Vision2Prompt
TΔNΘ
Web app for editing prompts or creating them from scratch. Under the hood — a prompt-forming instruction refined over 2 years: from SD to Nano Banana Pro. Optimized for Flux 1, Flux 2 [Max], and Nano Banana Pro. Allows converting prompts from standard Flux to Flux 2 or NBP.
Vision (I2P)
Upload any image from the web and extract a prompt from it — to generate the same image or understand prompt logic (e.g. for training your own LLM). Modes: Visual — soft image, ideal for landscapes; Cinematic — dirty lens, effects, pro camera work; Agent — most precise, outputs clear prompts for NBP.
Text
Create prompts from scratch or convert old queries for new models. Paste an old prompt and the system rewrites it. Or type an idea (e.g. "a cat making pancakes") — the service builds a full prompt. Modes: Visual — soft, may add overexposure; Cinematic — cinematic prompt; Agent — most precise, Natural Language, fits future generators.
Runs on Gemini 3 Flash. VPN and Google account may be required.