A 2022 project to build the Lookbook feature for the INDEPS NFT project. Existing video content limited how actively users could participate and explore, so a more immersive, interactive 3D environment was needed.
The goal was to rebuild scenes originally produced as video into interactive 3D scenes, delivering a rich experience where users could actively explore and interact.
Built the 3D environment on Vite and React-three/fiber, analyzing the video content and reconstructing it as 3D scenes. Developed a control system and navigation features for user interaction, and secured optimized 3D rendering performance for the web.
Delivered an interactive 3D Lookbook that fully replaced the video. As immersive content users could explore directly, it improved user engagement and brand experience compared to the original static video.
A 2022 project to build the PhotoBooth feature for the INDEPS NFT project. Users needed a playground where they could select an avatar, dress it, and apply various moods and poses to create a personalized scene. With over 200 3D meshes rendering simultaneously, performance optimization was critical.
The goal was to provide a creative tool where users could freely style scenes with their own avatar, while securing optimized 3D rendering performance that would run smoothly in the browser.
Built the 3D environment on Vite and React-three/fiber, and used InstancedMesh optimization to cut draw calls from 173 to 50 while rendering 200 meshes simultaneously. Applied Draco/KTX compression to the GLTF models, reducing model size from 32MB to 6MB and GPU memory from 675MB to 84MB, so it could run on mobile Safari as well.
Successfully built a PhotoBooth playground where users could freely create. Through draw-call optimization and model compression, achieved stable 3D rendering performance in the browser, delivering an immersive experience that extended to mobile as well.