ShowReady

AI stage-look planner for kids' performances. Pick your theme, get a virtual makeup + hair try-on powered by Perfect Corp, and order everything your star needs in one place.

Comment

The Moment That Started It

Anyone who's been backstage at a kids' dance recital knows the chaos. Parents frantically Googling "Swan Lake tutu girls size 6," makeup bags dumped out on folding chairs, someone's kid crying because the wrong color blush was bought. The show is in two weeks. Nobody knows what they're doing.

That's the problem ShowReady solves. Not in a theoretical "wouldn't it be nice" way — in a "here's exactly what to buy, here's what it looks like on your kid, here's the link" way.

What We Built

ShowReady is a 4-step AI stage-look planner for kids in recitals, talent shows, school plays, and cheer competitions.

  1. Pick the show type — ballet, dance, talent, fashion, play, cheer
  2. Choose a theme — 24 curated looks across all categories
  3. Virtual try-on — upload a photo and see the makeup, hair, and costume applied using Perfect Corp's YouCam APIs
  4. Shop the look — every item, curated and linked, with an AI concierge that can answer questions and build your order list on the spot

The AI concierge (powered by GPT-4o-mini) knows the full context of the look — the vibe, the color palette, the specific makeup steps, the hair style, the costume breakdown — and can answer questions like "what's budget-friendly?" or "what brushes do I need?" or just "order everything" and instantly generate a tappable order list.

How We Used Perfect Corp

We integrated three Perfect Corp YouCam APIs in parallel:

  • Look VTO (/s2s/v2.0/task/look-vto) — applies a full makeup look matched to the theme
  • Hair Transfer (/s2s/v2.1/task/hair-transfer) — styles the hair to match (buns for ballet, curls for Latin dance, etc.)
  • Cloth Try-On (/s2s/v3.0/task/cloth) — overlays a themed costume reference image onto the performer's photo

All three run concurrently via Promise.allSettled so the user sees results as fast as the slowest API — typically under 15 seconds for all three. We built a full async task + polling loop, routed user photos through imgbb for public URL hosting (a requirement of the S2S API), and mapped 24 themes to specific Perfect Corp template IDs sourced directly from the API's template library.

The Hard Parts

Getting the cloth API to work properly was the most stubborn challenge. The API needs a clean garment-only reference image — no people, no backgrounds, just the costume. Early attempts with lifestyle photography from Unsplash produced wildly wrong results (a hip-hop look reference image became a crop top on a Swan Lake dancer). The fix was sourcing Pexels images of dresses on hangers and mannequins — one per theme — which gave the model a clean garment signal to work with.

Hair template IDs weren't documented. The Perfect Corp playground shows display names like "Messy Bun" but the API requires internal IDs like female_messy_bun. We fetched the full template list from /s2s/v2.1/task/template/hair-transfer and mapped each theme manually.

The ordering flow taught us something real about the limits of agentic demos: Amazon doesn't expose a "add to cart on behalf of user" API without product ASINs. Rather than pretend otherwise, we built an honest UX — the agent generates a tappable order list with individual Shop buttons, each one tracked with a visual checkmark so the user knows what they've opened. Clean, transparent, actually useful.

What We Learned

The most valuable lesson was about reference data quality. AI try-on APIs are only as good as what you feed them. Spending time curating the right costume reference images — garment-only, good lighting, plain background — moved the quality of results from "confusing" to "genuinely impressive" without changing a single line of API code.

The second was about agentic UX. An AI that knows the full context of what a user is trying to accomplish — the show, the theme, the budget, the specific items — is dramatically more useful than a generic chatbot. Giving the concierge the complete shopping list and look data as system context meant it could answer any parent question without hallucinating products or prices.

Feasibility

ShowReady maps directly onto a real, underserved market. Dance studios alone in the US run tens of thousands of recitals per year. Every one of them has parents who need exactly this. The business model writes itself: affiliate revenue on every Amazon and Etsy purchase, white-label licensing to studios and costume retailers, and a premium tier with custom look design. The Perfect Corp API suite is the technical moat — it's what makes this a try-on experience rather than just a shopping list.

The app is live, fully functional, and deployed. Every feature shown in the demo video works in production, on real photos, with real API calls.

Built With

  • api
  • cloth
  • corp
  • gpt-4o-mini
  • hair
  • look
  • openai
  • perfect
  • react
  • transfer
  • try-on
  • typescript
  • vite
  • vto
  • youcam
Share this project:

Updates