Next.js developer — OpenAI image compositing feature for live app

I have a live AI-powered bedroom styling app called The Bedscape Editor (v0-bedscape-editor.vercel.app) built with Next.js and deployed on Vercel. I need one specific feature completed: image compositing. The user uploads two photos — their bed at home and a product they're considering buying. The app should generate a realistic image showing that product placed on their actual bed. This is the same technology ChatGPT Plus uses when it photoshops a product onto a photo — I need this working cleanly in my app. What's already done: The two-photo upload flow works perfectly at /match OpenAI API key is configured in Vercel environment variables The text verdict (does it match yes/no) works well Basic DALL-E compositing code exists but isn't producing clean results What I need: Debug and fix the existing compositing code OR reimplement using the best available OpenAI image editing endpoint The composited image should display above the text verdict labelled "How it could look" Should fail silently if generation fails