UNI: AN AI IMAGE GENERATOR


After analyzing current trends in AI image generation, I developed Uni, a web-based image generator prototype. Uni aims to strike a balance between accessibility and power, offering users a versatile tool for creating or editing images.

UI/UX Design
March 2024
Tools : Figma
Roles:  Researcher, UI/UX Designer, UX Workshop Facilitator, Usability Tester




The Case Study


The Problem



Existing AI image generators exhibit a clear issue: they are either highly potent yet challenging to navigate, or overly simplistic with limited functionality. This gap in the market highlights the need for a user-centered image generator that prioritizes accessibility while offering a wide range of customization features to cater to diverse user needs.

Research



Research Collaborators: 
Dunham DeRemer, Kai Shafer, Ilona Leclerc Devillard, and Jeff Murray
During my research phase, I learned that approximately 34 million images were generated daily using AI. While Adobe Firefly showed rapid growth, Midjourney had the largest amount of users. After conducting audits of Ideogram, Firefly, Krea, and Midjourney, I met with peers developing their own image generators in order to share our research and discuss the pros and cons of each product. Additionally, I found a Yale study indicating that individuals could only identify AI-generated images with 54% accuracy. Although it is important to note it will likely be more difficult to tell when images are AI generated as the technology develops. After this initial meeting, I formulated my problem to solve and solution. I also began to brainstorm what features could potentially be included in the prototype. Based on this initial progress, I crafted three personas to further guide me. But as the project developed, I refined them to represent distinct user proficiency levels: beginner, intermediate, and advanced.


To accommodate users with varying levels of proficiency, I conducted a thorough assessment of pain points across different image generators. Simplifying the onboarding process to rival that of Midjourney's interface was crucial, while ensuring customizable options were streamlined yet comprehensive.

Process Overview


After conducting thorough research, I allocated one week for branding, as the primary focus was on developing a high-fidelity prototype. Because I was uncertain about the brand identity, I prioritized crafting a slogan, ultimately settling on 'Your Ideas, Expanded by AI' after several iterations. This guiding phrase informed subsequent decisions regarding the name, wordmark, typography, and color palette with clarity and confidence.

My remaining time was dedicated to developing the prototype in Figma. I began by compiling a list of potential features and sketching basic wireframes, but struggled to prioritize features and structure the website effectively. To address this, I conducted a UX Workshop, collaborating with classmates to define essential features for the prototype. Through this collaborative process, key features such as aspect ratios, upscaling, preset styles, background removal, image variations, and an infinite canvas akin to Figma were identified.

Using this feedback, I crafted user flows that integrated these essential features. Subsequent weeks were dedicated to iterative prototyping, incorporating user feedback to refine the design. In March, I presented the final result. As the spring quarter commenced in April, I remained committed to refining the project and honing my proficiency in Figma.

The UX Workshop



The Solution


My proposed solution is to develop a user-centered web-based image generator named 'Uni' Uni prioritizes accessibility by offering an intuitive interface and user-friendly navigation. Furthermore, Uni provides a diverse array of customization options, allowing users to tailor their images to their specific preferences and needs. By finding a balance between accessibility and customization, Uni aims to empower users to effortlessly create high-quality images.