# Datameister

> AI research and deployment lab building production systems for spatial and visual intelligence. Computer vision, 3D generative AI, and MLOps from Ghent, Belgium.

Datameister is an AI research and deployment lab in Ghent, Belgium. We build and ship production systems for spatial and visual intelligence — computer vision, 3D generative AI, point-cloud perception, custom LLMs, and MLOps.

Every blog post and glossary entry below also has a plain-Markdown twin at `index.md` under its URL (e.g. https://datameister.ai/glossary/computer-vision/index.md). The full text of all of them is bundled in https://datameister.ai/llms-full.txt, and new posts appear on the RSS feed at https://datameister.ai/blog/feed.xml.

## Core pages

- [Research & Core Tech](https://datameister.ai/what-we-do/): Datameister's research and core tech: proprietary 3D AI engine, hardened academic research, and core technology tracks across capture, optimization, and creation.
- [Process](https://datameister.ai/how-we-do-it/): Discovery workshops, feasibility POCs, and full product deployment. See how Datameister delivers custom AI solutions with predictable pricing and full IP ownership.
- [Team](https://datameister.ai/team/): Meet the founders and engineers behind Datameister - an AI lab bridging frontier research and production deployment, founded in Ghent, Belgium.
- [Careers](https://datameister.ai/jobs/): Join Datameister's team in Ghent, Belgium. Open roles to help ship production AI systems for spatial and visual intelligence.
- [Contact](https://datameister.ai/contact-us/): Get in touch with Datameister for custom AI API development, visual intelligence consulting, and deployment modernization. We respond within one business day.
- [AI & 3D Glossary](https://datameister.ai/glossary/): Key concepts in computer vision, 3D generative AI, MLOps, and spatial intelligence - explained by the engineers who build production systems with them.
- [Blog](https://datameister.ai/blog/): Technical articles on 3D generative AI, computer vision, MLOps, and visual intelligence from the Datameister research and deployment lab.

## Glossary

- [Computer Vision](https://datameister.ai/glossary/computer-vision/): Computer vision is the AI field that trains machines to interpret images and video - powering object detection, tracking, segmentation, and scene understanding.
- [3D Generative AI](https://datameister.ai/glossary/3d-generative-ai/): 3D generative AI uses deep learning to create 3D meshes, textures, and scenes from text prompts or images - accelerating content work once done by hand.
- [Gaussian Splatting](https://datameister.ai/glossary/gaussian-splatting/): Gaussian splatting renders photorealistic 3D scenes in real time from millions of 3D Gaussian primitives - faster than NeRFs, with no neural inference.
- [Neural Radiance Field (NeRF)](https://datameister.ai/glossary/neural-radiance-field/): A neural radiance field (NeRF) is a neural network that learns a 3D scene from posed photos, enabling photorealistic novel-view synthesis via volume rendering.
- [Point Cloud](https://datameister.ai/glossary/point-cloud/): A point cloud is a set of 3D points from LiDAR, depth cameras, or photogrammetry - the raw spatial data behind scanning, segmentation, and reconstruction.
- [Object Detection](https://datameister.ai/glossary/object-detection/): Object detection is the computer vision task of locating and classifying objects in images and video with bounding boxes - from YOLO to detection transformers.
- [Retopology](https://datameister.ai/glossary/retopology/): Retopology rebuilds a 3D model's surface with clean, efficient polygon topology - turning dense scans and sculpts into production-ready meshes with proper UVs.
- [MLOps](https://datameister.ai/glossary/mlops/): MLOps is the practice of deploying, monitoring, and maintaining machine learning models in production - bridging model development and reliable operation.
- [Digital Twin](https://datameister.ai/glossary/digital-twin/): A digital twin is a virtual replica of a physical object or environment, kept in sync with real-world data for monitoring, simulation, and analysis.
- [LiDAR](https://datameister.ai/glossary/lidar/): LiDAR measures distance with laser pulses to build dense, accurate 3D point clouds of the environment - the backbone of 3D scanning, mapping, and perception.

