Pretrained: 4xPSNR

Texture retaining denoiser and sharpener for digital artwork

Sample images: https://1drv.ms/u/s!Aip-EMByJHY28xKFp93VIGydNsoN?e=vN2WYh

ArtClarity

A fast upscaler that is a replacement for standard methods of cleaning noise from the picture. Ideal for BDRemux 720p and 1080p, but only if the noise is monochrome or close to it. So far, attempts are underway to create a dataset that could eliminate color noise at the same speed.

1x BW Denoise

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/1xDeH264_realplksr)

# 1xDeH264_realplksr  
Scale: 1  
Architecture: [RealPLKSR](https://github.com/muslll/neosr/?tab=readme-ov-file#supported-archs)  
Architecture Option: [realplksr](https://github.com/muslll/neosr/blob/master/neosr/archs/realplksr_arch.py)  

Author: Philip Hofmann  
License: CC-BY-0.4  
Purpose: Restoration, H264  
Subject: Photography  
Input Type: Images  
Release Date: 08.08.2024  

Dataset: [nomosv2](https://github.com/muslll/neosr/?tab=readme-ov-file#-datasets)  
Dataset Size: 6000  
OTF (on the fly augmentations): No  
Pretrained Model: [1xDeJPG_realplksr_otf](https://github.com/Phhofm/models/releases/tag/1xDeJPG_realplksr_otf)  
Iterations: 210'000  
Batch Size: 8  
Patch Size: 64  

Description:   
A 1x de h264 model to remove h264 compression.  

Showcase:  
[Slowpics](https://slow.pics/c/71ulMz8P?canvas-mode=fit-height)  
[Imgsli](https://imgsli.com/Mjg1NTQy)

1xDeH264_realplksr

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/1xDeJPG_realplksr_otf)

1xDeJPG_realplksr_otf   
Scale: 1    
Architecture: [RealPLKSR](https://github.com/muslll/neosr/?tab=readme-ov-file#supported-archs)   
Architecture Option: [realplksr](https://github.com/muslll/neosr/blob/master/neosr/archs/realplksr_arch.py)   

Author: Philip Hofmann  
License: CC-BY-0.4  
Purpose: JPEG, Restoration  
Subject: Photography  
Input Type: Images  
Release Date: 09.07.2024  

Dataset: [nomosv2](https://github.com/muslll/neosr/?tab=readme-ov-file#-datasets)  
Dataset Size: 6000  
OTF (on the fly augmentations): Yes  
Pretrained Model: This one used the 60 model as pretrain, while I used [4xNomos2_realplksr_dysample](https://github.com/Phhofm/models/releases/tag/4xNomos2_realplksr_dysample) as a pretrain for the 60 model.  
Iterations: 196'000  
Batch Size: 24  
GT Size: 64  

Description:   
A 1x dejpg model, trained with otf down to jpg 40 in both degradation processes, see the yml config file for details.  
I also release the 1xDeJPG_realplksr_otf_60 model additionally, which had been trained with less jpg compression in comparison, down to 60 in both cases, where the hope was that it could give a bit higher quality on less compressed inputs.  
The 40 model is released as default since it will be able to handle more compressed inputs in general.  
Since these are trained with otf will slightly deblur. Examples here are compressed, and then recompressed, with jpg 40.  

Showcase:  
[Slowpics](https://slow.pics/c/c0Wft9H4)  
[Imgsli](https://imgsli.com/Mjc3NzMw)

1xDeJPG_realplksr_otf

[Link to Github Release](https://github.com/Phhofm/models/releases/tag/1xDeNoise_realplksr_otf)

# 1xDeNoise_realplksr_otf  
Scale: 1  
Architecture: [RealPLKSR](https://github.com/muslll/neosr/?tab=readme-ov-file#supported-archs)  
Architecture Option: [realplksr](https://github.com/muslll/neosr/blob/master/neosr/archs/realplksr_arch.py)  

Author: Philip Hofmann  
License: CC-BY-0.4  
Purpose: Restoration, Denoise    
Subject: Photography  
Input Type: Images  
Release Date: 08.08.2024  

Dataset: [nomosv2](https://github.com/muslll/neosr/?tab=readme-ov-file#-datasets)  
Dataset Size: 6000  
OTF (on the fly augmentations): Yes  
Pretrained Model: [1xDeJPG_realplksr_otf](https://github.com/Phhofm/models/releases/tag/1xDeJPG_realplksr_otf)  
Iterations: 200'000  
Batch Size: 8  
Patch Size: 64  

Description:   
A 1x realplksr model to denoise, trained with the realesrgan-otf pipeline, also handles a bit of jpg compression (if stronger jpg compression handling is needed, [1xDeJPG_realplksr_otf](https://github.com/Phhofm/models/releases/tag/1xDeJPG_realplksr_otf) can be used).

1xDeNoise_realplksr_otf 

Film-Degrainer_1-000

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

ISO denoise v1

ISO denoise v2

Removes banding, blocking, dithering, aliasing, noise and color tint on DDS Compressed Skin Diffuse Textures.

This should work extremely well on most modern DDS compression types. The training set was compressed with BC3/DXT5, BC3/DXT5 Fast, BC2/DXT3, BC2/DXT3 Fast, and a small number of JPEG compressed images to cover outliers.

