Skip to content
The Chat Archive
Human-written demonstration (OpenAssistant volunteers)· OpenAssistant (human-written) OpenAssistant Conversations (oasst2)

How would I make a procedurally generated desktop background on Linux?

In OpenAssistant data, “assistant” turns were written by human volunteers as demonstrations. They are not model outputs.
  1. User

    How would I make a procedurally generated desktop background on Linux?

  2. Assistant

    Here's a script to do so:

    import numpy as np
    from PIL import Image
    
    def generate_background():
        # Generate a random array of RGB values
        rgb = np.random.randint(0, 256, size=(1080, 1920, 3), dtype=np.uint8)
        
        # Create an image from the array
        img = Image.fromarray(rgb)
        
        # Save the image to disk
        img.save("background.jpg")
        
        # Set the image as your desktop background
        # Replace "feh" with your preferred tool for setting the desktop background
        # on your Linux system
        !feh --bg-fill background.jpg
    
    if __name__ == "__main__":
        generate_background()
    
    
  3. User

    This just generates random noise.

    Could you provide a procedural generation algorithm that will generate a mountain terrain with blue skies dotted with white, fluffy clouds?

  4. Assistant

    Here is a program to generate images based on your custom prompt "a mountain terrain with blue skies dotted with white, fluffy clouds".

    1. Install packages on your selected python interpreter
    pip install diffusers==0.12.1 torch==1.13.1
    
    
    1. Now we are ready to create a stable diffusion pipeline
    import torch as T
    from difusers import StableDiffusionPipeline as SDP
    
    # define pipeline and prompt
    prompt = "a mountain terrain with blue skies dotted with white, fluffy clouds"
    pipeline = SDP.from_pretrained("runwayml/stable-diffusion-v1-5", torch_dtype=T.float16)
    pipeline = pipeline.to("cuda")
    
    
    1. use the instantiated pipeline object to generate an image and save it in the current directory
    # generate and save image
    image = pipeline(prompt).images[0]
    image.save("./" + prompt + ".png")
    
    

    This program uses runwayml/sable-diffusion-v1-5 from huggingface and is implemented in torch, if you want me to switch to different versions please let me know !

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.