---
library_name: diffusers
pipeline_tag: text-to-image
---

# Style Portrait

![preview](sample.png)

Style Portrait is a SD-v1.5 based text-to-image generation model. It can generate high-quality 512*832 images of Asia portrait.

This repository contains checkpoints for diffusers and SD-WebUI.

# Checkpoints

- style-portait-v1-5-fp16.safetensors: Checkpoint for Stable-Diffusion-WebUI.
- Others: Checkpoint for Diffusers.


# Diffusers Usage

```python3
import torch

from diffusers import DiffusionPipeline


model_id = "Tramac/style-portrait-v1-5"
pipeline = DiffusionPipeline.from_pretrained(model_id, torch_dtype=torch.float16)
pipeline = pipeline.to("cuda")

prompt = "1girl, long hair, close-up, elegant, upper body, natural skin texture, highly detailed"
negative_prompt = "cloned face, ugly, out of frame, mutated, extra arms, cross-eye, blurry, bad art, bad anatomy, badhands, watermark, text, long neck, worst quality, low quality"
image = pipeline(
    prompt=prompt,
    negative_prompt=negative_prompt,
    height=832,
    width=512,
).images[0]

image.save("demo.png")
```

# WebUI Usage

1. Download the checkpoint (style-portait-v1-5-fp16.safetensors) to /Stable-Diffusion-WebUI/models/Stable-diffusion.
2. Start up WebUI and load checkpoint.