Upload README.md
Browse files
README.md
CHANGED
|
@@ -1,23 +1,57 @@
|
|
| 1 |
---
|
| 2 |
-
license:
|
| 3 |
-
datasets:
|
| 4 |
-
- AlexZheng/galactic-animation
|
| 5 |
-
language:
|
| 6 |
-
- en
|
| 7 |
-
pipeline_tag: text-to-image
|
| 8 |
tags:
|
| 9 |
-
-
|
|
|
|
| 10 |
---
|
|
|
|
|
|
|
|
|
|
| 11 |
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
|
| 15 |
-
|
| 16 |
-
|
| 17 |
-
|
| 18 |
-
|
| 19 |
-
|
| 20 |
-
|
| 21 |
-
|
| 22 |
-
|
| 23 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
+
license: creativeml-openrail-m
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3 |
tags:
|
| 4 |
+
- stable-diffusion
|
| 5 |
+
- text-to-image
|
| 6 |
---
|
| 7 |
+
# Arcane Diffusion
|
| 8 |
+
This is the fine-tuned Stable Diffusion model trained on images from the TV Show Arcane.
|
| 9 |
+
Use the tokens **_arcane style_** in your prompts for the effect.
|
| 10 |
|
| 11 |
+
**If you enjoy my work, please consider supporting me**
|
| 12 |
+
[](https://patreon.com/user?u=79196446)
|
| 13 |
+
|
| 14 |
+
### 🧨 Diffusers
|
| 15 |
+
|
| 16 |
+
This model can be used just like any other Stable Diffusion model. For more information,
|
| 17 |
+
please have a look at the [Stable Diffusion](https://huggingface.co/docs/diffusers/api/pipelines/stable_diffusion).
|
| 18 |
+
|
| 19 |
+
You can also export the model to [ONNX](https://huggingface.co/docs/diffusers/optimization/onnx), [MPS](https://huggingface.co/docs/diffusers/optimization/mps) and/or [FLAX/JAX]().
|
| 20 |
+
|
| 21 |
+
```python
|
| 22 |
+
#!pip install diffusers transformers scipy torch
|
| 23 |
+
from diffusers import StableDiffusionPipeline
|
| 24 |
+
import torch
|
| 25 |
+
|
| 26 |
+
model_id = "nitrosocke/Arcane-Diffusion"
|
| 27 |
+
pipe = StableDiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
|
| 28 |
+
pipe = pipe.to("cuda")
|
| 29 |
+
|
| 30 |
+
prompt = "arcane style, a magical princess with golden hair"
|
| 31 |
+
image = pipe(prompt).images[0]
|
| 32 |
+
|
| 33 |
+
image.save("./magical_princess.png")
|
| 34 |
+
```
|
| 35 |
+
|
| 36 |
+
# Gradio & Colab
|
| 37 |
+
|
| 38 |
+
We also support a [Gradio](https://github.com/gradio-app/gradio) Web UI and Colab with Diffusers to run fine-tuned Stable Diffusion models:
|
| 39 |
+
[](https://huggingface.co/spaces/anzorq/finetuned_diffusion)
|
| 40 |
+
[](https://colab.research.google.com/drive/1j5YvfMZoGdDGdj3O3xRU1m4ujKYsElZO?usp=sharing)
|
| 41 |
+
|
| 42 |
+
|
| 43 |
+
|
| 44 |
+
### Sample images from v3:
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
### Sample images from the model:
|
| 48 |
+
|
| 49 |
+
### Sample images used for training:
|
| 50 |
+
|
| 51 |
+
|
| 52 |
+
**Version 3** (arcane-diffusion-v3): This version uses the new _train-text-encoder_ setting and improves the quality and edibility of the model immensely. Trained on 95 images from the show in 8000 steps.
|
| 53 |
+
|
| 54 |
+
**Version 2** (arcane-diffusion-v2): This uses the diffusers based dreambooth training and prior-preservation loss is way more effective. The diffusers where then converted with a script to a ckpt file in order to work with automatics repo.
|
| 55 |
+
Training was done with 5k steps for a direct comparison to v1 and results show that it needs more steps for a more prominent result. Version 3 will be tested with 11k steps.
|
| 56 |
+
|
| 57 |
+
**Version 1** (arcane-diffusion-5k): This model was trained using _Unfrozen Model Textual Inversion_ utilizing the _Training with prior-preservation loss_ methods. There is still a slight shift towards the style, while not using the arcane token.
|