StableDiffusionTelegram is a telegram bot that allows to generate images using the Stable Diffusion AI from a telegram bot, in a much more comfortable and simple way. This bot can generate images from a text input or from an image with caption. In addition, given any response from the bot, a new try or a variation of the attempt can be generated.
Try the bot with colab, create a token of Hugging Face Hub and Telegram and deploy the bot with colab.
-
Install PyTorch (https://pytorch.org/get-started/).
-
Install requirements:
pip install -r requirements.txt
-
Register in Hugging Face Hub and create a token (https://huggingface.co/docs/hub/security-tokens).
-
Use the token to login in huggingface-cli:
$ huggingface-cli login
-
Talk to BotFather and create a bot (https://t.me/BotFather).
-
Create a .env file with the telegram token and the safe content option (if false, explicit content will be displayed, otherwise set to true). If you have memory problems, you can lower the width and height to lower numbers, such as 448 or 320. An example of .env is the following:
TG_TOKEN="YOUR_TOKEN_IS_HERE"
SAFETY_CHECKER="false"
HEIGHT="512"
WIDTH="512"
- Run the bot
python bot.py
| Generating image from text | Generating a variation | Generating a new image from an user photo |
|---|---|---|