Deploy JSON-LD v3 pipeline from jjohare/logseq @ 3d7eca0398d798bb4ca32a22dafa98c856c8e4ef jjohare/logseq@3d7eca0

Author: jjohare

api/markdown/3-d-content-creation.md

@@ -0,0 +1,14 @@
+- ### Definition
+  - 3D Content Creation is the multidisciplinary practice of producing three-dimensional digital assets — from individual objects and characters to complete environments — using [[3D Modeling Software]], sculpting tools, texturing suites, and [[3D Rendering Engine]] platforms. It integrates artistic decisions around form, colour, and material with technical constraints such as polygon budgets, level-of-detail hierarchies, and [[Physically Based Rendering]] workflows. The output feeds into [[3D Content Pipeline]] systems that target real-time engines, film compositing, [[XR Applications]], and [[Digital Twin Creation]].
+
+- ### Relationships
+  - 3D Content Creation uses [[3D Modeling Software]] such as Blender, Autodesk Maya, and ZBrush as primary tools, and relies on [[3D Rendering Engine]] technology to preview and finalise output. [[Physically Based Rendering]] underpins modern material authoring, ensuring assets respond convincingly to lighting. The discipline enables [[3D Content Pipeline]] automation, powers [[Digital Twin Creation]] workflows, and supplies the assets consumed by [[XR Applications]]. It is closely related to [[3D Asset]] management, emerging [[3D Generation]] techniques, and [[Motion Capture]] for character animation.
+
+- ### Content
+  - 3D content creation emerged from computer-aided design and film visual-effects research in the 1970s–1980s, maturing through the introduction of tools such as Wavefront Technologies' software (predecessor to Maya), 3D Studio, and the open-source Blender project in 1995. Early pipelines were entirely hand-crafted, requiring artists to manually place vertices and paint textures. The discipline grew rapidly as the games industry demanded real-time 3D assets in the 1990s and cinema adopted digital characters and environments through the 2000s.
+
+  - Modern 3D content creation is built around polygon and subdivision-surface modelling, parametric modelling with non-destructive workflows, and high-resolution sculpting later baked down to real-time meshes. UV unwrapping maps surface coordinates for texture application, while physically based rendering material systems encode base colour, metallic, roughness, and normal data in standardised maps. Real-time engines such as Unity and Unreal Engine consume these assets directly, with level-of-detail generation and occlusion culling managing runtime performance.
+
+  - The ecosystem includes dedicated tooling at each stage: Autodesk Maya and 3ds Max for rigging and animation, ZBrush and Mudbox for sculpting, Substance Painter and Quixel Mixer for texturing, Houdini for procedural geometry and visual effects, and Blender as an all-in-one open-source alternative. Asset interchange relies on standard formats including FBX, glTF 2.0, and USD, allowing cross-application workflows. Cloud-based collaboration platforms and version-control integrations have modernised studio pipelines to handle geographically distributed teams.
+
+  - Through 2024–2025, AI-assisted generation is reshaping the discipline: text-to-3D and image-to-3D models can produce draft geometry and textures in seconds, with [[3D Generation]] tools such as TripoSR, Stable Zero123, and Meshy integrated into production pipelines. Photogrammetry and [[3D Scanning]] further accelerate asset creation by capturing real-world geometry. The convergence of procedural generation, AI drafting, and human artistry is raising both quality ceilings and throughput, making high-fidelity 3D content accessible to smaller teams and individual creators.

