lidge-jun/ima2-gen

Minimal CLI + web UI for OpenAI GPT Image 2 generation. Dual auth: API Key (paid) or OAuth via ChatGPT (free). Text-to-image, image-to-image, parallel gen, custom sizes.

What it solves

ima2-gen provides a local desktop-like web application that replicates the image and video generation workflows found in high-end AI chat interfaces (like ChatGPT or Grok) but with advanced professional tools. It eliminates the need to switch between multiple web tabs and API dashboards by consolidating generation, iterative editing, and asset management into a single local studio.

How it works

The project acts as a local server and client that interfaces with various AI providers via OAuth or API keys. It supports multiple "paths" for generation:

  • OAuth Paths: Connects to ChatGPT and Grok accounts for free/standard access without needing manual API keys.
  • API Paths: Integrates with OpenAI, xAI (Grok), and Google Gemini (via direct API or the Antigravity CLI).

It uses a single Server-Sent Events (SSE) connection to multiplex progress updates for multiple concurrent generation jobs, preventing browser connection limits from stalling the gallery.

Who it’s for

It is designed for designers, frontend engineers, and AI artists who need a structured, iterative workflow for producing visual assets, including those who want to maintain character consistency across frames or perform targeted image cleanup.

Highlights

  • Iterative Generation Modes: Includes "Classic" for single results, "Node Mode" for branching different creative directions from a single image, and "Multimode" for generating several candidates at once.
  • Video Production: Supports text-to-video, image-to-video, and reference-to-video via Grok models, including keyframe extraction.
  • Storyboard Mode: Maintains character and scene continuity across sequential frames for both images and videos.
  • Canvas Mode: A dedicated cleanup tool for zooming, panning, annotating, and erasing backgrounds with alpha/matte export options.
  • Agent Skills: Provides pre-packaged Markdown instruction sets (Core, Frontend, UI/UX) to help AI coding agents automate asset production pipelines.
  • Local Gallery: Stores generated assets locally with session-aware history and metadata.