Shan Jin
Creative Technologist
AI Video for Empower
AI video production for wealth management platform Empower's national campaign: “Get good at money. So you can be a little bad.” Leveraged generative AI tools to produce a surreal, comedic social spot featuring a pampered dog enjoying a high-end spa day, visualizing the carefree lifestyle that comes with smart financial planning.
Client: Empower
Made with: Google Veo, External Voice Over
Year: 2026
AI Video, GSP Art Show
Created an experimental, deeply personal AI video for an internal agency art exhibition responding to the theme “ID Check: What makes you, you?”
By compiling and processing personal archival photographs from childhood to adulthood through generative video tools, the piece creates a fluid, morphing visual timeline that reflects on memory, growth, and what truly shapes an individual.
Made with: Hailuo, Kling
Year: 2025
AI Video, exploration
An experimental motion study focused on the intersections of real-world video capture and open-source generative AI tools.
Using a real-world video clip of myself as the initial conditioning seed, I mapped a fluid animation pipeline that gradually morphs my physical form into entirely new objects and environments.
Made with: Deforum, Automatic 1111
Year: 2023
Lunchabuild LoRA
Developed a custom-trained LoRA model to scale the “Lunchabuild This” campaign into an infinite generator. Utilizing a proprietary dataset curated by professional food stylists, the model learns the exact textures and structural logic of Lunchables components, empowering users to turn any prompt into a high-fidelity, brand-accurate food sculpture instantly.
Client: Lunchables
Made with: LoRA training
Year: 2023
AI Likeness Exploration
A technical R&D project mapping the evolution of identity-preserving AI pipelines for personalized UGC campaigns. Moving from early face-cropping and DALL-E outpainting workflows to custom Stable Diffusion embeddings, hypernetworks, and advanced ComfyUI IP-Adapter pipelines, this study tackles the complex challenge of maintaining precise facial likeness within infinitely variable generative environments.
Year: 2023 - 2024
Augmented Reality
An R&D compilation exploring spatial tactility using ARKit's people occlusion and body tracking features. By aligning human body depth maps with virtual objects to calculate real-time intersection data, these prototypes unlock a controller-free "touch" mechanic—allowing virtual elements to react dynamically to physical human contact.
Year: 2019 - 2020
Generative Visuals
A multidisciplinary compilation of real-time visual experiments and generative art. Leveraging a diverse toolkit including Three.js, WebGL, TouchDesigner, openFrameworks, and custom GLSL shaders, this collection bridges the gap between raw math and immersive art, exploring responsive physics, procedural geometry, and interactive digital spaces.
Year: 2016-2018
Drawings
A lifelong archive of non-technical visual explorations. Moving chronologically from early graphite sketches and watercolors to digital painting in Procreate, this collection culminates in my current study of Yancai (岩彩) and Guofeng (国风). By working with heavily textured mineral pigments and traditional Chinese art styles, these pieces explore the intersection of ancient material culture and modern visual composition.