Pretty big update on 12/17.
Soupy does more stuff now.
- Chat with soupy by @tagging it in any channel that it can access.
- Specify a specific channel where soupy responds to all messages.
- !generate an image based on user input.
- !transform an image based on user input.
- Describe the image attached to a message.
A generated image:
The instruction to transform the image:
The transformed image:
An image description:
Note: These are all from the channel in which soupy responds to all messages. That's an environment variable.
To describe an image, simply attach it to a message and @tag the bot.
Pretty big update on 12/10/2023 -- soupy can now analyze images sent to the channel. It will also follow instructions on how you want the image to be analyzed. For example, "describe this as a 5 year old would describe it" or "give a poetic description of this image."
To use this functionality, just tag the bot in a message with an attached image, and optionally give an instruction about how you would like the image to be analyzed.
Soupy-combined combines the functionality of the chat bot with the functionality of the image generation bot.
Currently, soupy-combined.py uses the chat model specified in the .env, and the image !generate uses a hard-coded model which IS NOT referenced in the .env.
- Install dependencies.
- Create a .env and populate it.
- Have fun.