Diffusion Models for Image Generation

Have you ever wanted to see your dog on the moon? Diffusion models can generate images from a prompt text and with our research we proved we could control and manipulate what the model generates.

Comment

Inspiration

We were inspired by our curiosity regarding image generation and generative AI and specifically about diffusion models.

What it does

Our project generates images from a given prompt that we can manipulate to control what is being generated.

How we built it

We combined PEZ and Dreambooth techniques to generate the images.

Challenges we ran into

Since both these techniques are opensource, we ran into challenges with manipulating the code to accomplish our goal such as editing images, and loading in the models.

Accomplishments that we're proud of

We created some really cool images of our robot Bob the Bagel Guy!

What we learned

We learned how to create and manipulate diffusion model for image generation.

What's next for Diffusion Models for Image Generation

Fine-tuning the techniques to allow for more accurate images to a given prompt.

Built With

  • google-collab
  • huggingface
  • python
Share this project:

Updates