CHROMA: Your Colours. Your Signature Look.

Inspiration

A makeup shade can look perfect on a product page and completely different when it reaches someone's face.

That gap is one of the frustrating parts of buying beauty products online. Product images and generic swatches cannot always communicate how a colour will complement an individual's complexion. People are often left guessing, experimenting, or buying products that don't match the look they imagined.

We wanted to make that experience more personal.

CHROMA was inspired by a simple idea: beauty technology should help people discover what works with their individual colouring, rather than asking everyone to fit the same beauty standards.

Instead of treating a selfie as just an image, we built CHROMA around the idea of turning it into the starting point for a personalized beauty experience.

What It Does

CHROMA brings AI-powered skin-tone analysis and personalized beauty discovery into one experience.

Users provide a selfie, and the application uses Perfect Corp's YouCam API to obtain skin-tone analysis. CHROMA then interprets the analysis within its own application logic to help users understand their colouring and explore beauty possibilities.

The experience is designed to connect three stages:

  1. Understand: Analyze the user's skin tone using YouCam technology.
  2. Interpret: Turn the analysis into a more understandable, personalized colour experience.
  3. Explore: Connect that understanding to virtual makeup visualization, helping users preview potential beauty looks.

The goal is not to tell people what they should look like. It is to give them better information and more room to experiment.

How We Built It

We built CHROMA as a web application using Next.js, React, and TypeScript, with Perfect Corp's YouCam API providing the skin-analysis capability.

The application follows a structured workflow:

  • The user uploads a selfie through the CHROMA interface.
  • The backend handles the analysis request and communicates with the YouCam API.
  • The returned analysis is processed into a structured result.
  • CHROMA presents the result through its own personalized interpretation experience.
  • The architecture includes signed provenance tickets to associate downstream rendering with the analysis that initiated it.

We also treated security and reliability as core product requirements, not features to add later.

We implemented server-side API handling, request-origin protections, Redis-backed rate limiting and spend controls, and HMAC-SHA256 signed tickets. These safeguards help protect the application from unauthorized requests, excessive API usage, and forged or altered analysis-to-render relationships.

The application is deployed on Vercel, with Upstash Redis supporting its production controls.

Challenges We Faced

One of our biggest challenges was making the experience more than a simple API demonstration.

It would have been easy to build a page that accepts a selfie and displays an API response. But that would not solve the actual user problem. We needed to make the analysis meaningful within a complete beauty-discovery experience.

We also faced several technical challenges:

  1. Connecting the analysis to the experience

Raw analysis data is not automatically useful to everyday consumers. We had to design an application layer that could make the results easier to understand and connect them to the wider beauty experience.

  1. Protecting paid API usage

AI-powered services can introduce unpredictable costs if endpoints are exposed without proper controls. We implemented rate limiting, budget enforcement, and server-side protections to reduce the risk of unauthorized or excessive requests.

  1. Preserving trustworthy results

A result should not be treated as trustworthy simply because it appears on screen. We introduced signed tickets to establish a verifiable relationship between an analysis and its associated rendering workflow.

  1. Building a polished product under time constraints

We wanted CHROMA to feel like a coherent beauty product rather than a generic technical dashboard. That meant refining the interface, simplifying the first-time experience, and keeping technical explanations out of the way of the primary user journey.

What We Learned

Building CHROMA taught us that successful AI products need more than a capable model or API.

The quality of the experience depends on how well the technology is translated into something people can understand and use. We learned to think about AI outputs as part of a product journey, not as the final product itself.

We also learned that trust has to be designed into the architecture. Protecting user requests, controlling API spending, and preserving the relationship between analysis and subsequent results are just as important as the visible interface.

Most importantly, we learned that personalization is not simply about collecting more data. It is about using relevant information to make an experience feel more useful, understandable, and individual.

What's Next

CHROMA is designed to grow beyond skin-tone analysis into a more complete personalized beauty-discovery experience.

We see opportunities to expand colour recommendations, improve makeup visualization, and eventually help beauty brands and retailers make online product discovery more relevant to individual customers.

The bigger idea is simple: online beauty shopping should not be a guessing game. Your face is individual. Your beauty experience should be, too.

Built With

  • analysis
  • api
  • artificial
  • beauty
  • color
  • computer
  • e-commerce
  • intelligence
  • next.js
  • react
  • skin
  • technology
  • theory
  • tone
  • try-on
  • typescript
  • virtual
  • vision
  • web
  • youcam
Share this project:

Updates

Submission history