pytorch diffusers stable-diffusion text-to-image diffusion-models-class dreambooth-hackathon wildcard

DreamBooth model for the hasbulla concept trained by carlosabadia on the carlosabadia/hasbulla dataset. DreamBooth Hackaton's Winner! 🏆

This is a Stable Diffusion model fine-tuned on the hasbulla concept with DreamBooth. It can be used by modifying the instance_prompt: hasbulla person

This model was created as part of the DreamBooth Hackathon 🔥. Visit the organisation page for instructions on how to take part!

Description

This is a Stable Diffusion model fine-tuned on Hasbulla images for the wildcard theme.

It was also featured in Hasbulla's Twitter account!

<blockquote class="twitter-tweet"><p lang="sv" dir="ltr">Hasbulla Van Gogh <a href="https://t.co/5f0uPKhi6U">pic.twitter.com/5f0uPKhi6U</a></p>— Hasbulla 🐐 (@HasbullaHive) <a href="https://twitter.com/HasbullaHive/status/1610729157268234240?ref_src=twsrc%5Etfw">January 4, 2023</a></blockquote> <script async src="https://platform.twitter.com/widgets.js" charset="utf-8"></script>

Images generated by model

<img src=https://i.imgur.com/Sqfr7ae.jpg width=70% height=70%>

Gradio & Colab

Model supported in a Gradio Web UI and Colab: Open In Spaces Open In Colab

Usage

import torch
from diffusers import StableDiffusionPipeline, DPMSolverMultistepScheduler

pipe = StableDiffusionPipeline.from_pretrained(
    "carlosabadia/hasbulla", 
    scheduler = DPMSolverMultistepScheduler.from_pretrained("carlosabadia/hasbulla", subfolder="scheduler"),
    torch_dtype=torch.float16,
).to("cuda")

guidance_scale = 7
prompt = "A portrait of hasbulla person"

images = pipe(prompt, num_images_per_prompt=1, num_inference_steps=25, guidance_scale=guidance_scale).images
image = images[0]
image.save("hasbulla.png")

visitors