uforz/pai-tong-kuan

Pai Tong Kuan is a browser extension for Chrome and Microsoft Edge that helps you recreate images you find online. When you see a picture on a webpage that you like, you can use this extension to figure out what text description, or "prompt," was likely used to generate it. Then, it can produce a similar image for you. The idea is simple: you see something you like, and the tool helps you make your own version of it. The extension lets you hover over any web image and click a button to reverse-engineer its prompt. You can also drag a box around part of a webpage to capture a specific area as your source image. It supports multiple languages for its interface, including Chinese, English, Japanese, and Korean. Beyond just copying an image, you can do group generations or swap out characters and objects within a scene. All your results are saved in a local gallery where you can search, view full-size images, and compare them side by side with the originals. To use it, you need to provide your own API keys from supported AI platforms. The extension works with services like OpenAI, ModelScope, SiliconFlow, and several others listed in the settings. You can also add compatible services if you prefer. The developer does not run a middleman server. Your API keys, source images, prompts, and generated results all stay stored on your own computer. Data only leaves your machine when you actively start a task, and it goes directly to whichever AI platform you selected. Installing the source version takes a few steps. You download the code, open your browser's extensions page, turn on developer mode, and load the folder containing the manifest file. After that, you configure your API keys and pick your models. The developer recommends using the official store version for regular use, but the source code is available for those who want to run it directly. The project is licensed under AGPL-3.0, which means anyone who modifies and shares it must also share their source code.