https://github.com/abra5umente/auto1111
terminal based wrapper for Stable Diffusion Web UI, written in Go
https://github.com/abra5umente/auto1111
automatic1111 automation cli-tool generative-ai-projects go image-generation stable-diffusion stable-diffusion-webui
Last synced: 24 days ago
JSON representation
terminal based wrapper for Stable Diffusion Web UI, written in Go
- Host: GitHub
- URL: https://github.com/abra5umente/auto1111
- Owner: abra5umente
- License: mit
- Created: 2025-07-06T04:49:39.000Z (11 months ago)
- Default Branch: main
- Last Pushed: 2025-07-15T16:51:18.000Z (11 months ago)
- Last Synced: 2025-07-16T12:52:55.708Z (11 months ago)
- Topics: automatic1111, automation, cli-tool, generative-ai-projects, go, image-generation, stable-diffusion, stable-diffusion-webui
- Language: Go
- Homepage:
- Size: 6.92 MB
- Stars: 1
- Watchers: 0
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Go Auto1111 Launcher & Web GUI
A unified toolkit for Stable Diffusion on your own hardware. Run it **headless** from the desktop or via a **browser‑friendly Web GUI**—both now share one code‑base and identical JSON‑driven settings.
---
## ✨ What’s New (July 2025)
| Area | Added / Changed |
| ---------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Desktop CLI** | • *Automatic* PNG naming in `/output`
• Robust error handling (non‑zero exit codes bubble up) |
| **Web GUI** | • Drag‑and‑drop **`.json` settings** upload
• Dropdown recall of uploaded presets
• Generates images with a single click (prompt + selected JSON) |
| **Config split** | • `.env` → **system‑level** stuff (paths, model folders, env BATs)
• `settings.json` → **generation parameters** (sampler, scheduler, width × height, steps, cfg scale) |
---
## 1. Directory Layout
```text
project/
│ README.md
│ requirements.txt ← Python deps (legacy FastAPI)
│ webapp.exe ← built full-Go web server (optional)
│
├─ backend/ ← generator.exe + legacy FastAPI
│ ├─ generator.exe ← built Go generator
│ ├─ settings.json ← symlink / copy chosen preset
│ └─ settings/ ← uploaded *.json presets
│
├─ output/ ← generated PNGs (auto‑timestamped)
│
├─ source/
│ ├─ *.go ← generator sources
│ └─ cmd/webapp/ ← full-Go web server (serves UI + API)
│
└─ auto1111_webapp/
├─ frontend/ ← React + Vite + Tailwind (shadcn/ui)
└─ app/ ← build output served at /app
```
---
## 2. Installation & Build
### ⚙️ Prerequisites
* **Go 1.21+** (for the CLI generator)
* **Python 3.10+** (for FastAPI & AUTOMATIC1111)
* **Stable Diffusion WebUI** (AUTOMATIC1111) cloned locally
* *Windows users*: `ENVIRONMENT_BAT` & `AUTO1111_BAT` paths configured in `.env`
### 🐍 Python deps (Web GUI)
```bash
python -m venv venv
venv\Scripts\activate # or `source venv/bin/activate`
pip install -r requirements.txt
```
`requirements.txt` generated today:
```text
fastapi
uvicorn[standard]
python-multipart
pydantic
```
### 🛠️ Build & Run — Desktop CLI
```bash
cd backend
# Build once
go build -o generator.exe .
# Single image
./generator.exe --prompt "a cat in space" --settings settings/sample.json
```
Images appear in `output/`.
### 🌐 Build & Run — Web GUI (legacy FastAPI)
```bash
cd backend
# 1. build Go binary (same as above)
# 2. start FastAPI server (reload disabled for Go exe stability)
uvicorn main:app --host 0.0.0.0 --port 8000
```
Open [http://localhost:8000/app/](http://localhost:8000/app/) in your browser.
On Windows, you can use `startup.bat` to launch both the frontend (Vite dev server) and backend together. It installs frontend deps if missing and opens the dev server in a new window.
```bat
startup.bat
```
#### Frontend (React + shadcn/ui)
The frontend lives in `auto1111_webapp/frontend` and builds into `auto1111_webapp/app` (served by FastAPI at `/app`). The legacy static HTML has been removed.
