Skip to content

README.en.md

Latest commit

 

History

History
154 lines (108 loc) · 6.71 KB

README.en.md

File metadata and controls

154 lines (108 loc) · 6.71 KB
Raw

简体中文 · English

Intertwine-Ai-Image-Enhanced

The image workspace for everything after Generate.
Prompt. Reference. Refine. Animate. Ship.

A local-first, task-first image workspace for creators who want more than a one-shot result.

Vue 3 FastAPI Python 3.11+ AGPL-3.0-or-later

Keep the source. Keep the context. Keep the good version. Intertwine-Ai-Image-Enhanced turns every generation into a task you can inspect, recover, refine, and ship.

Intertwine-Ai-Image-Enhanced workspace

A real workspace capture — not a marketing mockup.

Image creation is not a one-click problem

The click is easy. The work after the click is where the craft lives: compare references, tune the prompt, preserve the winning version, retry a failed request, remove the background, create transparent assets, build a GIF, or come back to the same task next week.

Intertwine-Ai-Image-Enhanced is built for that part.

It gives creators a focused Vue workspace and gives operators a Python backend that owns tasks, files, authentication, and provider connections. Bring your own model and provider. Keep the workflow you trust. Take the final assets with you.

One workspace. The whole creative loop.

When you need to… You get…
Start from an idea Text-to-image, Chinese prompt presets, and high-quality defaults
Create from references Image-to-image, multiple references, editing, and masks
Finish a usable asset Transparent PNG/WebP, size, format, and compression controls
Iterate without losing context Persistent task history, preview, retry, continue, cancel, and delete
Make a still image move Animation frames, GIF generation, preview artifacts, and export
Build a desktop pet Row-based frames, contact sheets, spritesheets, and pet.json packaging
Connect your own stack Images / Responses protocols, model discovery, and capability-aware controls

Why it holds up beyond the first result

Tasks, not disposable requests

Every task has an explicit lifecycle: queued, running, succeeded, failed, or cancelled. Inputs and task metadata are persisted before remote work begins. Refresh the page and the workspace can reconstruct the state from the backend.

References and prompts stay together

Prompt presets, source images, masks, outputs, and animation artifacts are organized around the task. Continue from a result instead of trying to remember what produced it.

Bring your own provider

Use Images- or Responses-compatible upstreams, discover available models, and expose controls based on the active provider’s capabilities. Your workflow is not locked to a single site or a single model pool.

From generation to delivery

Transparent PNG/WebP, compression, sizing, GIFs, frame previews, contact sheets, and ZIP exports are part of the same task-oriented workflow. Generation is the beginning; a reusable asset is the finish line.

Local-first, with a clear security boundary

  • The browser calls only local /api/* endpoints.
  • Provider URLs, API keys, and raw upstream Authorization headers stay in the backend.
  • SQLite stores task metadata and relative paths; image and animation binaries stay in the configured data directory.
  • Generated files are scoped to the authenticated user and downloaded through authenticated endpoints.
  • For public deployment, terminate TLS with your own reverse proxy. Plain HTTP is suitable only for temporary, non-private setups.

Architecture

Vue 3 creative workspace
        │ local /api requests
        ▼
Python + FastAPI task service
        ├── Images / Responses provider clients
        ├── model and capability discovery
        ├── task management, retry, and cancellation
        ├── SQLite task metadata
        └── user-scoped image, frame, and GIF artifacts

One Python service can serve the built frontend and API from the same port. The frontend receives capability and status information, never raw provider secrets.

Quick start

Requirements

  • Node.js 18+ and npm
  • Python 3.11+
  • Dependencies from backend/requirements.txt

Install, build, run

npm install
npm run build
./start.sh

Windows:

npm install
build.bat
start.bat

Open the local address printed by the launcher, then configure a provider from the administration/settings UI.

Configuration and data

  • backend/.env or process environment: host, port, data directory, timeouts, and optional defaults.
  • Admin/provider settings: upstream base URL, wire protocol, model pool, and provider tokens.
  • SQLite + data directory: tasks, source material, generated images, animation frames, and exports.

Do not commit .env files, credentials, runtime data, logs, caches, build output, or local release archives.

Repository map

src/                         Vue application and task workspace
backend/app/                 Python configuration, storage, providers, tasks, and API
backend/tests/               Backend regression and security tests
public/                      Logo, favicon, and static assets
docs/project/                Product and architecture guides
docs/build/                  Build, verification, and release instructions
docs/assets/                 Public README screenshots

Verify

python -m compileall backend/app
python -m unittest discover backend/tests
npm run lint:check
npm run build

More documentation:

License

Intertwine-Ai-Image-Enhanced is released under the GNU Affero General Public License v3.0 or later. See LICENSE for the complete terms. If you deploy a modified version as a network service, AGPL-3.0-or-later requires users to be offered the corresponding source code.