Hover
Move across any image and ImgX surfaces a single, calm prompt button.
Chrome extension for AI creators
ImgX
Hover any image. ImgX reads its visual grammar and turns it into bilingual prompts for Midjourney, Stable Diffusion, DALL·E, and any modern image model.
8
specialized prompt templates
12
structured visual fields
0
intermediary servers
Real extension UI
Analysis stays where the image lives.
ImgX injects a lightweight floating interface into the page: a hover bubble, a live queue, and a result card with Chinese, English, and JSON views.
Three-second flow
From reference image to usable prompt.
Analyze
The classifier picks the right visual template before your VL model reads the image.
Reuse
Copy bilingual prompts, inspect JSON, export a share card, or recall history without another API call.
Built for prompt quality
Not a caption generator. A visual grammar extractor.
ImgX asks the model for composition, lighting, spatial layers, material cues, color structure, style tags, and negative prompts. It also guards against invented cameras, artists, brands, and IP names.
Bilingual by default
Chinese natural paragraph plus English prompt paragraph, tuned for creator workflows.
Batch queue
Click multiple images and let the background engine process them with retries and rate-limit awareness.
Share-ready output
Export a polished 1080px prompt card directly from the result panel.
History recall
Analyzed images can restore previous results instantly with zero extra model cost.
Auto-routed intelligence
Eight prompt templates, one invisible classifier.
A product hero shot should not be decoded like an anime illustration or a mobile UI. ImgX routes each image to the right expert prompt before generation.
Fallback coverage for all visual dimensions.
Lighting type, expression, gaze, focal feel.
Time of day and foreground, midground, background.
Materials, interior space, perspective discipline.
Commercial lighting, background, shadow treatment.
Platform, grid, design language, component list.
Style taxonomy, palette, line handling.
Type hierarchy, grid, contrast, whitespace.
Privacy by design
Your key, your endpoint, your browser.
ImgX has no intermediary server. Your API key stays in local browser storage and image data is sent only to the OpenAI-compatible endpoint you configure.
Local keyStored in Chrome, not on an ImgX backend.
No telemetryNo tracking pipeline or usage analytics.
Compatible APIsWorks with any OpenAI-compatible vision endpoint — SiliconFlow, OpenAI, Gemini, DeepSeek and more.
Start in minutes
Install ImgX and bring your own VL model key.
- Load the extension in Chrome.
- Open the three-step onboarding page.
- Set endpoint, API key, and model, then test the connection.