Stable Diffusion Text To Image Demo

Local Setup Instructions

This app demo app was developed with python 3.9 as the target version, however newer version are likely to work as well. It is recommended to use virtual environment locally.

Here are the quick setup commands (for linux or macOS):

# create and activate virtual environment (called venv)
python -m venv venv
source venv/bin/activate

# install dependencies
pip install -r requirements.txt

Environment Variables and Configuration

In the current version of the app we use OctoML's OctoAI service for image generation. The app expect to find 2 environment variables, either set through the command line e.g. export OCTOAI_TOKEN=<token>, or in a .env file.

• OCTOAI_TOKEN needs to have your OctoAI token that you need to have to use the service
  • Instructions to set it up can be found at https://docs.octoai.cloud/docs/how-to-create-an-octoai-access-token
• OCTOAI_PUBLIC_ENDPOINT needs to be set to an OctoAI demo inference point, either a template or one that you owned.
  • This initial model app assumes this endpoint follows the "Stable Diffusion 1.5" template (https://octoai.cloud/templates/stable-diffusion-demo)

Example .env file

OCTOAI_TOKEN=<token>
OCTOAI_PUBLIC_ENDPOINT=<endpoint>

This app also requires XetHub environment variables to support writing back to the project. Make sure to export the following:

XET_USER_NAME=
XET_USER_TOKEN=
XET_USER_EMAIL=