ChaiNNer (software)
Script error: No such module "Draft topics".
Script error: No such module "AfC topic".
 | |
 | |
| Original author(s) | Joey Ballentine |
|---|---|
| Developer(s) | Community contributors Joey Ballentine Michael Schmidt theflyingzamboni |
| Initial release | January 17, 2022 |
| Repository | https://github.com/chaiNNer-org/chaiNNer |
| Written in | Python |
| Engine | |
| Operating system | Microsoft Windows, macOS, Linux |
| Platform | x86-64 and Apple M1 |
| Available in | English |
| License | GNU General Public License version 3[1] |
| Website | https://chainner.app/ |
Search ChaiNNer (software) on Amazon.
chaiNNer is an open-source node-based image manipulation software[2]. It was originally developed to use machine learning models for image restoration tasks[3] with a GUI, such as image super-resolution[3], background removal[3], image generation through Stable Diffusion[4], and image inpainting[3]. chaiNNer was developed in an electron-based node system[5], which allows flexibility to do complex image manipulation tasks. The software also includes integration with various OpenCV filters.[6] Integration with multiple inference frameworks like Pytorch[7], NCNN[8] and ONNX[9] allows chaiNNer to work on both Nvidia and AMD graphics cards. It's a cross-platform software, working on Microsoft Windows, macOS, Linux, on both x86-64 and Apple M1 architectures.[3]
Background[edit]
Before being adapted graphical solutions, machine learning models for image restoration are executed through python scripts with the use of a terminal[10][11][12]. This can be a barrier for people not used to command line tools. As a solution, multiple software adapted to use a graphical interface have been developed, such as IEU (Image Enhancing Utility)[13] and cupscale.[14] However, earlier attempts stopped being maintained by original authors.
Initial discussion about the development of a new software started on the "Enhance Everything" (originally "Game Upscale") discord server, a community dedicated to media restoration and machine learning.[15] The main developer, Joey, released the alpha version few months after that discussion[16] and since then the project received multiple contributions from the community.[17]
Features[edit]
Features include inference of multiple networks for image restoration:
| CNN-based SISR | Transformer-based SISR | Face Super-Resolution | Background Removal | Inpainting | Image Generation | OpenCV Filters |
|---|---|---|---|---|---|---|
| ESRGAN[18][19] | SwinIR[20][21] | GFPGAN[22][23] | rembg[24] (based on U2-Net[25][26]) | lama[27] | Stable Diffusion (through AUTOMATIC1111 webui[28]) | Resize, Crop, Rotation |
| SRVGGNet (Compact)[29] | Swin2SR[30] | RestoreFormer[31] | MAT[32][33] | Basic Color Correction tools | ||
| SPSR[34] | HAT[35] | CodeFormer[36][37] | Blur and Noise | |||
| Swift-SRGAN[38] | Unsharp Mask |
Additionally, chaiNNer has integration with ONNX[9], which means multiple models compatible can potentially be executed. Although not all networks can be run on AMD graphics cards due to lack of compatibility, chaiNNer has integration with NCNN[10], a framework developed by Tencent that allows conversion of models to be run on multiple kinds of hardware.
References[edit]
- ↑ "ChaiNNer". GitHub. Retrieved 18 March 2023.
- ↑ "chaiNNer". chainner.app. Retrieved 2023-03-16.
- ↑ 3.0 3.1 3.2 3.3 3.4 chaiNNer, chaiNNer, 2023-03-16, retrieved 2023-03-17
- ↑ "chaiNNer/backend/src/nodes/nodes/external_stable_diffusion at main · chaiNNer-org/chaiNNer". GitHub. Retrieved 2023-03-17.
- ↑ chaiNNer, chaiNNer, 2023-03-16, retrieved 2023-03-17
- ↑ "chaiNNer/backend/src/nodes/nodes/image_filter at main · chaiNNer-org/chaiNNer". GitHub. Retrieved 2023-03-17.
- ↑ "chaiNNer/backend/src/nodes/nodes/pytorch at main · chaiNNer-org/chaiNNer". GitHub. Retrieved 2023-03-17.
- ↑ "chaiNNer/backend/src/nodes/nodes/ncnn at main · chaiNNer-org/chaiNNer". GitHub. Retrieved 2023-03-17.
- ↑ 9.0 9.1 "chaiNNer/backend/src/nodes/nodes/onnx at main · chaiNNer-org/chaiNNer". GitHub. Retrieved 2023-03-17.