api/markdown/3-d-generation.md

@@ -0,0 +1,14 @@
+- ### Definition
+  - 3D Generation is the process of algorithmically producing three-dimensional geometry, textures, and scene structure — often from text prompts, single images, or partial observations — using generative machine learning models. Core techniques include [[NeRF]]-based implicit representations, [[Gaussian Splatting]] for explicit radiance fields, and [[Diffusion Model]] architectures adapted for 3D data. The field intersects with [[Text-to-3D]] synthesis, image-conditioned reconstruction, and [[Neural 3D Generation]] from large-scale pretraining, collectively positioning 3D generation as a transformative layer within [[3D Content Creation]] pipelines.
+
+- ### Relationships
+  - 3D Generation uses [[Diffusion Model]] frameworks (such as Score Distillation Sampling applied to NeRFs) and [[Gaussian Splatting]] representations to produce geometry and appearance. It enables [[3D Content Creation]] workflows by supplying draft assets, accelerates [[Digital Twin Generation]] from imagery, and fuels [[Game Asset Generation]] pipelines. The field is closely related to [[Text-to-3D]] interfaces, [[Neural 3D Generation]] research, and [[3D Reconstruction]] from multi-view observations.
+
+- ### Content
+  - Interest in automated 3D generation dates to procedural modelling tools of the 1990s, but the modern deep-learning era began with generative adversarial networks applied to voxel grids and point clouds around 2016–2018. DeepMind and academic labs demonstrated shape generation from latent spaces, while occupancy networks and signed-distance-function networks offered continuous implicit representations. These early methods struggled with quality and diversity outside narrow object categories.
+
+  - The pivotal technical shift came with Neural Radiance Fields (NeRF, 2020) and subsequent work on instant training (Instant-NGP, 2022), which demonstrated high-fidelity novel-view synthesis from photographs. Score Distillation Sampling (DreamFusion, 2022) then coupled 2D diffusion priors with NeRF optimisation to enable text-conditioned 3D generation without 3D training data. Gaussian splatting (2023) further advanced the field by replacing neural implicit fields with explicit 3D Gaussians, yielding real-time rendering speeds and simpler editing.
+
+  - The ecosystem diversified rapidly: TripoSR, Zero123++, Stable Zero123, and Shap-E produced single-image-to-3D models; OpenAI and Stability AI released text-to-3D APIs; and commercial tools such as Meshy, Luma AI, and Kaedim targeted production artists. Integration with Blender via add-ons and with Unreal Engine through importer plugins lowered adoption barriers. Point-cloud and mesh generation networks trained on Objaverse — a dataset of over 800k 3D assets — gave models broad object-level coverage.
+
+  - In 2024–2025, 3D generation is transitioning from research curiosity to production tool. Models such as Trellis (2024) and CraftsMan produce detailed meshes with UV-ready topology. Scene-level generation — producing entire environments rather than single objects — remains an open challenge, with work on compositional generation and world-model approaches. Industry adoption is accelerating in gaming, film pre-visualisation, e-commerce, and spatial computing, with major studios incorporating AI-drafted geometry into their pipelines while retaining human artist refinement.

