Originally posted on HuggingFace by 22h
Vintedois (22h) Diffusion model trained by Predogl and piEsposito with open weights, configs and prompts (as it should be)
This model was trained on a large amount of high quality images with simple prompts to generate beautiful images without a lot of prompt engineering.
You can enforce style by prepending your prompt with estilovintedois if it is not good enough. It also works well with different aspect ratios, such as 2:3 and 3:2.
It should also be very dreamboothable, being able to generate high fidelity faces with a little amount of steps.
You can use this model commercially or whatever, but we are not liable if you do messed up stuff with it.
Model card
Everything from Stable Diffusion v1-5, plus the fact that this is being built by two indie devs, so it was not extensively tested for new biases.
You can run this concept via diffusers Colab Notebook for Inference