A 1x model designed to reduce artifacts and restore detail to images upscaled by 4xFrankendata_FullDegradation_SRFormer. It could possibly work with other upscaling models too.

Frankenfixer_SRFormerLight

[Github Release Link](https://github.com/Phhofm/models/releases/Ludvae200)

Name: Ludvae200   
License: CC BY 4.0   
Author: Philip Hofmann   
Network: [LUD-VAE](https://github.com/zhengdharia/LUD-VAE)   
Scale: 1   
Release Date: 25.03.2024   
Purpose: 1x realistic noise degradation model    
Iterations: 190'000   
H_size: 64   
n_channels: 3   
dataloader_batch_size: 16    
H_noise_level: 8   
L_noise_level: 3   
Dataset: [RealLR200](https://drive.google.com/drive/folders/1L2VsQYQRKhWJxe6yWZU9FgBWSgBCk6mz)   
Number of train images: 200   
OTF Training: No   
Pretrained_Model_G: None   

Description:
1x realistic noise degradation model, trained on the [RealLR200](https://drive.google.com/drive/folders/1L2VsQYQRKhWJxe6yWZU9FgBWSgBCk6mz) dataset as found released on the [SeeSR](https://github.com/cswry/SeeSR) github repo.
Next to the ludvae200.pth model file, I provide a ludvae200.zip file which not only contains the code but also an inference script to run this model on the dataset of your choice.
Adapt the ludvae200_inference.py script accordingly by adjusting the file paths at the beginning section, to your input folder, output folder, the folder path holding the ludvae200.pth model, and a folder path where you want the text file to be generated. I made the textfile generation the same way as I did in [Kim's Dataset Destroyer](https://github.com/Kim2091/helpful-scripts/tree/main/Dataset%20Destroyer), which means you will have each image file logged with each of the values used to degrade that specific image file in the resulting text file, which will append only and never overwrite.

You can also adjust the strength settings inside the inference script file to fit to your needs. If you in general want less strong noise for example, you should adjust the temperature upper limit from 0.4 to 0.2 or go even lower.
So in line 96 change "temperature_strength = uniform(0.1,0.4)" to "temperature_strength = uniform(0.1,0.2)" just to give an example.

These values are defaulted to my needs of my last dataset degradation workflow I used, but feel free to adjust these values. You can also do the same as I did, temporarily using deterministic values with multiple runs to determine the min and max values of noise generation you deem suitable for your dataset needs.

See the examples of what this looked like for my last dataset workflow I used my model in.

Ludvae200

Category: Foliage/Ground
Purpose: bad upscale

Images of plants, trees or other foliage upscaled with Photoshop Preserve Details 2.0. Sharpens and "subdivides" details and noise so it doesn't look upscaled.

Plants

Purpose: Correct old color photos that are tinted red

Correct old color photos that are tinted red. Model only tested in nature.

RedImage10000

A 2x photo upscaling compact model based on microsofts image pairs. This was one of the earliest models I started training and finished it now for release.

Parimg Compact

Description: 4x photo upscaler for faces, trained on the FaceUp dataset. These models are an improvement over the previously released 4xFFHQDAT and are its successors. These models are released together with the FaceUp dataset, plus the accompanying [youtube video](https://www.youtube.com/watch?v=TBiVIzQkptI)

This model comes in 4 different versions:  
4xFaceUpDAT (for good quality input)  
4xFaceUpLDAT (for lower quality input, can additionally denoise)  
4xFaceUpSharpDAT (for good quality input, produces sharper output, trained without USM but sharpened input images, good quality input)  
4xFaceUpSharpLDAT (for lower quality input, produces sharper output, trained without USM but sharpened input images, can additionally denoise)  

I recommend trying out 4xFaceUpDAT

4xFaceUpDAT

4xFaceUpSharpDAT

A normal ESRGAN model without degradations and without any pretrain, simply an RRDBNet model trained on paired dataset (4x downsampled) on the full LSDIR dataset (84,991 images / 165 GB)

4xLSDIR

Upscale small good quality photos to 4x their size

My first ever model 😄 Well, it’s not the best, but, it’s something 😉 
I provide some 15 examples from the validation set here for you to visually see the generated output (with chaiNNer), photo dimensions are in the name

LSDIR Compact

Upscale photos to x4 their size.

4xLSDIRCompactv2 supersedes the previously released models, it combines all my progress on my compact model. Both CompactC and CompactR had received around 8 hours more training since release with batch size 10 (CompactR had only been up to 5 on release), and these two were then interpolated together. This allows v2 to handle some degradations, while preserving the details of the CompactC model. Examples: https://imgsli.com/MTY0Njgz/0/2

LSDIR Compact v2

Upscale small photos with compression to 4x their size.

Trying to extend my previous model to be able to handle compression (JPG 100-30) by manually altering the training dataset, since 4xLSDIRCompact can't handle compression. Use this instead of 4xLSDIRCompact if your photo has compression (like an image from the web).

LSDIR Compact C

Upscale compressed photos to x4 their size.

Able to handle JPG compression (30-100).

