For old version, see v3 branch and https://paint.leafyee.xyz

Painter Leaf

A image creator based on free Cloudflare AI and HuggingFace APIs. Features include:

• Prompt-to-image: supports a variety of models (e.g. Flux.1、StableDiffusion 3.5)
• Image-to-Prompt: convert local images to prompts
• AI Translation: use Chinese prompts in any model
• Store: save your creations to IndexedDB

1 Deployment

1.1 Config Environment Variables

You can use either Fullstack or Client-Server mode.

1.1.1 Fullstack

Set following environment variables in .env file or Vercel.

Key	Value	Required
CF_USER_ID	Cloudflare user id	✅
CF_AI_API_KEY	Cloudflare AI api key	✅
HF_API_KEY	HuggingFace api key

The free plan of Vercel has a limit of 10s for each request, which may cause 504 error (especially when using HuggingFace models). You can subscribe to a Vercel paid plan, run the server locally, or use Client-Server mode.

1.1.2 Client-Server

Deploy the server to Cloudflare Workers (see this project) and set following environment variables in .env file or Vercel.

Key	Value	Required
NEXT_PUBLIC_WORKERS_SERVER	Server url (e.g. https://api.xxx.workers.dev)	✅

Once NEXT_PUBLIC_WORKERS_SERVER is set, all the other environment variables will be ignored.

1.2 Deploy to Vercel

Deploy this project to Vercel (remember to set environment variables).

1.3 Common Issues

• 429 error: You may have exceeded the HuggingFace api request limit. Please wait for a while, reduce the frequency of requests, and consider subscribing to a paid plan.
• 504 error: The request have exceeded the Vercel time limit. See above for resolution.

2 Development

2.1 Clone the repository

git clone https://github.com/LeafYeeXYZ/PainterLeaf.git
cd PainterLeaf

2.2 Install dependencies

bun install # or use other package manager you like

2.3 Local Development

bun run dev

2.4 Build

bun run build