BasicSR
Real-ESRGAN
BasicSR | Real-ESRGAN | |
---|---|---|
5 | 3 | |
6,198 | 399 | |
2.3% | 2.5% | |
0.0 | 0.0 | |
7 days ago | 19 days ago | |
Python | Python | |
Apache License 2.0 | BSD 3-clause "New" or "Revised" License |
Stars - the number of stars that a project has on GitHub. Growth - month over month growth in stars.
Activity is a relative number indicating how actively a project is being developed. Recent commits have higher weight than older ones.
For example, an activity of 9.0 indicates that a project is amongst the top 10% of the most actively developed projects that we are tracking.
BasicSR
- About Open Source Image and Video Restoration Toolbox
-
Super-Resolution Generative Adversarial Networks (SRGAN)
I think you might be interested in https://github.com/XPixelGroup/BasicSR
-
Just saw a demo of nvidias super resolution. Is the software already available to the end user and can one upsize ones old „family“ photos to something astonishingly crisp and detailed for prints?
It's not the same but GFPGan works quite well, you might want to check out BasicSR. The Remini mobile app is impressive too.
-
Anyone currently able to run Real-ESRGAN notebooks on colab currently?
!pip install git+https://github.com/xinntao/BasicSR.git
-
Training ESRGAN: Seemingly impossible
So, I was using a pretty high-end machine with an AMD RX 6900 XT GPU, ready to do some GPU accelerated computing. I had booted into Windows 10. I followed a guide for training on my own dataset. PyTorch was the name of the framework that served as the foundation for [BasicSR[(https://github.com/xinntao/BasicSR), which in turn provided the tools I needed.
Real-ESRGAN
-
Retrained Real-ESRGAN version used by ruDALL-E (a text-to-image AI) may be of interest
I noticed that the 256x256 -> 1024x1024 upscalings used by the ruDALL-E (text-to-image AI) demo site looked more detailed than other upscalers I've used. It apparently uses a retrained Real-ESRGAN. Here is a comparison I did of 2 synthetic images upscaled with this and 2 other upscalers. The GitHub repos are here and here. A web app version is here.
-
"a dog in a sweater. no watermark." (4 images) generated by CogView 2. ~256x256 cropped screenshot upscaled 4x with Real-ESRGAN from ruDALL-E demo.
I used this Colab for the upscaling. GitHub repo for the upscaler.
-
Real-ESRGAN (an upscaler) implementation used by ruDALL-E demo seems to create a lot more fine details than the other implementation of Real-ESRGAN that I used. Gallery contains upscaler comparisons for 2 input images. An implementation of SwinIR upscaler is also included.
Colab for Real-ESRGAN used by ruDALL-E demo. GitHub repo.
What are some alternatives?
GFPGAN - GFPGAN aims at developing Practical Algorithms for Real-world Face Restoration.
Real-ESRGAN-colab - A Real-ESRGAN model trained on a custom dataset
Real-ESRGAN - Real-ESRGAN aims at developing Practical Algorithms for General Image/Video Restoration.
Image-Super-Resolution-via-Iterative-Refinement - Unofficial implementation of Image Super-Resolution via Iterative Refinement by Pytorch
ESRGAN - ECCV18 Workshops - Enhanced SRGAN. Champion PIRM Challenge on Perceptual Super-Resolution. The training codes are in BasicSR.
RealSR - Toward Real-World Single Image Super-Resolution: A New Benchmark and A New Model (ICCV 2019)
dalle-flow - 🌊 A Human-in-the-Loop workflow for creating HD images from text
onnx-web - web UI for GPU-accelerated ONNX pipelines like Stable Diffusion, even on Windows and AMD
realesrgan-gui - 实用、美观的 Real-ESRGAN 图形界面,同时支持 Windows、Ubuntu 和 macOS 平台。(Cross-platform GUI for image upscaler Real-ESRGAN with additional features.)
EGVSR - Efficient & Generic Video Super-Resolution