Four ways to start using Open Source DALL-e today

Dall-E is a neural network(an Artificial Intelligence technology) that can generate images from text inputs in natural language.

Here are ways to use open source version of Dall-E today, with easy of use easy to harder as we go down the list:

  • Use Dall-E from HuggingFace Website
  • Use Dall-E Playground
  • Install Dall-E Playground on your computer
  • Install Dall-E Playground and backend on your computer

Let’s go through them one by one

Use Dall-E from HuggingFace Website without installing anything

Skill Level: Beginner

  1. Open HuggingFace Dall-E Application
  2. Enter a prompt and press generate your images

That’s it. This is the most easy way to use open source Dall-E currently.

However, if you are running into too much traffic error, you can try to install Dall-E Playground on your computer as listed in below section

Use Dall-E Playground without installing anything

Skill Level: Intermediate

For this method, it helps if you know your way around Google Colab.

There are two main parts of having DALL-E Playground running on your computer:

  1. DALL-E backend
  2. DALL-E Playground or DALL-E Web App

Below steps cover both of them.

Get DALL-E backend up and running on Google Colab

  1. Open Dall-e backend notebook in Google Colab
  2. Click on Runtime menu and click on Run all
  3. Once finished, get the link from the bottom of the Google Colab notebook. It should look like this: `https://xxxxxx.loca.lt)
    If you run into problems, try selecting a different model.

Get DALL-E Playground in your browser

  1. Open the Dall e Playground in your browser
  2. Input the url from the backend setup
  3. Enter your prompt and click enter to generate your dall e images

Install Dall-E Playground on your computer

Skill Level: Intermediate

For this method, it helps if you know your way around command line terminal.

Let’s setup the backend first.

  1. Open Dall-e backend notebook in Google Colab
  2. Click on Runtime menu and click on Run all
  3. Once finished, get the link from the bottom of the Google Colab notebook. It should look like this: https://xxxxxx.loca.lt
    If you run into problems, try selecting a different model.

Now, time to install the Dall-E Playground frontend application

  1. Open your terminal

  2. get the code for the open source Dall-E Playground by running the following command:
    git clone https://github.com/saharmor/dalle-playground.git
    If it says git: command not found, you need to install git on your computer. Here is a guide to install git.

  3. cd to the directory you just cloned

  4. cd to the interface directory

  5. run below command
    npm install
    This installs the required node modules.

  6. If everything goes well, so far, start the Dall-E Playground by running the following command:
    npm start
    by default this will start the playground on port 3000

  7. Open your browser and visit - http://localhost:3000

  8. Input the url from the Google colab.

  9. enter your prompt and click enter to generate your dall e images

Install Dall-E Playground and backend on your computer

Skill Level: Advanced

In this method, it helps if you know your way around using command line and can install python packages.

Also, you need to have a computer or virtual machine that has a GPU attached.

As in previous methods, you need to have a backend and a frontend running. Difference in this method is both the backend and the frontend are running on your computer.

The steps to install the front end web application are the same as in the previous method. We will just cover the backend installation.

DALL-E backend installation:

  1. Open your terminal
  2. get the code for the open source Dall-E backend code by running the following command:
    git clone https://github.com/saharmor/dalle-playground.git
    If it says git: command not found, you need to install git on your computer. Here is a guide to install git.
  3. cd to the directory you just cloned above
  4. cd to the backend directory
  5. Install required python packages by running the following command:
    ‘pip install -r requirements.txt’
  6. Run the backend server by running the following command:
    ‘python app.py –port 8000 –model_version mini’
    If everything goes well, your Dall-E backend is running on http://local:8000/.

You can now use the above URL in the Dall-E Playground you setup earlier.

Conclusion

Generating art using AI is very exciting and fun. Dall-E is a great way to generate art using AI. One of the above methods will deifnitely be the best way to use Dall-E today.