---
title: swin2SR-realworld-sr-x4-64-bsrgan-psnr
canonical_url: "https://www.modelscope.cn/models/Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr"
md_url: "https://www.modelscope.cn/models/Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr.md"
repository: Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr
last_updated: 2025-10-01
pipeline_tag: image-to-image
tasks:
  - image-to-image
model_type:
  - swin2sr
architectures:
  - Swin2SRForImageSuperResolution
base_model:
  - caidas/swin2SR-realworld-sr-x4-64-bsrgan-psnr
base_model_relation: finetune
library_name:
  - onnx
  - pytorch
frameworks:
  - pytorch
downloads: 1727
stars: 0
---

# swin2SR-realworld-sr-x4-64-bsrgan-psnr

> swin2SR-realworld-sr-x4-64-bsrgan-psnr - Xenova 在 ModelScope 开源的模型。https://huggingface.co/caidas/swin2SR-realworld-sr-x4-64-bsrgan-psnr with ONNX weights to be compatible with Transformers.js.

Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr 是 ModelScope 魔搭社区上的image-to-image模型，基于 caidas/swin2SR-realworld-sr-x4-64-bsrgan-psnr 构建。

- **Repository**: Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr
- **Tasks**: image-to-image
- **Base model**: caidas/swin2SR-realworld-sr-x4-64-bsrgan-psnr
- **Downloads**: 1727
- **Stars**: 0
- **Last updated**: 2025-10-01

Source: https://www.modelscope.cn/models/Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr

---

https://huggingface.co/caidas/swin2SR-realworld-sr-x4-64-bsrgan-psnr with ONNX weights to be compatible with Transformers.js.

## Usage (Transformers.js)

If you haven't already, you can install the [Transformers.js](https://huggingface.co/docs/transformers.js) JavaScript library from [NPM](https://www.npmjs.com/package/@huggingface/transformers) using:
```bash
npm i @huggingface/transformers
```

**Example:** Upscale an image with `Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr`.
```js
import { pipeline } from '@huggingface/transformers';

// Create image-to-image pipeline
const upscaler = await pipeline('image-to-image', 'Xenova/swin2SR-realworld-sr-x4-64-bsrgan-psnr', {
    dtype: 'fp32', // Options: 'fp32', 'fp16', 'q8', 'q4'
});

// Upscale an image
const url = 'https://huggingface.co/datasets/Xenova/transformers.js-docs/resolve/main/butterfly.jpg';
const output = await upscaler(url);
// RawImage {
//   data: Uint8Array(3145728) [ ... ],
//   width: 1024,
//   height: 1024,
//   channels: 3
// }

// (Optional) Save the upscaled image
output.save('upscaled.png');
```

<details>
  <summary>See example output</summary>

  Input image:
  
  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b253b7ac5ecaae3d1efe0c/eqLyvsErNQvXAFDD2MylF.png)

  
  Output image:

  ![image/png](https://cdn-uploads.huggingface.co/production/uploads/61b253b7ac5ecaae3d1efe0c/_yYB-1NRfobL2wWMIcy3k.png)

</details>

---

Note: Having a separate repo for ONNX weights is intended to be a temporary solution until WebML gains more traction. If you would like to make your models web-ready, we recommend converting to ONNX using [🤗 Optimum](https://huggingface.co/docs/optimum/index) and structuring your repo like this one (with ONNX weights located in a subfolder named `onnx`).