so_tech

🚀 Want to see how we do real-to-sim from a single input image? We’re releasing the code for I-Scene #CVPR2026! ✨Highlights: - Stronger scene generalization trained on randomly composed objects - Scalable data generation for downstream tasks - Supports both 3D Gaussian Splatting and mesh outputs Try the online Hugging Face demo and play with I-Scene yourself! GitHub: github.com/LuLing06/I-Sce… Demo: huggingface.co/spaces/LuLing/… Project: luling06.github.io/I-Scene-web-pa…

SuperSplat's voxel collision is a game changer for 3D Gaussian splat creators. Generate a highly accurate, inescapable voxel representation of any splat, regardless of quality. Generation is performed on the GPU, so it's super-fast. The voxel representation is tiny - only 418KB for this scene. And all this magic is free and open source, courtesy of the splat-transform library from @PlayCanvas.

📢 New paper out! We introduce Learn2Splat — a meta-learned optimizer for 3D Gaussian Splatting. ⚡ Faster early convergence than Adam 🔁 Stable over long training horizons (no LR schedules!) 🌍 Zero-shot generalization across scenes & resolutions 1/6 🧵

地理参照3D Tiles（spzの3DGS）フォルダーをWebアプリにD&Dすると、自動で地図に配置。 高度は高めになる（理由あり）ので変更し、3DGSの表示範囲を調整という作業をします。 地面解析（コライダー生成）も自動で行われアプリに記憶されます。 ここまで数十秒。全てローカルのブラウザ内処理。

首が捥げるほど同意。 泥臭さと非効率は紙一重で、何でもかんでも努力を美談にしてしまうのは違和感がある。

例えば弊社も研究開発には無限の情熱を注ぎ込むけど、実は営業やブランディングにはあんまり時間を割いてない。 営業はプル型のみ。広報もこのXと、たまのメディア取材ぐらい。 開発加速が至上命題なので、HPすら更新してない。 強烈なコミットメントも、正しいポイントに焦点を当てているかが肝要

I have just tested the released VGGT-Omega checkpoint from the VGGT team. 🚀 TL;DR: It is an absolute beast for outdoor scenes or scenes with larger depth ranges. Here are some results👇 (1/10)

Try colmap and pycolmap with prebuilt Caspar GPU solver here. 2x+ mapping speed: github.com/lyehe/build_gp…

Underwater SLAM is one of the hardest unsolved problems in robotics. And an open-source project called AQUA-SLAM is taking it on. Here's why underwater SLAM is so brutal compared to standard environments: → No GPS, zero external positioning reference → Poor visibility, cameras struggle where light doesn't reach → Sensor drift, errors compound fast with no corrections → Feature scarcity, almost nothing to anchor a map to (open sea or ocean lol) AQUA-SLAM combines stereo cameras, IMU and a Doppler Velocity Log to tackle localisation and mapping in conditions that defeat conventional systems. It ships with sensor calibration, loop closure, full ROS integration and real underwater datasets out of the box. Why does this matter beyond the lab? Underwater robotics is one of the most underloved and underinvested areas in the entire robotics stack, and one of the most commercially important. Pipeline inspection. Offshore energy infrastructure. Deep-sea mining. Environmental monitoring. Search and rescue. All of it requires robots that can navigate reliably where humans cannot go and GPS does not work. Open-source projects like this are how the field moves forward. Someone builds the foundation. The community builds on top of it. If you're working in underwater robotics, sensor fusion or visual-inertial SLAM, this one is worth exploring. VERY INTERESTING. 🌊 Here's the GH: github.com/SenseRoboticsL… ~~ ♻️ Join the weekly robotics newsletter, and never miss any news → ziegler.substack.com

open-sourcing a 3D gen toolkit for Claude Code input image → environment, meshes, physics, lighting, & audio

3 major SuperSplat upgrades shipping today 🚀 🧱 One-Click Collision Generation We've wired the SplatTransform 2.0 collision pipeline into SuperSplat Studio's backend. Open your splat ➡️ Assets panel ➡️ Hit Generate Walk-ready splats in seconds. No command line. No fuss. [1 / 4]

こいつに任せてワイは隠居するか… [2605.06607] AI CFD Scientist: Toward Open-Ended Computational Fluid Dynamics Discovery with Physics-Aware AI Agents arxiv.org/abs/2605.06607

チューリングの自動運転、銀座を走ってみました。初めてなのにすごく上手い！自分たちでもビックリしてます

論文を読んで終わりにしないローカル研究ジムResearch Dojoを作った｜Ayato Kaku zenn.dev/ayarnte/articl… #zenn

ユーザー様がAprilTag+AVATA360を使ってテストして頂き、その動画もYoutubeに掲載して頂きました。ありがとうございました。誤差は0.4%(True:5m, Actual:5.02m)とのこと。安価なMetashape Standardでも実スケールが出せます。またXMP出力も可能なので、RealityScanにエクスポートしてフォトグラでメッシュ化も可能です。

3D Gaussian SplattingをiPhone/iPadのカメラ映像に重ねて、ARでプレビュー＆録画できるアプリ「GSCam」をリリースしました！ PLYを読み込んで、位置・回転・スケールを調整しながら、その場で合成チェックできます。 VFX、AR、ロケハン、3DGS制作の検証にぜひ。 apps.apple.com/jp/app/gscam/i… #3DGS #GaussianSplatting #AR #VFX #iPhone

Working Python code for SLAM, path planning, and localization github.com/AtsushiSakai/P…

Course materials for robotic mapping and SLAM with ROS github.com/ariarobotics/r…

@tom_doerr ros mapping is its own kind of madness but ngl this looks well structured building that grid from scratch is the real flex here

画像生成モデルがVisionタスクの汎用性も持つよ&むちゃ強いよっていう研究で、内容がすごすぎて横転 SAM3, Depth Anythingを上回るらしい arxiv.org/abs/2604.20329… 4/22 by Google Deepmind