A telegram bot integrating with the AUTOMATIC1111 stable diffusion interface to bring genAI images directly to your chats.
| Command | Parameters | Function |
|---|---|---|
| /picgen | prompt | Generate a new image based on the prompt |
| /variation | an image + a text prompt | Generate a variation of the supplied image |
| /variation | replied to an image + a text prompt | Generate a variation of the supplied image |
| /safemode | on|off | Turn spoiler filtered images on or off |
| /teach | %keyword replacement text | Teach a word that you can substitue in prompts using the % sign as a marker |
| /forget | %keyword | Frogets the taught keyword |
| /mywords | Display all your known words |
- The bot works based on user + chat combinations. On first image generation, the user and chat combination will be logged into a table in the database, and will be used for the custom commands
- Safemode status and words are user+chat specific
- The user/chat combination is only logged on image generation; safemode and aliasing will not work until at least one image has been generated
Stable Diffusion is not part of the package; you will need to first set up the SD Web UI, and launch it in API mode (--noui). Once you have it set, create a telegram bot using @BotFather. You can set all details as you like, the bot is 100% in your control.
Set the bot commands to these:
start - Start the bot picgen - Create new image from prompt variation - Create a variation from the image teach - Teach aliases to be replaced in the prompts forget - Forget a learned alias mywords - Get a list of all your aliases safemode - Toggle spoiler filter mode Grab your bot token, and bot username. and paste them into the docker-compose.yml env variables:
version: "3.9"services: dilly-dalle-sd: container_name: "dilly-dalle-sd-bot"build: .restart: alwaysenvironment: - "LOGLEVEL=ERROR"# DEBUG, INFO, WARNING, ERROR, CRITICAL - "TELEGRAM_BOT_TOKEN=" - "STABLE_DIFFUSION_URL=" - "DATABASE_URL=/app/data/sqlite/dilly-dalle-sd.db"# Only modify if you want different volume mappings. If you change the filename update entrypoint.sh - "STEPS=20"# The number of steps when creating the imagevolumes: - ./data:/app/data # sqlite db and generated imagesMake sure to adjust the STABLE_DIFFUSION_URL to point to your host address if you're running SD on a different machine.
Afterwards, start the container with:
docker compose up -dThis will build the image and start a container. The generated images and the back-end sqlite database will be stored in the data directory created when starting.
This code is being released under CC BY-NC-SA 4.0. See the LICENSE file for details.
Additionally, the following third party libraries are used, without modification:
| Library | Version | License |
|---|---|---|
| python-telegram-bot | 21.2 | GNU Lesser General Public License v3.0 |
| requests | 2.31.0 | Apache License 2.0 |
| Pillow | 10.2.0 | Historical Permission Notice and Disclaimer (HPND) |
| apsw | 3.45.1.0 | Open Source License |
General Notices This software dynamically links to libraries that are licensed under licenses differing from CC BY-NC-SA 4.0. Users are informed of their rights under this license and can obtain the source code of these libraries from the respective repositories linked above.
Please use the github issues page to report bugs and problems.
If you want to contribute, feel free to buy me a coffee.