This model is trained to remove the slight green color tint that DDS compression tends to add to skin textures, so the model output will not match the original color tone of the input image. This is the desired result though, as DDS compression shifts the colors to a sickly green tint and this model corrects that to more natural color tones.

The training set included faces, body parts, eyes, mouths and hair in a variety of skin types and tones so it should work well on most related diffuse textures.

However it's not just limited to skin, many other images and textures can be cleaned with this model. Designed to be used as a first step cleaning pass before applying additional models after. Check out the other ITF Models.

ITF SkinDiffDDS v1

[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

Takes composite/RF/VHS NES footage and attempts to restore it to RGB quality. Assumes footage has been properly deinterlaced via field duplication from 240p to 480p/720p/etc. Note that:

* All footage was captured in 240p/480p/720p NTSC.
* RGB footage was captured via an AV Famicom with the RGB Blaster via the Retrotink 2x or GBS Control.
* The model was trained exclusively on individual frames, so it can't fix things like dropouts.
* The even and odd fields of NES composite tend to be a bit...different from each other, so there will be some jitter at 60fps.
* I don't have access to an NES Toploader, so I wouldn't expect it to fix the jailbars very well.

Revision History:

* 1.5.0 (06/29/2025): Applied additional augmentations to reduce overfitting. Also added a small amount of 720p training data.
* 1.0.0 (11/03/2024): Initial release.

NES Composite to RGB

Takes composite/RF/VHS NES footage and attempts to restore it to RGB quality. Assumes footage has been properly deinterlaced via field duplication from 240p to 480p/720p/etc. Note that:

* All footage was captured in 240p/480p/720p NTSC.
* RGB footage was captured via an AV Famicom with the RGB Blaster via the Retrotink 2x or GBS Control.
* The model was trained exclusively on individual frames, so it can't fix things like dropouts.
* The even and odd fields of NES composite tend to be a bit...different from each other, so there will be some jitter at 60fps.
* I don't have access to an NES Toploader, so I wouldn't expect it to fix the jailbars very well.

Compared to the OmniSR variant of this model, this version runs more quickly (despite the larger model file size), and should do a good job on footage captured directly from the console using composite. If your footage is from a VHS tape or was captured over RF, the slower OmniSR model is probably a better choice.

Revision History:

* 1.5.0 (09/17/2025): Added "Small" version of model

NES-Composite-2-RGB-Small

NoiseToner Poisson

Attempts to remove the damages done from noise. Successor of sorts to Noisetoner_Poisson_150000_G

NoiseToner Poisson Detailed

NoiseToner Poisson Soft

NoiseToner Uniform

NoiseToner Uniform Detailed

NoiseToner Uniform Soft

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

A fast upscaler that is a replacement for standard methods of cleaning noise from the picture. Ideal for BDRemux or WEB-DL 720p or 1080p it. It gets rid of floral noise well, the smaller its grain size, the more uniform the resolution for all videos will be. If you need a more stable video, then you can also add 2x-DigitalFlim-SuperUltraCompact after this upscaler. My upscaler can slightly simplify the geometry of very small objects.

1x RGB max Denoise

Takes composite/RF/RGB/VHS Genesis footage and attempts to restore it to RGB quality. Assumes footage has been properly deinterlaced via field duplication from 240p to 480p/720p/etc. Note that:

* All footage was captured in 240p/480p/720p NTSC.
* Ground truth RGB footage was captured via a triple-bypassed Model 2 Genesis with new capacitors via the Retrotink 2x or GBS Control. Other footage was captured from various Model 1 Genesis consoles (with and without the "High Definition Graphics" mark).
* The model was trained exclusively on individual frames, so it can't fix things like dropouts.
* Should help fix color, blur, and jailbars.
* Cannot un-dither fully blended pixels.

Compared to the OmniSR variant of this model, this version runs more quickly (despite the larger model file size), and should do a good job on footage captured directly from the console using composite or RGB. If your footage is from a VHS tape or was captured over RF, the slower OmniSR model is probably a better choice.

Revision History:

1.0.0 (09/16/2025): Initial release.

Sega Genesis Cleanup-Small

Purpose: Anti-aliasing, Restoration

I was bored, so I did this. This model uses DPID as the scaling algorithm for the HRs. The original images were 8k or 12k. It's significantly sharper than Box/Area scaling, yet does a great job with aliasing. This allows for a very sharp model with minimal artifacts, even on the SPAN version.

The main model is trained on 12k images captured with Nvidia Ansel. It took about 2 days capturing manual 4k and 12k pairs for this model. The 4k captures were used as the LR, the 12k captures were resized to 4k with DPID with randomized lambda values, then trained on as HRs. 

The Alt model is trained exclusively on 8k images from my 8k dataset, resized to 4k with dpid. This provides a clearer result with less noise, but it doesn't handle long edges well at all.

Thanks to CF2lter for advice on preparing the dataset, and umzi2 for creating the [rust version of DPID](<https://github.com/umzi2/pepedpid>). 

**Showcase:** https://slow.pics/c/TCyqje9K
![Animation (2)](https://github.com/user-attachments/assets/fb9f010a-a4e1-4537-8b23-9a69974011c6)

Architecture	RealPLKSR
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	2024-08-08
Dataset	Nomosv2
Dataset size	6000
Training iterations	200000
Training batch size	8
Training HR size	256
Training OTF	Yes

1xDeNoise_realplksr_otf

1xDeNoise_realplksr_otf

Similar Models