---
license: cc-by-4.0
pipeline_tag: image-to-image
tags:
- pytorch
- super-resolution
---

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/4xNomosUniDAT_otf)

# 4xNomosUniDAT_otf

Name: 4xNomosUniDAT_otf  
Author: Philip Hofmann  
Release: 23.09.2023   
License: CC BY 4.0  
Network: DAT  
Scale: 4  
Purpose: 4x multipurpose upscaler otf version (resize, jpg, noise, blur)  
Iterations: 150000  
epoch: 70  
batch_size: 8  
HR_size: 128  
Dataset: nomosuni  
Number of train images: 2989  
OTF Training: Yes  
Pretrained_Model_G: DAT_x4.pth  

Description:
4x multipurpose dat upscaler that handles resize, jpg compression, noise and blur. Trained on DAT with Adan, U-Net SN, huber pixel loss, huber perceptual loss, vanilla gan loss and huber color loss, on paired nomos_uni (universal dataset containing photographs, anime, text, maps, music sheets, paintings ..).

Some Imgsli examples:  
[Real degraded photo](https://imgsli.com/MjA4NjMw)  
[JPG Compression](https://imgsli.com/MjA4NjI2)  
[Noisy Photo](https://imgsli.com/MjA4NjI4)  
[Manga](https://imgsli.com/MjA4NjMx)  

![Example1](https://github.com/Phhofm/models/assets/14755670/0bc23ace-c406-408a-b19b-2743319625b4)
![Example2](https://github.com/Phhofm/models/assets/14755670/c0dcde4f-d5cf-4e44-9a97-5aa2efc161a7)
![Example3](https://github.com/Phhofm/models/assets/14755670/7d787f9a-dd7d-4947-8399-8730b9bdc9b9)
![Example4](https://github.com/Phhofm/models/assets/14755670/755c8503-0f16-4898-bb2b-a23969dce11b)