佐々木竹充/SASAKI TAKERU

基準はヒト

@Azadux cgworld.jp/flashnews/01-2…

@urekat Definitely

Appleのあれかな

ほんとに？

はぇー

指先で指示

Received a dataset from a customer that optimized poorly. The realtime GNSS capture feature was used during the scan. Turns out, some receivers deliver data that's wildly off! The GNSS reports "RTK-fixed" and a 2 cm accuracy, yet the points are off by 30 meters! (reported high-accuracy points are green in the video) 🤷‍♂️ (1/3)

SLAM test at the Mall! Insight 9 handles smooth floors and complex lighting, building dense 3D point cloud maps in real time. Laptop is only for visualization. All VSLAM and point cloud processing runs fully on-board Insight 9, streamed via ROS topics! #SLAM #Robotics #PointCloud

GIS×BIM×XR：次世代デジタルツイン GISの広域コンテキストとBIMの精緻な設計データをXRで統合。 現実を写し取るReality Mappingを基盤に、設計と現状を同一空間で比較・分析可能にする。 単なる視覚化を超え、信頼性の高いデータに基づいた意思決定を支援するデジタルツイン基盤。

CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN=1

[CVPR 2026 (Highlight)] Scal3R: Scalable Test-Time Training for Large-Scale 3D Reconstruction This paper addresses the task of large-scale 3D scene reconstruction from long video sequences. Recent feed-forward reconstruction models have shown promising results by directly regressing 3D geometry from RGB images without explicit 3D priors or geometric constraints. However, these methods often struggle to maintain reconstruction accuracy and consistency over long sequences due to limited memory capacity and the inability to effectively capture global contextual cues. In contrast, humans can naturally exploit the global understanding of the scene to inform local perception. Motivated by this, we propose a novel neural global context representation that efficiently compresses and retains long-range scene information, enabling the model to leverage extensive contextual cues for enhanced reconstruction accuracy and consistency. The context representation is realized through a set of lightweight neural sub-networks that are rapidly adapted during test time via self-supervised objectives, which substantially increases memory capacity without incurring significant computational overhead. The experiments on multiple large-scale benchmarks, including the KITTI Odometry and Oxford Spires datasets, demonstrate the effectiveness of our approach in handling ultra-large scenes, achieving leading pose accuracy and state-of-the-art 3D reconstruction accuracy while maintaining efficiency.

Gaussian Splatting is steadily finding its way into e-commerce. 🧥 It's a great alternative for capturing tricky products, offering a more straightforward workflow for many use cases. Shout-out to @voxeloai for streamlining this tech and making it accessible! #GaussianSplatting

エージェントコーディングってなに

@Fumiya_Kume 間違いだったのか！

@urekat 修正されたっぽい

ま

高度なソフトウェアを作るのに必要なものがわかっているシニアのエンジニアにしか見えない領域で、実装する気力と手数だけが足りなかった分野がまず最初の競争市場になりそうですね。 コンパイラ/DB/OS/FS/VM/ブラウザとか、基盤ソフトウェアに革命が起きるのが第一段階かなーと。もう体力勝負。

zenn.dev/kun432/scraps/… そりゃそうだろ今まで捨ててたんかい

preserve_thinking