88 categories
โข Functional part segmentation + articulation
โข Per-part material + mass
โข Procedural interiors
#CVPR2026 โ Poster 164 Session 2 & ActiVis Workshop
Website + paper below ๐งต
Website: 3dlg-hcvc.github.io/artiversโฆ
Paper: arxiv.org/pdf/2605.24403
3
142
