## Blog

- [Parametric Scene Reconstruction: From LiDAR Scan to Editable BIM](https://datameister.ai/blog/parametric-scene-reconstruction-lidar-to-editable-bim/): Turn LiDAR point clouds into editable CAD: why meshes fail when you edit, and how Datameister reconstructs queryable walls and floors in minutes.
- [Datameister #1 in Intrinsic's robotic-manipulation challenge qualifier](https://datameister.ai/blog/intrinsic-ai-for-industry-challenge-qualifying-first/): Datameister ranked #1 of ~160 teams in the Intrinsic AI for Industry Challenge qualifier, a robotic cable-insertion benchmark judged by Google DeepMind and NVIDIA.
- [Autonomous from Scratch: Building Datameister's Physical AI Foundation](https://datameister.ai/blog/autonomous-from-scratch-physical-ai-foundation/): How Datameister built an autonomous navigation and data-capture stack from scratch on a Unitree GO2, NVIDIA Jetson, and Livox MID-360, unified under ROS 2.
- [The Monorepo as AI Factory](https://datameister.ai/blog/the-monorepo-as-ai-factory/): How Datameister runs on a single monorepo: default knowledge-sharing, faster developer velocity and a consistent quality bar as the team scales.
- [Indoor Semantic Segmentation of Point Clouds: From LiDAR Capture to Real-World Use](https://datameister.ai/blog/indoor-semantic-segmentation-lidar-pipeline/): Turn LiDAR point clouds into indoor spatial intelligence: why outdoor models fail indoors and how to build a reliable segmentation pipeline.
- [From Studio to Robot: Well-Integrated 3D Generation](https://datameister.ai/blog/blender-controlled-generative-3d-generation-addon/): A Blender add-on for controlled 3D generation, letting creators steer results with precise go-zones and no-go zones inside the modeling environment.
- [Why the Future of 3D Generative AI is Programmatic](https://datameister.ai/blog/why-the-future-of-3d-generative-ai-is-programmatic/): How programmatic 3D generative AI reshapes workflows by producing editable, script-based assets instead of opaque static meshes.
- [Three challenges in finetuning Trellis](https://datameister.ai/blog/three-challenges-in-finetuning-trellis/): Practical lessons from finetuning Trellis for image-conditioned 3D generation: data quality, memory bottlenecks and overfitting.
- [Trellis 2: Scaling 3D Generation with Improved Efficiency and Control](https://datameister.ai/blog/trellis-2-controlled-3d-generation/): How Trellis 2 uses native 3D Omni-Voxels and efficient latent compression to enable scalable, physically grounded 3D generation.
- [Why DETRs are replacing YOLOs for real-time object detection](https://datameister.ai/blog/detection-transformers-real-time-object-detection/): Real-time Detection Transformers as a superior, Apache-2.0-licensed alternative to YOLOs for object detection: RT-DETR, D-FINE and DEIMv2.
- [Automated Retopology for 3D Assets](https://datameister.ai/blog/ai-automated-retopology/): Retopomeister is a prototype AI tool for automated 3D asset retopology, preserving anatomy, symmetry and UVs while cutting manual work.
- [Datameister at SIGGRAPH 2025: Insights and Trends](https://datameister.ai/blog/datameister-at-siggraph-2025-insights-and-trends/): SIGGRAPH 2025 recap: practical AI in computer graphics, covering 3D generation, simulation, pipeline fit and takeaways studios can apply now.
- [Constraint-Aware 3D Generative Design: Editable, Iterable, Manufacturable](https://datameister.ai/blog/constraint-aware-3d-generative-design-editable-iterable-manufacturable/): Constraint-aware 3D generation for industrial design: explore variations while hard points stay fixed, via masked generation and differentiable rendering.
- [Datameister Turns Two 🎂](https://datameister.ai/blog/datameister-two-year-anniversary/): Datameister's two-year anniversary: tour the new Ghent HQ, meet the doubled AI team, explore platform upgrades and catch the celebration recap.
- [Constraint-driven 3D Generative AI - Computational Design Symposium](https://datameister.ai/blog/constrained-driven-3d-generative-ai-computational-design-symposium/): A CDFAM Amsterdam preview: how Datameister embeds real engineering constraints into generative design to cut design lock-in and speed iteration.
- [Datameister Platform: Accelerating AI Deployment for Visual Data](https://datameister.ai/blog/datameister-platform-accelerating-ai-deployment-for-visual-data/): How the Datameister Platform accelerates MLOps for visual AI: fast deployment, debugging and cost-efficient GPU scaling for image, video and 3D.
- [3D Generative AI: Image-based 3D reconstruction](https://datameister.ai/blog/3d-generative-ai-image-based-3d-reconstruction/): How image-based 3D reconstruction evolved from NeRFs to Trellis, compared against Rodin, Tripo, SPAR3D and Hunyuan3D-2.
- [Datameister @ECCV 2024: Building a foundation](https://datameister.ai/blog/datameister-at-eccv-2024-building-a-foundation-part-i/): Highlights from ECCV 2024 in Milan: 3D Gaussian Splatting advances like WildGaussians and Gaussian Frosting reshaping real-time rendering.
- [Celebrating our first year](https://datameister.ai/blog/celebrating-our-first-year/): Datameister marks its first anniversary with a growing team and an evening of drinks, BBQ and music at Zebrabeach in Ghent.
- [Reflecting on the first AI for Digital Arts/Entertainment/Game Dev meetup](https://datameister.ai/blog/reflecting-on-the-first-ai-for-digital-artsentertainmentgame-dev-meetup/): A recap of the first AI for Digital Arts, Entertainment and Game Dev meetup in Ghent, where techies and creatives explored the future of game development.
- [AI-Driven Breakthroughs in Image-Based Rendering: Light Fields, SMoE, Gaussian Splatting, NeRFs and beyond](https://datameister.ai/blog/ai-driven-breakthroughs-image-based-rendering/): An overview of AI-driven advances in image-based rendering: light fields, SMoE, Gaussian Splatting and NeRFs.
- [Making the case for custom LLMs and custom LLM deployments](https://datameister.ai/blog/making-the-case-for-custom-llms-and-custom-llm-deployments/): Why custom LLMs and self-deployed models win: gain control, build IP, cut costs and protect your data, with Datameister.
- [Meet The Meisters](https://datameister.ai/blog/meet-the-meisters/): Meet the founders of Datameister, Axel Vlaminck and Ruben Verhack: why they started a deep-tech AI lab in Ghent and what they build, from 3D AI to Physical AI.