README.md

---
license: openrail++
language:
- en
thumbnail: "https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/thumbnail.png"  
pipeline_tag: text-to-image
tags:
- stable-diffusion
- stable-diffusion-diffusers
inference: true
widget:
- text: >-
    masterpiece, best quality, 1girl, brown hair, green eyes, colorful, autumn,
    cumulonimbus clouds, lighting, blue sky, falling leaves, garden
  example_title: example 1girl
- text: >-
    masterpiece, best quality, 1boy, medium hair, blonde hair, blue eyes,
    bishounen, colorful, autumn, cumulonimbus clouds, lighting, blue sky,
    falling leaves, garden
  example_title: example 1boy
library_name: diffusers
---

<style>
  .title-container {
    display: flex;
    justify-content: center;
    align-items: center;
    height: 100vh; /* Adjust this value to position the title vertically */
  }
  .title {
    font-size: 3em;
    text-align: center;
    color: #333;
    font-family: Arial, sans-serif;
    text-transform: uppercase;
    letter-spacing: 0.05em;
    padding: 0.5em 0;
    box-shadow: 0px 0px 20px 0px rgba(0,0,0,0.15);
    background: transparent;
  }
  .title span {
    background: -webkit-linear-gradient(45deg, #fe6b8b 30%, #ff8e53 90%);
    -webkit-background-clip: text;
    -webkit-text-fill-color: transparent;
  }
  .image-grid {
    display: grid;
    grid-template-columns: repeat(3, 1fr);
    gap: 0.5em;
  }
  .image-item {
    box-shadow: 0px 0px 10px 0px rgba(0,0,0,0.15);
    padding: 10px;
  }
  .image-item img {
    width: 100%;
    height: 100%;
    object-fit: cover;
    border-radius: 10px;
    transition: transform .2s;
  }
  .image-item img:hover {
    transform: scale(1.1);
  }
  .custom-table {
    table-layout: fixed;
    width: 100%;
    border-collapse: collapse;
  }
  .custom-table td {
    width: 50%;
    vertical-align: top;
    padding: 10px;
    box-shadow: 0px 0px 10px 0px rgba(0,0,0,0.15);
  }
  .custom-image {
    width: 100%;
    height: 100%;
    object-fit: cover;
    border-radius: 10px;
    transition: transform .2s; 
  }
  .custom-image:hover {
    transform: scale(1.1);
  }
</style>

<h1 class="title"><span>Hermitage XL</span></h1>

<div class="image-grid">
  <div class="image-item">
    <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/sample1.png">
      <img src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/sample1.png">
    </a>
  </div>
  <div class="image-item">
    <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/sample2.png">
      <img src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/sample2.png">
    </a>
  </div>
  <div class="image-item">
    <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/sample3.png">
      <img src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/sample3.png">
    </a>
  </div>
  <div class="image-item">
    <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/sample4.png">
      <img src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/sample4.png">
    </a>
  </div>
  <div class="image-item">
    <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/sample5.png">
      <img src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/sample5.png">
    </a>
  </div>
  <div class="image-item">
    <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/sample6.png">
      <img src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/sample6.png">
    </a>
  </div>
</div>

<hr>

## Overview

Hermitage XL is a high-resolution, latent text-to-image diffusion model. The model has been fine-tuned using a learning rate of 4e-7 over 5000 steps with a batch size of 16 on a curated dataset of superior-quality anime-style images. This model is derived from Stable Diffusion XL 1.0.

e.g. **_1girl, white hair, golden eyes, beautiful eyes, detail, flower meadow, cumulonimbus clouds, lighting, detailed sky, garden_** 

- Use it with the [`Stable Diffusion Webui`](https://github.com/AUTOMATIC1111/stable-diffusion-webui)
- Use it with 🧨 [`diffusers`](https://huggingface.co/docs/diffusers/index)
- Use it with the [`ComfyUI`](https://github.com/comfyanonymous/ComfyUI)

<hr>

## Features

1. High-Resolution Images: The model trained with 1024x1024 resolution. The model is trained using [NovelAI Aspect Ratio Bucketing Tool](https://github.com/NovelAI/novelai-aspect-ratio-bucketing) so that it can be trained at non-square resolutions.
2. Anime-styled Generation: Based on given text prompts, the model can create high quality anime-styled images.
3. Fine-Tuned Diffusion Process: The model utilizes a fine-tuned diffusion process to ensure high quality and unique image output.