Setup and build:
```bash
cd auto1111_webapp/frontend
npm install
npm run build # outputs to ../app with base /app/
```
For local development (hot reload) without the batch script:
```bash
cd auto1111_webapp/frontend
npm run dev # http://localhost:5173
```
The dev server hits the same API endpoints (`/generate`, `/history`, etc.).
### 🖱️ One‑click Windows app (full Go)
Build a single Go web server that serves the UI, exposes the API, and shells out to `generator.exe`. No Python needed at runtime.
1) Build frontend (once, or whenever you change it):
```bash
cd auto1111_webapp/frontend
npm install
npm run build # outputs to ../app with base /app/
```
2) Build server (from repo root):
```bash
go build -o webapp.exe ./source/cmd/webapp
```
3) Run:
- Double‑click `webapp.exe` (or run it from a terminal)
- Opens http://127.0.0.1:8000/app/
- Press Ctrl+C to quit
Runtime requirements:
- `backend/generator.exe` present (build with `go build -o backend/generator.exe ./source`)
- `backend/.env` for AUTO1111 paths and keys (the generator reads `.env` beside itself)
- By default, the server serves UI from disk at `auto1111_webapp/app`.
- To bake the UI into the binary, first copy the built assets into `source/cmd/webapp/app`, then build with a tag:
```bash
REM after running npm run build (which outputs ../app)
xcopy /E /I /Y auto1111_webapp\app source\cmd\webapp\app
go build -tags embedui -o webapp.exe ./source/cmd/webapp
```
With `-tags embedui`, the binary contains the UI and does not need the `auto1111_webapp/app` folder at runtime.
---
## 3. Using the Web GUI
1. **Upload** a `*.json` preset via **Settings → Upload**. The file is saved to `backend/settings/` and instantly appears in the dropdown.
2. **Select** the preset from the dropdown. Its parameters (sampler, width, etc.) are shown for confirmation.
3. Type your **prompt** (or leave blank to auto‑enhance) and press **Generate**.
4. The resulting PNG streams back and is also written to `output/`.
> **Tip:** presets can be version‑controlled. Commit the `settings/` folder to Git for easy sharing.
---
## 4. Roadmap (July 2025)
| Feature | Status |
| ------------------------------------------ | --------------------------------------------------------------------------- |
| **Prompt History** | ✅ Planned – quick recall in GUI |
| **Style Presets** | 🚧 Expanding – now driven by uploaded JSON; will ship with curated defaults |
| **Negative Prompt Generation** | ✅ Keeps bad stuff out automatically |
| **Model Selection** | ✅ Currently relies on whatever is set in Automatic1111, but will obey `.env` / JSON overrides soon |
| **GUI Interface** | 🚧 MVP shipped; polishing UI & auth next |
| **Prompt Templates (JSON import)** | ✅ Core to new workflow |
| **Social Features** (share prompts / PNGs) | ✅ Still on the slate |
| **Containerisation** (Docker) | ✅ Working on this |
---
## 5. Troubleshooting
| Symptom | Likely Cause | Fix |
| ---------------------------------- | -------------------------------------------- | ------------------------------------------------------------------------------------------------------ |
| `generator.exe failed: …` | Invalid JSON (missing comma or wrong key) | Validate with `jq . settings/your.json` |
| Image 512×512 despite bigger width | JSON keys wrong *or* values saved as strings | Keys are `IMAGE_WIDTH/IMAGE_HEIGHT` or `width/height` (caps or lower). Numbers **must not** be quoted. |
| `No .env file found` from AUTO1111 | Working dir incorrect for the child process | Confirm `start_auto1111()` sets `cmd.Dir` to `backend/` |
| PNG missing but no error | Output path invalid | Ensure `output/` exists & that you have write permissions |
---
## 6. FAQ
> **Q:** Do I need both `.env` *and* `settings.json`?
>
> **A:** Yes. Think of `.env` as *system settings* (where Python lives, which SD model to load, API keys). `settings.json` is *per‑image parameters* (sampler, size, steps). Separate files keep your secrets out of Git while letting you share artistic presets.
---
© 2025 MIT‑licensed. Happy rendering! 🚀