Try any garment on any photo. Reference-accurate AI fashion try-on, in seconds.
Open-source virtual try-on + multi-image fashion editor. Drop a person + a flat-lay product shot, get a styled photo back. BYOK, local-first.
Quick Start · Modes · How it works · BYOK · Stack
Drape is a free, open-source AI fashion try-on studio you can self-host in two minutes. Drop a full-body photo, drop a garment reference (flat-lay product shot works best), and get back a photoreal image of the person wearing it — face and body preserved.
Two modes, one workflow:
- Try-on (strict) — IDM-VTON swaps one garment onto the person without touching face, hair, or body. Best for accurate product visualization.
- Fashion Edit (flexible) — nano-banana takes the person + up to 4 garment refs and a prompt. Stack a top, bottom, and accessory; change the scene; restyle the pose.
- 🛍️ E-commerce mockups without a model shoot
- 👗 Outfit pre-shopping — see how it actually looks before buying
- 🎬 Lookbook & campaign drafts for indie fashion brands
- 📸 Reference-accurate styling — exact color, pattern, silhouette preserved
- 🧪 A/B style tests — same person, swap garments with a fixed seed
git clone https://github.com/morsechimwai/drape.git
cd drape
pnpm install
pnpm devOpen http://localhost:3000, open the Settings dialog, paste your Replicate API key, and you're generating in under 60 seconds.
No
.envfiles. No deploy step. Your API key lives in your browser.
Twelve presets out of the box. Each one is a hand-tuned prompt — not a generic LoRA.
| Style | Best for | |
|---|---|---|
| 💼 | Corporate | LinkedIn, About pages, conference bios |
| 🌸 | Anime | Twitter, Bluesky, anime communities |
| 🍃 | Ghibli | Soft, warm, hand-painted vibes |
| 🧸 | 3D Pixar | Family-friendly avatars, kid profiles |
| 🤖 | Cyberpunk | Discord, gaming, devs who self-identify as "based" |
| 🎬 | Cinematic | Personal websites, portfolios, newsletters |
| 🎨 | Oil Painting | Profile pics that look like museum pieces |
| 📷 | B&W Film | Timeless, editorial, journalist energy |
| 🖌️ | Watercolor | Soft, artistic, calm |
| ⚔️ | Fantasy | D&D, RPG, fantasy authors |
| 🌴 | Vaporwave | Synthwave, retro, 80s nostalgia |
| ⚪ | Minimal | Clean editorial design, indie hackers |
Plus modifier chips to mix in lighting, mood, and background tweaks. And advanced controls if you actually know what guidance_scale does.
- Upload a clear, front-facing photo (the better the input, the better the output — physics still applies)
- Pick a style preset or write your own prompt
- Tweak with modifier chips (lighting, background, mood)
- Generate via Replicate using your own API key
- Download the image — it's yours
No accounts. No subscriptions. No image stored on our servers, because there are no servers — it's a Next.js app you run yourself.
Drape uses your Replicate account. That means:
- ✅ You pay only for what you generate (usually pennies per image)
- ✅ Your images stay between you and Replicate
- ✅ No middleman markup
- ✅ No "freemium" rug pulls
Add your key once in the in-app Settings dialog. It's stored locally in your browser.
Built on the modern Next.js stack — fast, type-safe, and easy to fork.
- Next.js 16 (App Router) + React 19
- Tailwind CSS v4 + shadcn-style components
- Replicate for AI image generation
- Zustand for state, Zod for validation
- Motion for the smooth bits
- TypeScript, ESLint, deploy-anywhere
One-click deploy to Vercel — the platform behind Next.js itself.
vercel deployOr self-host anywhere that runs Node.
PRs welcome — especially new style presets. If you've tuned a prompt that produces consistently great portraits, open a PR adding it to lib/presets.ts.
MIT — free to use, fork, remix, and ship as your own product. No attribution required (but a ⭐ is appreciated).
If Drape saved you a styling session, star the repo. ⭐
Keywords: AI profile photo generator · AI headshot generator · LinkedIn headshot AI · anime profile picture · Ghibli portrait AI · Pixar avatar generator · cyberpunk PFP · open source AI portrait · Next.js Replicate AI app · self-hosted AI headshot