AI Pipelines & Production Experience

Project Name

Year

2022.12 - 2026

Overview

Beyond my core expertise in concept art and art direction using DCC tools, I have strategically integrated tailored AI workflows across 20+ projects.

Applications span concept design, virtual-production backgrounds, interactive applications, film and episodic VFX, fully AI-generated animation, concert visuals, and graphic design

Here is an overview of my AI project experience and R&D work.

My Role

I utilize AI across my projects to generate everything from early pitches and assets to full AI animations and VFX.

Alongside production, I actively assist the team in adopting AI workflows by testing new tools, models, and pipelines.

I began exploring AI in 2022, right as it emerged in the visual industry.

Earl

Testing and Generating Assets

Early style-transfer test

Training SD/SDXL LoRAs
Generating pitch images and image assets
Frame-by-frame AI animation Testing

MidjourneyStable DiffusionControlNetEbSynthLoRAAdobe

From Concepts to Video

Concept work became easier and steadier
Open and closed source video models — not reliable yet, but usable on some projects
Trained LoRAs for open models to generate more consistent output.
Use open-source models to test more diverse and detailed applications, such as relighting, altering pose proportions, outpaint / inpaint and camera movement control.

Nano BananaRunwayKlingVeo3GrokComfyQwenFluxZ-ImageWanLTXTripoMeshyInfiniteTalkAdobeSunoElevenLabs

In Every Stage

Image and video generation stable enough to use at scale, controlled mostly by prompt and reference
Open-source models still valuable when testing more detailed controls or addressing specific project requirements.
Claude and Codex, plus MCP into each application — extending my range and building tools that smooth the pipeline

GPT ImageSeedanceComfyQwenFluxKreaMinimaxLTXTripoTopazClaudeCodexvarious MCP toolsAdobeSunoElevenLabs

AI For Concept & Background (2023 - 2025)

Concept art made in 2024 for the series Black Tide Island (2025)
I used Stable Diffusion and Photobashing to replace the land in live-action footage with the ocean, providing a reference for the VFX team.

ROG ProArt Commercial (2024)

MSI Cnd Series (2024)

ROG CES Product Lauch Event (2025)

PhotoshopMidjourneyNano BananaStable DiffusionComfy

Fully AI Animation - Have A Bad Luck (2026.2)

• We completed a 15-minute, fully AI-generated animation in just 3 weeks with a 3-person team.

• I led the AI Art Direction for this project, also producing over 120 images and dozens of AI-generated video clips.

• The project was originally designed for a 3D animation pipeline, so the storyboards and narrative content presented many challenges for AI generation.

Nano BananaGrokKlingComfyQwenWanPhotoshop

Character concept art by the concept team, and the AI-generated 3D-style character made by me

Fully AI Animation - Candy Heterotopia (2026.8)

Candy Heterotopia Lost on the Factory LineAI · Experience · Concept/Visdev

• I independently developed the script and art direction, and generated all the styleframes in 2024.

• In 2026, our two-person team delivered a 4½-minute, one-continuous-take, 60 fps 4K animation in 3 weeks, covering everything from 3D previs to final AI generation.

BlenderBlender MCPClaudeComfySeedanceGPT ImagePhotoshopQwenTripoNano BananaMinimaxFluxKrea

AI in Every Stage (2026-)

Hualien AI Music Lab ExhibitionAI · Experience

• Used AI to generate visual assets, utilized TouchDesigner for creating interactive content, and used Claude to assist the production workflow.a

• As the Art Director for this project, I planned the AI strategy and produced the exhibition's animations and interactive content.

TouchDesignerClaudeSeedanceMinimaxGPT Imagevarious MCP tools

Lu Yun Zepp Taipei Live ShowEvent · 3D · AI

• I independently directed and produced visual animations for four concert songs.

• Developed a real-time interactive live game using Claude.

• Created visuals with Blender and TouchDesigner, utilizing Claude to accelerate production.

BlenderTouchDesignervarious MCP toolsClaudeGPT ImageSeedance

AI Avatar Guide WebAppAI · Experience

• Built by our team for the studio showcase, as part of the event experience.

• Users can get their own stylised avatar and interact with them on the webapp.

• I led the art direction, UI design, and character design, while planning the AI strategy, assisting in building the AI workflow, and co-training LoRA models with a team member.

FigmaQwenWanLoRANano BananaComfy