<hr>

## Model Details

- **Developed by:** [Linaqruf](https://github.com/Linaqruf)
- **Model type:** Diffusion-based text-to-image generative model
- **Model Description:** This is a model that can be used to generate and modify anime-themed images based on text prompts. 
- **License:** [CreativeML Open RAIL++-M License](https://huggingface.co/stabilityai/stable-diffusion-2/blob/main/LICENSE-MODEL)
- **Finetuned from model:** [Stable Diffusion XL 1.0](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0)
<hr>

## How to Use:
- Download `Hermitage XL` [here](https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/hermitage-xl.safetensors), the model is in `.safetensors` format.
- You need to use Danbooru-style tag as prompt instead of natural language, otherwise you will get realistic result instead of anime
- You can use any generic negative prompt or use the following suggested negative prompt to guide the model towards high aesthetic generationse:
```
lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry
```
- And, the following should also be prepended to prompts to get high aesthetic results:
```
masterpiece, best quality, illustration, beautiful detailed, finely detailed, dramatic light, intricate details
```
<hr>

## 🧨 Diffusers 

Make sure to upgrade diffusers to >= 0.18.2:
```
pip install diffusers --upgrade
```

In addition make sure to install `transformers`, `safetensors`, `accelerate` as well as the invisible watermark:
```
pip install invisible_watermark transformers accelerate safetensors
```

Running the pipeline (if you don't swap the scheduler it will run with the default **EulerDiscreteScheduler** in this example we are swapping it to **EulerAncestralDiscreteScheduler**:
```py
import torch
from torch import autocast
from diffusers.models import AutoencoderKL
from diffusers import StableDiffusionXLPipeline, EulerAncestralDiscreteScheduler

model = "Linaqruf/hermitage-xl"
vae = AutoencoderKL.from_pretrained("stabilityai/sdxl-vae")

pipe = StableDiffusionXLPipeline.from_pretrained(
    model, 
    torch_dtype=torch.float16, 
    use_safetensors=True, 
    variant="fp16",
    vae=vae
    )

pipe.scheduler = EulerAncestralDiscreteScheduler.from_config(pipe.scheduler.config)
pipe.to('cuda')

prompt = "masterpiece, best quality, 1girl, green hair, sweater, looking at viewer, upper body, beanie, outdoors, watercolor, night, turtleneck"
negative_prompt = "lowres, bad anatomy, bad hands, text, error, missing fingers, extra digit, fewer digits, cropped, worst quality, low quality, normal quality, jpeg artifacts, signature, watermark, username, blurry"

image = pipe(
    prompt, 
    negative_prompt=negative_prompt, 
    width=1024,
    height=1024,
    guidance_scale=12,
    target_size=(1024,1024),
    original_size=(4096,4096),
    num_inference_steps=50
    ).images[0]

image.save("anime_girl.png")
```
<hr>

## Limitation 
1. This model inherit Stable Diffusion XL 1.0 [limitation](https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0#limitations)
2. This model is overfitted and cannot follow prompts well, because it's fine-tuned for 5000 steps with small scale datasets.
3. It's only a preview model to find good hyperparameter and training config for Stable Diffusion XL 1.0

<hr>

## Example

Here is some cherrypicked samples and comparison between available models:

<table class="custom-table">
  <tr>
    <td>
      <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/image1.png">
        <img class="custom-image" src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/image1.png" alt="sample1">
      </a>
      <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/image3.png">
        <img class="custom-image" src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/image3.png" alt="sample3">
      </a>
    </td>
    <td>
      <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/image2.png">
        <img class="custom-image" src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/image2.png" alt="sample2">
      </a>
      <a href="https://huggingface.co/Linaqruf/hermitage-xl/blob/main/sample_images/image4.png">
        <img class="custom-image" src="https://huggingface.co/Linaqruf/hermitage-xl/resolve/main/sample_images/image4.png" alt="sample4">
      </a>
    </td>
  </tr>
</table>