LSDIR Compact C3

Upscale good quality input photos to x4 their size

The original 4xLSDIRCompact a bit more trained, cannot handle degradation

LSDIRCompactN

Upscale small photos with compression, noise and slight blur to 4x their size.

Extending my last 4xLSDIRCompact model to Real-ESRGAN, meaning trained on synthetic data instead to handle more kinds of degradations, it should be able to handle compression, noise, and slight blur.

Here is a comparison to show that 4xLSDIRCompact cannot handle compression artifacts, and that these two models will produce better output for that specific scenario. These models are not ‘better’ than the previous one, they are just meant to handle a different use case.

LSDIR Compact R

Upscale (degraded) photos to x4 their size. 

Trained on synthetic data, meant to handle more degradations

LSDIR Compact R3

The RealESRGAN_x4plus finetuned with the big LSDIR dataset (84,991 images / 165 GB), with manually added jpg compression.

4xLSDIRplusC

The RealESRGAN_x4plus finetuned with the big LSDIR dataset (84,991 images / 165 GB), no degradation.

4xLSDIRplusN

[Link to Github Release](https://github.com/Phhofm/models/releases/4xNomos8k_atd_jpg)

Name: 4xNomos8k_atd_jpg   
License: CC BY 4.0   
Author: Philip Hofmann   
Network: [ATD](https://github.com/LabShuHangGU/Adaptive-Token-Dictionary)   
Scale: 4   
Release Date: 22.03.2024   
Purpose: 4x photo upscaler, handles jpg compression   
Iterations: 240'000   
epoch: 152   
batch_size: 6, 3   
HR_size: 128, 192   
Dataset: nomos8k   
Number of train images: 8492   
OTF Training: Yes   
Pretrained_Model_G: 003_ATD_SRx4_finetune   

Description:
4x photo upscaler which handles jpg compression. This model will preserve noise. Trained on the very recently released (~2 weeks ago) Adaptive-Token-Dictionary network.   

Training details: 
AdamW optimizer with U-Net SN discriminator and BFloat16.
Degraded with otf jpg compression down to 40, re-compression down to 40, together with resizes and the blur kernels.  
Losses: PixelLoss using CHC (Clipped Huber with Cosine Similarity Loss), PerceptualLoss using Huber, GANLoss, [LDL](https://github.com/csjliang/LDL) using Huber, YCbCr Color Loss (bt601) and Luma Loss (CIE XYZ) on [neosr](https://github.com/muslll/neosr).

7 Examples:
[Slowpics](https://slow.pics/s/uwnoI435)

 4xNomos8k_atd_jpg

I release my span otf series:
They come in three variations: weak, middle, and strong.
Mainly meant for photos (can be tried on other things of course).

(Also there is a non-otf span model I had been working on simultaneously that I will release shortly, should give better results on less degraded input in comparison to this span otf series)

Basically I trained the otf_strong for 90k iter and then medium and weak based off that, with some more training to de-learn (tone down) the (too?) strong degradations. Used discrim resets to correct occuring colorloss in all of them. gt size was for the most part 512 with batch 9 (since i hoped it would give better results) with 0.55 it/s training speed (first 40k at the beginning were gt size 256 with batch 20 with 0.58 it/s).

Nomos8k_span_otf_medium

Nomos8k_span_otf_strong

Nomos8k_span_otf_weak

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/4xDRCT-mssim-pretrains)

Since no [DRCT](https://github.com/ming053l/DRCT/tree/main) model releases exist yet, I am releasing my drct-s, drct and drct-l pretrains. 
These can be used to speed up and stabilize early training stages when training new drct models.
Trained with mssim (and augs, and color and lumaloss) on downscaled [nomosuni](https://github.com/muslll/neosr/tree/0ecbeb98dcefd97801da0ea39ca2cd03a87fce21?tab=readme-ov-file#datasets).
Training files are also provided additionally.

## 4xmssim_drct-l_pretrain

**Scale:** 4  
**Architecture:** [DRCT](https://github.com/ming053l/DRCT)  
**Architecture Option:** [DRCT-L](https://github.com/ming053l/DRCT/blob/8c13e63135f494913cd504c073e41ef52250d1d4/options/train/train_DRCT-L_SRx4_ImageNet_from_scratch.yml#L55)  

**Author:** Philip Hofmann  
**License:** CC-BY-0.4  
**Purpose:** Pretrained  
**Subject:** Anime, Photography  
**Input Type:** Images  
**Release Date:** 28.04.2024  

**Dataset:** [nomos_uni](https://github.com/muslll/neosr?tab=readme-ov-file#datasets)  
**Dataset Size:** 2989  
**OTF (on the fly augmentations):** No  
**Pretrained Model:** None (=From Scratch)  
**Iterations:** 108'000  
**Batch Size:** 2-6  
**GT Size:** 128-320  

**Description:** A pretrain to start drct-l model training, since there are no official released drct pretrains yet I trained these myself and release them here.

