Built at Simform

Nira App

High-performance 3D CAD and point cloud viewer enabling seamless collaboration on complex engineering datasets

Overview

Nira App is a cutting-edge web-based 3D visualization platform designed for engineers, architects, and construction professionals working with CAD models and point cloud datasets. The application handles massive 3D files—from architectural BIM models to LiDAR point clouds with millions of points—delivering smooth, interactive visualization directly in web browsers without requiring specialized software or high-end workstations.

Built with React, Three.js, and custom WebGL rendering pipelines, Nira provides professional-grade 3D viewing capabilities accessible from any device. The platform features advanced measurement tools, cross-section analysis, model annotation, real-time collaboration, and seamless integration with enterprise CAD workflows. The application optimizes viewing of files ranging from 10MB to 10GB+ through sophisticated streaming and level-of-detail techniques.

This technically demanding project required expertise in geometric algorithms, GPU optimization, large-scale data streaming, memory management, and building intuitive interfaces for complex engineering workflows used by professional users with high expectations for performance and accuracy.

Role: 3D Developer · Simform
Duration: 11 months
Team: Cross-functional team of 2 frontend engineers, 1 graphics engineer, 2 backend engineers, 2 designers, and 1 product manager

The Challenge

Engineering and construction industries rely on massive 3D datasets that traditionally require expensive specialized software and high-end workstations. Cloud-based alternatives suffered from poor performance, limited functionality, or security concerns around sensitive project data. The challenge was to build a web-based viewer capable of handling industrial-scale CAD models and point clouds with millions of vertices while maintaining smooth 60fps interaction, providing professional measurement tools, and ensuring data security for enterprise clients.

Render CAD models with 10M+ polygons and point clouds with 100M+ points maintaining smooth 60fps performance
Implement precision measurement tools accurate to millimeter-level for engineering validation and quality control
Build progressive streaming system loading multi-gigabyte 3D files without browser crashes or excessive memory usage
Support diverse CAD formats (IFC, Revit, STEP, STL, E57, LAS) with accurate conversion and geometry preservation

What I built

High-Performance Point Cloud Rendering

Engineered custom point cloud rendering engine handling 100M+ points with real-time interaction using WebGL2 and compute shaders. Implemented octree-based spatial indexing for hierarchical level-of-detail, rendering only visible points based on camera position and zoom level. Built adaptive point sizing system dynamically adjusting point dimensions maintaining visual continuity across zoom ranges. Created custom shader programs handling point cloud coloring modes (RGB, intensity, classification, elevation) with smooth transitions. The system achieves 60fps performance with datasets that would crash traditional viewers.

100M+ points rendered at 60fps

CAD Model Visualization

Developed sophisticated CAD rendering pipeline supporting complex BIM models with accurate material representation, assembly hierarchies, and metadata preservation. Built optimized mesh processing handling tessellation, normal generation, and edge detection for technical visualization. Implemented multiple display modes including shaded, wireframe, hidden line, and X-ray views essential for engineering workflows. Created dynamic cross-section tools with real-time clipping planes allowing interior inspection of complex assemblies. The system handles building-scale models with thousands of components while maintaining interactive performance.

Support for 15+ CAD formats

CAD model visualization with assembly hierarchy

Precision Measurement Tools

Implemented comprehensive measurement suite providing millimeter-level accuracy for professional validation. Built distance, area, and volume calculation tools with automatic snapping to geometry features (vertices, edges, faces). Created angle measurement for structural analysis, path measurement for HVAC/piping routing, and elevation profiling for terrain analysis. Implemented coordinate system management supporting multiple reference frames and unit conversions. The measurement system includes annotation capabilities with persistent markers, screenshots, and export to PDF reports.

Millimeter-level measurement accuracy

Environment and Lighting Controls

Developed sophisticated visualization controls allowing users to evaluate models under different environmental conditions. Built day/night lighting simulation with accurate sun position calculations based on project location and time. Implemented weather visualization modes including rain, fog, and varying sky conditions for contextual presentation. Created custom HDRI environment maps with dynamic exposure adjustment and shadow quality controls. The system helps architects visualize buildings under real-world conditions and engineers evaluate visibility and aesthetics.

12 environmental presets with real-time switching

Dynamic lighting and environmental controls

Progressive Streaming Architecture

Architected sophisticated streaming system loading multi-gigabyte 3D files efficiently without browser limitations. Implemented hierarchical data structure with progressive refinement—low-resolution model appears instantly while high-detail geometry streams in background. Built intelligent prioritization loading visible areas first based on camera frustum and user interaction patterns. Created custom binary format with optimal compression reducing transfer sizes by 80%. The system includes resume capability for interrupted downloads and works efficiently over varying network conditions.

Sub-2s initial visualization for 2GB+ files

How it works

Built on React with TypeScript for type-safe component development. Custom Three.js rendering pipeline extends base functionality with specialized loaders, shaders, and optimization techniques. WebGL2 enables compute shader-based point cloud rendering and advanced rendering techniques. WebWorkers handle geometry processing, file parsing, and spatial indexing off main thread. The architecture implements custom memory management with object pooling, geometry instancing, and automatic LOD generation. Backend services provide format conversion (15+ CAD formats to optimized web format), spatial indexing, and secure file storage. WebSocket connections enable real-time collaboration features. Testing infrastructure includes visual regression testing and performance profiling across device tiers.

Stack

React
Three.js
WebGL
CAD Integration
TypeScript

Hard parts

Managing GPU memory when loading multi-gigabyte 3D files in browser environment with strict memory limits
Achieving 60fps performance with 100M+ point clouds requiring billions of GPU operations per frame
Converting diverse CAD formats (IFC, Revit, STEP, etc.) to web-friendly format while preserving geometry accuracy and metadata
Implementing millimeter-accurate measurements on floating-point geometric data with inherent precision limitations

How I solved them

Developed streaming LOD system with automatic memory management loading/unloading geometry based on visibility and camera distance
Built custom octree-based culling with compute shader point rendering utilizing GPU parallelism processing 10M+ points per frame
Created backend conversion service using Open Cascade and IFC.js libraries with validation ensuring geometric accuracy within 0.1mm tolerance
Implemented double-precision camera calculations and high-precision measurement algorithms using integer-based arithmetic reducing floating-point errors

Results

200+
Engineering Firms: Using Nira for CAD visualization; -60%
Review Cycle Time: Through instant web access; 50,000+
Files Processed: 500TB+ of 3D data

Deployed by 200+ engineering firms and construction companies eliminating $2M+ annual CAD software licensing costs
Enabled 5,000+ engineers to collaborate on 3D models without specialized software or hardware investments
Reduced design review cycles by 60% through instant web-based access eliminating file sharing bottlenecks
Processed 50,000+ CAD files totaling 500TB+ of 3D data with 99.8% conversion success rate
Achieved 4.8-star rating from engineering professionals citing intuitive interface and reliable performance

Keep looking

Previous project

Golden City

Next project

MyVirtualStudio

Back to all projects