api/markdown/3-d-modeling-software.md

@@ -0,0 +1,14 @@
+- ### Definition
+  - 3D Modeling Software provides interactive environments for constructing and editing three-dimensional geometry, ranging from hard-surface polygon editing to organic sculpting and parametric procedural construction. Tools such as Blender, Autodesk Maya, 3ds Max, Cinema 4D, and ZBrush each occupy specific niches within the [[3D Content Creation]] workflow. They operate on data structures including polygon meshes, subdivision surfaces, and NURBS, and typically export to standard [[3D File Format]] containers such as FBX, OBJ, and USD for consumption by [[Game Engine]] or compositing pipelines.
+
+- ### Relationships
+  - 3D Modeling Software enables the production of [[3D Content Creation]] assets and exports [[3D Asset]] files consumed by downstream systems. It uses [[3D File Format]] standards for interchange, integrates [[Physically Based Rendering]] material workflows and [[Shader]] authoring. It relates closely to [[CAD Software]] for engineering-oriented geometric design, [[BIM Software]] for architectural digital twins, and [[Game Engine]] environments that consume its output.
+
+- ### Content
+  - The history of 3D modelling software begins with Evans & Sutherland's pioneering work in the 1960s and the subsequent commercialisation of workstation-class tools in the 1980s. Wavefront Technologies and Alias Systems developed what became Autodesk Maya, while Kinetix produced 3D Studio (later 3ds Max). These tools dominated film and games pipelines through the 1990s. Blender was launched as free software in 1995, growing from a small Dutch studio tool into the leading open-source 3D application by the 2010s.
+
+  - Technically, 3D modelling software operates on scene graphs — hierarchical structures of meshes, lights, cameras, and material assignments. Core editing paradigms include box modelling (extruding and bevelling from primitive shapes), edge-loop sculpting, retopology of high-poly sculpts to game-ready meshes, and non-destructive modifier stacks. Parametric modelling workflows allow constraint-driven geometry that updates automatically when dimensions change, bridging creative and engineering use cases. UV unwrapping tools map 3D surfaces to 2D texture space for painting and baking.
+
+  - The ecosystem is highly fragmented by use case: Autodesk Maya and Houdini dominate visual-effects and animation studios; Blender has become the community standard for independent artists and is growing in studios; ZBrush and Mudbox handle high-frequency sculpting; Maxon Cinema 4D is widely used in motion graphics; and Rhino 3D and SolidWorks serve product design and architecture. Substance Painter and Quixel Mixer handle the texturing stage downstream. Interoperability between these tools depends on shared formats and translation layers.
+
+  - In 2024–2025, AI integration is accelerating transformation of the modelling workflow. AI-assisted retopology tools, generative geometry nodes, text-to-mesh features, and automated UV unwrapping are being incorporated into Blender's add-on ecosystem and commercial tools. Cloud-based collaborative modelling — pioneered by Autodesk's cloud platforms and NVIDIA Omniverse — allows multi-user editing of shared scenes. Real-time viewport previews using hardware ray-tracing bring final-quality feedback directly into the modelling environment, reducing iteration cycles.

api/markdown/3-d-perception.md

@@ -0,0 +1,14 @@
+- ### Definition
+  - 3D Perception is the process by which a computing system builds a structured understanding of a three-dimensional environment from raw sensor inputs. It combines [[Depth Sensing]] from stereo cameras, time-of-flight sensors, and LiDAR with [[Sensor Fusion]] to produce reliable spatial representations. Outputs include [[Point Cloud]] models, depth maps, occupancy grids, and semantic scene graphs used by downstream modules for navigation, manipulation, and [[6DoF Tracking]] in XR and robotics contexts. Deep learning architectures — including 3D convolutional networks, point-cloud transformers, and multi-view networks — have become central to state-of-the-art 3D perception pipelines.
+
+- ### Relationships
+  - 3D Perception uses [[Depth Sensing]] hardware and [[Sensor Fusion]] algorithms to combine complementary modalities, and processes [[Point Cloud]] data as a primary representation. It enables [[Simultaneous Localisation and Mapping]] for environment mapping, supports [[6DoF Tracking]] for XR and robotics, and drives [[Object Detection]] in autonomous systems. The discipline is closely related to [[3D Scanning]] for digitisation, [[Pose Estimation]] for object and human body understanding, and [[Semantic Segmentation]] for scene labelling.
+
+- ### Content
+  - Early computational approaches to 3D perception, dating to the 1970s and 1980s, focused on stereo vision — recovering depth by matching corresponding points across two camera images — and structured-light depth sensors. The introduction of time-of-flight cameras and rotating LiDAR units in the 2000s provided richer depth data for autonomous vehicles and robotics. The Microsoft Kinect (2010) democratised real-time depth sensing for consumer applications and fuelled academic research in human body tracking and indoor reconstruction.
+
+  - Modern 3D perception pipelines are predominantly driven by deep learning. PointNet (2017) demonstrated that neural networks operating directly on unordered point clouds could achieve strong classification and segmentation results. Subsequent architectures including PointNet++, DGCNN, VoxNet, and transformer-based networks such as Point Transformer have progressively improved accuracy and efficiency. Multi-modal fusion — combining RGB images with LiDAR point clouds — has become the standard for autonomous-driving perception, enabling reliable detection across lighting and weather conditions.
+
+  - Key applications include autonomous vehicle perception (object detection, lane segmentation, and free-space estimation), robotic manipulation (grasp pose estimation and bin-picking), augmented reality (AR surface detection and anchor placement), and industrial inspection (defect localisation on surfaces). Benchmarks such as KITTI, nuScenes, and ScanNet provide standardised evaluation data. Real-time constraints demand efficient model architectures and hardware acceleration on GPUs and specialised neural processing units.
+
+  - Through 2024–2025, the field is advancing on several fronts: large-scale pretraining on synthetic and real-world data improves generalisation; occupancy prediction networks used in Tesla Autopilot and other systems replace explicit object detection with dense volumetric output; and 4D perception — tracking objects through time — is maturing. Integration with foundation models enables open-vocabulary 3D recognition, while edge deployment on XR headsets such as the Apple Vision Pro and Meta Quest demonstrate that high-fidelity 3D perception is achievable within compact wearable form factors.

