Right-click any image on the web and PromptLens reverse-engineers a generation prompt — written in the native dialect of Midjourney, Seedance, Kling, Flux, SDXL and more.
BYOK · your Gemini, Claude, OpenAI or OpenRouter key · stored locally · sent only to the provider you choose
How it works
Choose “Generate prompt from image” from the context menu — or drop, paste, or pick a file in the toolbar popup for protected images.
Each target model gets a prompt in its own grammar — parameters for Midjourney, shot structure for Seedance, tags and negatives for SDXL. Switching is cached and instant.
One button. The prompt carries the source's measured aspect ratio and a full-detail mandate — specific colors, light, composition, no filler adjectives.
Seven dialects
Single-line v7 prompt with exact --ar from the source image, --stylize tuned to photographic or illustrative looks.
Treats the image as a first frame: Shot / Scene / Motion / Style structure for a 5–10s clip.
One flowing cinematography paragraph — camera movement, action, light — that extends the still into motion.
Dense natural prose the way diffusion-transformer models like it: concrete, visual, no vague adjectives.
Comma-separated tags ordered by weight, plus a matched negative prompt.
An art-director's cover brief: concept, palette, typography mood, one signature element.
A maximally faithful, model-agnostic paragraph you can adapt anywhere.
Bring your own key
PromptLens has no backend, no accounts, no analytics, and collects nothing. Your API key
is stored in chrome.storage.local on your machine and used for exactly one
thing: calling the vision API you selected — Google Gemini, Anthropic Claude, OpenAI, or
any model on OpenRouter. Images are converted to base64 in your browser and sent directly
to that provider. Read the full privacy policy.