Do NOT download the file on the right, it's just a placeholder.

At first for Ideogram4, but now also support Moss, Flux3 Klein and Krea3 coming soon.

CHANGELOG

v1.1 Also support converting ideogram4 structured json prompt to natural language prompt, so it can be used on Krea2 Z-IMAGE etc. In this way, you can also make the composition more stable (tested on Krea2).

An enhanced Ideogram4 prompt tool, still a very primitive tool prototype. New features will be gradually improved and added during use.

Online demo

github: https://github.com/SurpassHR/ideogram4-editor

You can also deploy your own online canvas and customize your AI provider.

Overview

Ideogram4 Editor is a web-based visual tool for crafting structured JSON prompts for the [Ideogram 4](https://ideogram.ai/) image generation model. Instead of writing complex JSON by hand, you interact with a canvas to define image regions, configure visual properties, and generate standardized prompts — all with real-time visual feedback.

Core workflow: Drag to create bounding boxes on a canvas → Configure each box (mode, text, description, colors) → Set global parameters (aesthetics, lighting, medium) → Generate normalized JSON prompt → Render via ComfyUI API.

Features

- 🖊️ Canvas Interaction — Drag to create, move, and resize bounding boxes with pixel-precise positioning

- 🔍 Artboard Controls — Scroll-wheel zoom (centered on cursor), middle-button pan, 10%–500% zoom range

- 🎨 Color Palettes — Global palette (up to 16 colors) + per-box independent palette

- 📝 Dual Mode — Photo mode MODE_PHOTO) and Art Style mode MODE_ARTSTYLE); each box can be obj or text

- 🌐 i18n — Built-in Chinese/English interface with persistent language preference

- 📋 JSON Generation / Loading — One-click JSON export, paste to import existing prompts

- 🖼️ Image Drag & Drop — Import PNG backgrounds; auto-extract embedded ComfyUI metadata

- 🔌 ComfyUI Integration — Direct API calls for image generation with seed control and quality presets

- 🤖 LLM Assistance — Multi-provider support (OpenAI, Anthropic, Gemini, OpenAI-compatible) for AI-powered prompt optimization

- 💬 AI Chat Panels — Per-box AI chat (✨ button) and canvas-level composition chat with streaming SSE output

- 🧠 Chain of Thought — Auto-detects and displays model reasoning in collapsible blocks

- 🖼️ Canvas Snapshots — Each AI reply includes a canvas thumbnail from the moment of sending

- 🎭 Dark Theme — Custom CSS-driven design system with zero external UI libraries

- ⌨️ Keyboard Shortcuts — Ctrl+D/X/C/V for box operations, Delete to remove

- ⚡ SSE Streaming — Token-by-token streaming for all LLM interactions

- ✨ Interactive Glow Grid — Dynamic dot-matrix glow background via CSS Masking + JS coordinate mapping