David Egwu
Turning Complex Crafts into AI Training Data & viral content
New to Contra
David is ready for their next project!
The Challenge: > Transforming intricate, hand-crafted physical art into high-performance digital assets for 3D environments.
The Solution: > I developed a workflow that captures the geometric complexity of physical origami and translates it into an optimized 3D mesh. This project highlights the potential for brands to take handmade physical products and deploy them into AR (Augmented Reality) shopping experiences.
Technical Focus:
•Geometric Fidelity: Maintaining the sharp edges and folding patterns unique to paper art.
•Lightweight Mesh: Ensuring the final .glb file is under 5MB for fast mobile web loading.
•Cross-Platform Ready: Tested for iOS Quick Look and Android Scene Viewer.
1
1
47
The goal of this project is to demonstrate the full pipeline of converting a static 2D portrait into a high-fidelity, AR-ready 3D asset. In the evolving landscape of 2026, where AI and spatial computing require precise depth and geometry, this project serves as a benchmark for Digital Twin accuracy and optimized topology.
The Technical Process🛠️
I executed this project in three distinct phases to ensure maximum quality for technical integration:
•Phase 01: Geometry & Wireframe: Established clean mesh topology to ensure the asset is lightweight yet structurally sound for real-time rendering.
•Phase 02: Surface Topology (Clay Render): Focused on organic form and volumetric accuracy, stripping away textures to verify the integrity of the 3D sculpt.
•Phase 03: Texturing & PBR Mapping: Applied high-resolution textures and Physically Based Rendering (PBR) materials to achieve a realistic, "Digital Twin" finish.
The Deliverable✅
The final output is a fully optimized .glb asset, cross-compatible with Augmented Reality (AR) platforms, AI vision training datasets, and interactive web environments.
Key Skills Applied✅
•3D Reconstruction: Turning 2D imagery into 3D volume.
•Asset Optimization: Ensuring low-latency performance for mobile and web.
•Technical Visuals: Providing wireframe and clay audits for quality assurance.
1
1
51
This project focused on the 'Hook-Hold-Action' framework for short-form video. I took a 4 minute complex technical process and condensed it into a 14-second high-impact tutorial. Using CapCut’s advanced features including motion tracking, color grading for night-mode aesthetics, and rhythmic cutting.
I created a piece of content optimized for both TikTok engagement and technical clarity. The result is a 'final result transition' that requires no translation to be understood globally.
Model: :Trash Box 🚮
1
54
A self-initiated project to demonstrate the intersection of intricate manual dexterity and high-quality data annotation. This video serves as a benchmark dataset for training vision-based AI models in complex manipulation.
1
64