stable-diffusion text-to-image image-to-image

PoW : Something

Showcase

Theme

Today it loomed. It loomed over you in a way that was unexplainable. No matter how hard you had tried to articulate yourself, be precise, be… yourself? Something wasn’t working. Your mind felt filled with the weight of sudden change. In a blank room you sat, with nothing but a single mirror before you.

Except… this reflection was not of you… was it? Engraved into the mirror's simplistic design was the etched word: “Something”.

You looked around for a sign or clue to leave, but all that remained was what was before you.

Your first chapter focuses on the feeling of being lost and disconnected from oneself. You are encouraged to explore your own identity and inner selves through artwork that represents your "something" - the feeling or emotion that is difficult to articulate but looms over you. A journey of self-expression and self-discovery - the goal is to capture the essence of the emotion and express it in a way that is unique to you.

Create the beginnings towards a journey of self-expression and self-discovery.

Find your something. Pull it out from within yourself. And embark on this journey with us this week:

“Something, Somewhere, Someone.”

This will be a 3 part story depicting the challenges of facing and overcoming our “somethings”-- and finding “everything”.

Let’s take the first step together.

Model description

This is a model related to the "Picture of the Week" contest on Stable Diffusion discord.

I try to make a model out of all the submission for people to continue enjoy the theme after the even, and see a little of their designs in other people's creations. The token stays "SDArt" and I balance the learning on the low side, so that it doesn't just replicate creations.

The total dataset is made of 50 pictures. It was trained on Stable diffusion 1.5. I used EveryDream to do the training, 100 total repeat per picture. The pictures were tagged using the token "SDArt", and an arbitrary token I choose. The dataset is provided below, as well as a list of usernames and their corresponding token.

The recommended sampling is k_Euler_a or DPM++ 2M Karras on 20 steps, CFGS 7.5 .

Trained tokens

Download links

SafeTensors

CKPT

Dataset

Old version

🧨 Diffusers

This model can be used just like any other Stable Diffusion model. For more information, please have a look at the Stable Diffusion.

You can also export the model to ONNX, MPS and/or FLAX/JAX.

from diffusers import StableDiffusionPipeline
import torch

model_id = "Guizmus/SDArt_something"
pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
pipe = pipe.to("cuda")

prompt = "SDArt oxi"
image = pipe(prompt).images[0]

image.save("./SDArt.png")