4xmssim_drct-l_pretrain

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/4xDRCT-mssim-pretrains)

Since no [DRCT](https://github.com/ming053l/DRCT/tree/main) model releases exist yet, I am releasing my drct-s, drct and drct-l pretrains. 
These can be used to speed up and stabilize early training stages when training new drct models.
Trained with mssim (and augs, and color and lumaloss) on downscaled [nomosuni](https://github.com/muslll/neosr/tree/0ecbeb98dcefd97801da0ea39ca2cd03a87fce21?tab=readme-ov-file#datasets).
Training files are also provided additionally.

## 4xmssim_drct_pretrain

**Scale:** 4  
**Architecture:** [DRCT](https://github.com/ming053l/DRCT)  
**Architecture Option:** [DRCT](https://github.com/ming053l/DRCT/blob/8c13e63135f494913cd504c073e41ef52250d1d4/options/train/train_DRCT_SRx4_ImageNet_from_scratch.yml#L55)  

**Author:** Philip Hofmann  
**License:** CC-BY-0.4  
**Purpose:** Pretrained  
**Subject:** Anime, Photography  
**Input Type:** Images  
**Release Date:** 28.04.2024  

**Dataset:** [nomos_uni](https://github.com/muslll/neosr?tab=readme-ov-file#datasets)  
**Dataset Size:** 2989  
**OTF (on the fly augmentations):** No  
**Pretrained Model:** None (=From Scratch)  
**Iterations:** 95'000  
**Batch Size:** 2-6  
**GT Size:** 128-320   

**Description:** A pretrain to start drct model training, since there are no official released drct pretrains yet I trained these myself and release them here.

4xmssim_drct_pretrain

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/4xDRCT-mssim-pretrains)

Since no [DRCT](https://github.com/ming053l/DRCT/tree/main) model releases exist yet, I am releasing my drct-s, drct and drct-l pretrains. 
These can be used to speed up and stabilize early training stages when training new drct models.
Trained with mssim (and augs, and color and lumaloss) on downscaled [nomosuni](https://github.com/muslll/neosr/tree/0ecbeb98dcefd97801da0ea39ca2cd03a87fce21?tab=readme-ov-file#datasets).
Training files are also provided additionally.


## 4xmssim_drct-s_pretrain.pth  

**Scale:** 4  
**Architecture:** [DRCT](https://github.com/ming053l/DRCT)  
**Architecture Option:** [DRCT-S](https://github.com/muslll/neosr/blob/0ecbeb98dcefd97801da0ea39ca2cd03a87fce21/neosr/archs/drct_arch.py#L923)  

**Author:** Philip Hofmann  
**License:** CC-BY-0.4  
**Purpose:** Pretrained  
**Subject:** Anime, Photography  
**Input Type:** Images  
**Release Date:** 28.04.2024  

**Dataset:** [nomos_uni](https://github.com/muslll/neosr?tab=readme-ov-file#datasets)  
**Dataset Size:** 2989  
**OTF (on the fly augmentations):** No  
**Pretrained Model:** None (=From Scratch)  
**Iterations:** 75'000  
**Batch Size:** 6  
**GT Size:** 128-320  

**Description:** A pretrain to start drct-s model training, since there are no official released drct pretrains yet I trained these myself and release them here.

4xmssim_drct-s_pretrain

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

[Neosr](https://github.com/muslll/neosr)'s latest update from yesterday included a [new adaptation of the multi-scale ssim loss](https://github.com/muslll/neosr/wiki/Losses#mssim_opt).     
This was an experiment to test out the difference between making a SPAN pretrain with pixel loss with L1 criteria (as often used in research) vs mssim loss as its only loss.     
Models are provided so they can be used for tests or also used as a pretrain for another SPAN model.  

---

## 4xmssim_span_pretrain

Scale: 4  
Architecture: SPAN  

Author: Philip Hofmann    
License: CC-BY-4.0  
Purpose: Pretrain  
Subject: Realistic, Anime  
Date: 10.04.2024  

Dataset: [nomos_uni](https://github.com/muslll/neosr)  
Dataset Size: 2989   
OTF (on the fly augmentations): No  
Pretrained Model: None  
Iterations: 80'000  
Batch Size: 12  
GT Size: 128  

Description: 4x SPAN pretrain trained on [neosr](https://github.com/muslll/neosr)'s [new adaptation of the multi-scale ssim loss](https://github.com/muslll/neosr/wiki/Losses#mssim_opt) from yesterdays update on downsampled nomos_uni dataset using kim's [dataset destroyer](https://github.com/Kim2091/helpful-scripts/tree/main/Dataset%20Destroyer) with down_up,linear,cubic_mitchell,lanczos,gauss,box (while down_up used the same and with range = 0.15,1.5). 
The new augmentations except CutBlur have also been used (since CutBlur is meant to be applied to real-world SR and may cause undesired effects if applied to bicubic-only). 
Config and training log provided for more details.
 
Showcase: 
[7 Slowpics Examples](https://slow.pics/c/zyilXhKU)

Architecture	LUDVAE
Scale	1x
Color Mode	RGB
License	CC-BY-4.0 Private use Commercial use Distribution Modifications Credit required State Changes No Liability & Warranty Disclaimer
Date	2024-03-25
Dataset	RealLR200
Dataset size	200
Training iterations	190000
Training batch size	16
Training HR size	64
Training OTF	No

Ludvae200

Similar Models