Product Design Studio

Converts hand-drawn 2D sketches into editable 3D models using computer vision and deep learning inference. The system likely employs a multi-stage pipeline: sketch image preprocessing (normalization, line extraction), feature detection to identify geometric primitives (circles, lines, curves), 3D shape inference using trained neural networks to predict depth and volume from 2D line patterns, and mesh generation to produce an editable 3D representation. The output is a parametric or mesh-based 3D model that can be further refined within the editor.

Provides a multi-user design environment where team members can simultaneously view, edit, and comment on 3D models with live cursor tracking and presence indicators. The system likely uses WebSocket or similar real-time protocol for synchronizing model state, viewport changes, and annotations across connected clients. Operational transformation or conflict-free replicated data types (CRDTs) likely manage concurrent edits to prevent conflicts. Presence awareness (showing who is viewing/editing and where their cursor is) reduces communication overhead and enables natural collaboration without explicit turn-taking.

Allows users to edit and refine 3D models generated from sketches through a parametric or direct-manipulation interface. Users can adjust dimensions, proportions, curves, and geometric features post-conversion. The system likely maintains an editable representation (parametric constraints, mesh deformation, or feature-based modeling) that allows non-destructive changes. Real-time 3D viewport updates provide immediate visual feedback as parameters are adjusted, enabling rapid iteration without re-running the sketch-to-3D conversion.

Exports refined 3D models from Pietra to industry-standard file formats (GLTF, OBJ, STEP, STL, FBX, or similar) for downstream use in CAD, rendering, 3D printing, or manufacturing workflows. The export pipeline likely performs format-specific optimizations (e.g., mesh decimation for OBJ, STEP assembly generation, STL repair for 3D printing). Export may be available through the UI or API, with options for quality/resolution trade-offs and metadata preservation.

Enables team members to leave comments, annotations, and feedback directly on 3D models at specific locations or on model elements. Comments are likely threaded (allowing replies and discussion) and spatially anchored to the 3D geometry or viewport. The system tracks comment status (resolved, pending, etc.) and may notify relevant team members of new feedback. Annotations may include text, sketches, or reference images to clarify design intent or issues.

Provides workspace and project management features for organizing multiple design files, versions, and team assets. Users can create projects, organize models into folders or collections, and manage access permissions for team members. The system likely tracks file metadata (creation date, last modified, owner) and may support basic versioning or snapshots. Asset libraries or templates may be available for reuse across projects.

Provides AI-generated design suggestions, variations, or optimizations based on the current model and design context. The system may suggest proportional adjustments, alternative forms, or design refinements using trained models or heuristics. Suggestions are likely presented as alternatives or overlays in the 3D viewport, allowing users to accept, reject, or iterate on recommendations. This capability may leverage computer vision and generative models to propose design improvements without explicit user input.

Implements a freemium business model where core sketch-to-3D conversion and basic editing are available for free, with advanced features (export formats, collaboration limits, storage, API access) restricted to paid tiers. The system likely tracks usage metrics (file count, storage, collaborators) and enforces soft limits (e.g., limited exports per month) or hard limits (e.g., max 3 collaborators) on free accounts. Paid tiers unlock additional features and higher quotas.