Pixie
smart photo App that learns trendy styles from influencers and stylizes pictures
Overview
With the rise of social media, visual branding becomes increasingly important. Individuals need new styles and aesthetic photos to demonstrate their personalities. Small business owners and individual sellers need consistently stylized photos to establish brand image.
Responding to the ongoing need for visual branding and individual expression, Pixie is a smart photo App that uses machine learning to learn trendy styles from artists and influencers’ photos. The styles are then defined and generated as assets for Pixie’s users to purchase and use, in order to edit their own photos. In other words, Pixie is trying to copyright styles.
Interactive Demo
Time
Apr 2018
Project Type
USC Viterbi ABC Innovation Prize Top 3 Finalist // Mobile UI/UX
Tools
Sketch, InVision, Adobe Suite (Ai, Ps)
Team
2 engineers, 1 designer, 1 business person
My role: design lead, research, UI/UX and branding
Problem
With the rise of social media, visual branding becomes increasingly important.
Individuals need new styles and aesthetic photos to demonstrate personalities
Small business owners and individual sellers need consistently stylized photos to establish brand image, and boost visitors or sales
How might we facilitate consistent and compelling visual branding while accurately capturing ongoing trends and styles embedded in popular consumer culture?
Research & Ideation
Survey (Quantitative)
Interview (Qualitative)
“Consistently good-looking photos are a large part of my online representation.“
— Amanda W.
“I usually took over 50 photos for something, and it’s very hard for me to organize and stylize them because of the large scale.“
— Chloe X.
“I want to follow the style trends but existing photo filters are too generic. It usually takes me over 20 minutes to edit a photo and achieve a style effect that I like.“
— Joanna, D.
Current Solution Research
Current photo editing tools:
Users tend to get lost in thousands of filters and attribute settings - hard to achieve the desired outcome;
Generic - unable to capture popular styles and ongoing consumer trends
Hire a Professional photographer:
Costly on time and money
Limitation and inflexibility on sites, equipment and etc.; hard to communicate
Competitive Analysis
Mood Board
Low Fidelity Prototype & Wireframe
Technology Details
Design Outcome
Onboarding
Stylize a Photo
Users can choose any photo in their camera roll. Then Pixie will use image recognition to analyze the categorization of this photo (food, portrait, landscape, etc.), and then suggest styles based on similar structure and color distribution search, as well as ongoing popularity from Instagram influencers. The selected photo will then be stylized through the use of style transfer.
Discover & Me
Discover: Users are also able to browse trendy styles and purchase them through the app’s Discover Page. The Pixie app has a built-in E-commerce system in a UGC (User-Generated Content) format. Monetization provides artists incentives to create styles and make them consistent.
Me: Users can access their personal gallery with recently styled photos, purchase history, and style creation.
