2025
Makiverse
Makiverse is a platform that lets users generate, customize, and train AI manga characters using reference images, prompts, and personalization tiers. I led the end-to-end design for its core character generation feature, from flow architecture to visual components and onboarding.
AI
Entertainment
Project Overview
Designing the future of character creation: modular, customizable, and AI-powered.®
The Problem
Generating AI characters that feel consistent and personalized requires multiple technical steps: reference image processing, prompt crafting, and LoRA training. All of which are complex for everyday users. We needed to build a seamless, beginner-friendly UX that didn’t sacrifice control for power users.
Design Goals
Design Goals
Make advanced AI generation workflows approachable without dumbing them down.
Clearly differentiate between “generation” and “training”, a frequent point of user confusion.
Create a visually engaging and mobile-first interface that reflects anime/manga aesthetics.
Build a foundation for inventory tagging, tiered upgrades, and visual identity over time.
Key UX Features
Key UX Features
Character Generation Flow
Designed a multi-step UX: description → reference image upload → quality tier → result.
Introduced guidance text and tooltips to help users distinguish between base prompt generation vs. LoRA training for personalization.
Created fallback states for empty prompts, missing uploads, and unclear generations.
Training Tier Selection
Users could choose different generation “tiers” each with cost, time, and quality implications.
UX used pricing blurbs and visual hierarchy to set expectations around output differences.
Designed toggleable previews for comparing quality levels before committing to training.
Inventory & Character Sheet
Designed a collectible-style “Character Sheet” UI where users could view their created characters, tag them, and prep them for training.
Blurred out “coming soon” states for gear/inventory panels to tease future functionality.
Designed modular, scalable card components with rarity and status badges.
Branding & Visual Design
Leveraged manga-style layout principles: strong paneling, minimal UI chrome, and contrast-heavy visuals.
Used light parchment textures, anime-inspired fonts, and layered gradients to evoke a digital manga workshop vibe.
Prioritized UI clarity over maximalism to ensure wide accessibility.
Results
Results
Successfully launched beta generation tool for early adopters
Reduced user confusion around training vs. generating by introducing tier labels and tooltips
Helped shape the foundation for the upcoming LoRA training system, linked directly to character history
Makiverse taught me how to design for nonlinear, multi-step AI workflows while maintaining clarity and user control. I had to deeply consider user mental models, onboarding drop-off points, and how to visualize otherwise invisible technical processes.
The experience also helped me refine my ability to balance aesthetic world-building with pragmatic UI clarity, something I’ve carried into every project since.
More Works
(truo® — 02)
©2025
2025
Makiverse
Makiverse is a platform that lets users generate, customize, and train AI manga characters using reference images, prompts, and personalization tiers. I led the end-to-end design for its core character generation feature, from flow architecture to visual components and onboarding.
AI
Entertainment
Project Overview
Designing the future of character creation: modular, customizable, and AI-powered.®
The Problem
Generating AI characters that feel consistent and personalized requires multiple technical steps: reference image processing, prompt crafting, and LoRA training. All of which are complex for everyday users. We needed to build a seamless, beginner-friendly UX that didn’t sacrifice control for power users.
Design Goals
Make advanced AI generation workflows approachable without dumbing them down.
Clearly differentiate between “generation” and “training”, a frequent point of user confusion.
Create a visually engaging and mobile-first interface that reflects anime/manga aesthetics.
Build a foundation for inventory tagging, tiered upgrades, and visual identity over time.
Key UX Features
Character Generation Flow
Designed a multi-step UX: description → reference image upload → quality tier → result.
Introduced guidance text and tooltips to help users distinguish between base prompt generation vs. LoRA training for personalization.
Created fallback states for empty prompts, missing uploads, and unclear generations.
Training Tier Selection
Users could choose different generation “tiers” each with cost, time, and quality implications.
UX used pricing blurbs and visual hierarchy to set expectations around output differences.
Designed toggleable previews for comparing quality levels before committing to training.
Inventory & Character Sheet
Designed a collectible-style “Character Sheet” UI where users could view their created characters, tag them, and prep them for training.
Blurred out “coming soon” states for gear/inventory panels to tease future functionality.
Designed modular, scalable card components with rarity and status badges.
Branding & Visual Design
Leveraged manga-style layout principles: strong paneling, minimal UI chrome, and contrast-heavy visuals.
Used light parchment textures, anime-inspired fonts, and layered gradients to evoke a digital manga workshop vibe.
Prioritized UI clarity over maximalism to ensure wide accessibility.
Results
Successfully launched beta generation tool for early adopters
Reduced user confusion around training vs. generating by introducing tier labels and tooltips
Helped shape the foundation for the upcoming LoRA training system, linked directly to character history
Makiverse taught me how to design for nonlinear, multi-step AI workflows while maintaining clarity and user control. I had to deeply consider user mental models, onboarding drop-off points, and how to visualize otherwise invisible technical processes.
The experience also helped me refine my ability to balance aesthetic world-building with pragmatic UI clarity, something I’ve carried into every project since.
More Works
(truo® — 02)
©2025
2025
Makiverse
Makiverse is a platform that lets users generate, customize, and train AI manga characters using reference images, prompts, and personalization tiers. I led the end-to-end design for its core character generation feature, from flow architecture to visual components and onboarding.
AI
Entertainment
Project Overview
Designing the future of character creation: modular, customizable, and AI-powered.®
The Problem
Generating AI characters that feel consistent and personalized requires multiple technical steps: reference image processing, prompt crafting, and LoRA training. All of which are complex for everyday users. We needed to build a seamless, beginner-friendly UX that didn’t sacrifice control for power users.
Design Goals
Make advanced AI generation workflows approachable without dumbing them down.
Clearly differentiate between “generation” and “training”, a frequent point of user confusion.
Create a visually engaging and mobile-first interface that reflects anime/manga aesthetics.
Build a foundation for inventory tagging, tiered upgrades, and visual identity over time.
Key UX Features
Character Generation Flow
Designed a multi-step UX: description → reference image upload → quality tier → result.
Introduced guidance text and tooltips to help users distinguish between base prompt generation vs. LoRA training for personalization.
Created fallback states for empty prompts, missing uploads, and unclear generations.
Training Tier Selection
Users could choose different generation “tiers” each with cost, time, and quality implications.
UX used pricing blurbs and visual hierarchy to set expectations around output differences.
Designed toggleable previews for comparing quality levels before committing to training.
Inventory & Character Sheet
Designed a collectible-style “Character Sheet” UI where users could view their created characters, tag them, and prep them for training.
Blurred out “coming soon” states for gear/inventory panels to tease future functionality.
Designed modular, scalable card components with rarity and status badges.
Branding & Visual Design
Leveraged manga-style layout principles: strong paneling, minimal UI chrome, and contrast-heavy visuals.
Used light parchment textures, anime-inspired fonts, and layered gradients to evoke a digital manga workshop vibe.
Prioritized UI clarity over maximalism to ensure wide accessibility.
Results
Successfully launched beta generation tool for early adopters
Reduced user confusion around training vs. generating by introducing tier labels and tooltips
Helped shape the foundation for the upcoming LoRA training system, linked directly to character history
Makiverse taught me how to design for nonlinear, multi-step AI workflows while maintaining clarity and user control. I had to deeply consider user mental models, onboarding drop-off points, and how to visualize otherwise invisible technical processes.
The experience also helped me refine my ability to balance aesthetic world-building with pragmatic UI clarity, something I’ve carried into every project since.
More Works
©2025