api/markdown/3-d-scanning.md

@@ -0,0 +1,14 @@
+- ### Definition
+  - 3D Scanning is the systematic capture of real-world geometry using active or passive sensing methods to produce digital [[Point Cloud]] or mesh representations. Active methods include structured light (projecting known patterns and reading deformation), time-of-flight [[3D LiDAR]], and laser triangulation scanners, while passive methods rely on [[Photogrammetry]] — computing geometry from multiple overlapping images. The resulting data enables [[3D Scene Reconstruction]] for documentation, reverse engineering, film production, and [[Digital Twin Creation]]. [[Point Cloud Processing]] pipelines refine raw scan data into usable geometry through filtering, registration, and meshing stages.
+
+- ### Relationships
+  - 3D Scanning uses [[3D LiDAR]] for large-scale outdoor capture, [[Photogrammetry]] for image-based reconstruction, and [[Depth Sensor]] hardware for close-range scanning. It enables [[3D Scene Reconstruction]] of environments, supports [[Digital Twin Creation]] from physical assets, and produces data for [[Point Cloud Processing]]. The technology is closely related to the [[Point Cloud]] data format, [[Structure-from-Motion]] algorithms, and [[3D Perception]] systems that interpret scanned representations in real time.
+
+- ### Content
+  - Industrial 3D scanning began with coordinate measuring machines (CMMs) in the 1960s and early laser triangulation scanners developed for aerospace quality control in the 1970s–1980s. The first commercial structured-light scanners appeared in the 1990s, enabling sub-millimetre accuracy for reverse engineering and medical prosthetics. Ground-penetrating and airborne LiDAR scanning were adopted for large-scale topographic survey and archaeological documentation from the late 1990s onwards.
+
+  - Modern 3D scanning technologies span a resolution and range continuum: handheld structured-light scanners (e.g., Artec Eva) capture objects up to a few metres with sub-millimetre precision; terrestrial LiDAR systems (e.g., Leica BLK, FARO Focus) scan architectural spaces to centimetre accuracy at ranges of hundreds of metres; and aerial/mobile mapping rigs mount LiDAR alongside GNSS/IMU for kilometre-scale terrain capture. Photogrammetry using drone imagery and consumer cameras has become a high-throughput complement, producing textured meshes from photo collections with software such as Agisoft Metashape and RealityCapture.
+
+  - Applications of 3D scanning span heritage preservation (scanning of artefacts and monuments before restoration or replication), film and games production (body and face scanning for digital doubles), construction and BIM (as-built capture for project verification), automotive and aerospace quality control, and medical imaging (orthopaedic implant fitting and surgical planning). The scan-to-BIM workflow, which converts point-cloud data into parametric building information models, has become a standard practice in architecture and facilities management.
+
+  - In 2024–2025, smartphone-class depth sensing (using LiDAR on iPhone and iPad Pro, and structured-light face scanners) has democratised basic 3D scanning. Gaussian splatting and NeRF-based reconstruction tools allow high-quality captures from video sequences without dedicated scanning hardware. AI-driven hole-filling and noise-reduction are incorporated into standard post-processing tools. Integration between scanning platforms and digital-twin management software is maturing, and real-time collaborative scanning workflows are enabling distributed capture projects.