- ↑ 10.0 10.1 "Tencent NCNN". GitHub. 17 March 2023.
- ↑ ONNX Runtime Inference Examples, Microsoft, 2023-03-17, retrieved 2023-03-17
- ↑ "Saving and loading models for inference in PyTorch — PyTorch Tutorials 2.0.0+cu117 documentation". pytorch.org. Retrieved 2023-03-17.
- ↑ ptrsuder (2023-03-15), IEU - Image Enhancing Utility, retrieved 2023-03-17
- ↑ N00MKRAD (2023-03-17), Cupscale, retrieved 2023-03-17
- ↑ "Upscale Wiki". upscale.wiki. Retrieved 2023-03-17.
- ↑ "Join the chaiNNer Discord Server!". Discord. Retrieved 2023-03-17.
- ↑ "Contributors to chaiNNer-org/chaiNNer". GitHub. Retrieved 2023-03-17.
- ↑ Xintao (2023-03-17), xinntao/ESRGAN, retrieved 2023-03-17
- ↑ Wang, Xintao; Yu, Ke; Wu, Shixiang; Gu, Jinjin; Liu, Yihao; Dong, Chao; Loy, Chen Change; Qiao, Yu; Tang, Xiaoou (2018-09-17). "ESRGAN: Enhanced Super-Resolution Generative Adversarial Networks". arXiv:1809.00219 [cs.CV].
- ↑ Liang, Jingyun (2023-03-16), SwinIR: Image Restoration Using Swin Transformer, arXiv:2108.10257, retrieved 2023-03-17
- ↑ Liang, Jingyun; Cao, Jiezhang; Sun, Guolei; Zhang, Kai; Van Gool, Luc; Timofte, Radu (2021-08-23). "SwinIR: Image Restoration Using Swin Transformer". arXiv:2108.10257 [eess.IV].
- ↑ TencentARC/GFPGAN, ARC Lab, Tencent PCG, 2023-03-17, retrieved 2023-03-17
- ↑ Wang, Xintao; Li, Yu; Zhang, Honglun; Shan, Ying (2021-06-10). "Towards Real-World Blind Face Restoration with Generative Facial Prior". arXiv:2101.04061 [cs.CV].
- ↑ Gatis, Daniel (2023-03-17), Rembg, retrieved 2023-03-17
- ↑ Qin, Xuebin; Zhang, Zichen; Huang, Chenyang; Dehghan, Masood; Zaiane, Osmar R.; Jagersand, Martin (October 2020). "U$^2$-Net: Going Deeper with Nested U-Structure for Salient Object Detection". Pattern Recognition. 106: 107404. arXiv:2005.09007. Bibcode:2020PatRe.10607404Q. doi:10.1016/j.patcog.2020.107404. Unknown parameter
|s2cid=ignored (help) - ↑ Qin, Xuebin (2023-03-17), U2-Net: U Square Net, retrieved 2023-03-17
- ↑ "Resolution-robust Large Mask Inpainting with Fourier Convolutions". advimman.github.io. Retrieved 2023-03-17.
- ↑ AUTOMATIC1111 (2023-03-17), Stable Diffusion web UI, retrieved 2023-03-17
- ↑ Wang, Xintao; Xie, Liangbin; Yu, Ke; Chan, Kelvin C.K.; Loy, Chen Change; Dong, Chao (February 2022), BasicSR: Open Source Image and Video Restoration Toolbox, retrieved 2023-03-17
- ↑ Conde, Marcos V.; Choi, Ui-Jin; Burchi, Maxime; Timofte, Radu (2022-09-22). "Swin2SR: SwinV2 Transformer for Compressed Image Super-Resolution and Restoration". arXiv:2209.11345 [cs.CV].
- ↑ Wang, Zhouxia; Zhang, Jiawei; Chen, Runjian; Wang, Wenping; Luo, Ping (2022-06-25). "RestoreFormer: High-Quality Blind Face Restoration from Undegraded Key-Value Pairs". arXiv:2201.06374 [cs.CV].
- ↑ fenglinglwb (2023-03-17), MAT: Mask-Aware Transformer for Large Hole Image Inpainting (CVPR 2022 Best Paper Finalist, Oral), retrieved 2023-03-17
- ↑ Li, Wenbo; Lin, Zhe; Zhou, Kun; Qi, Lu; Wang, Yi; Jia, Jiaya (2022-06-26). "MAT: Mask-Aware Transformer for Large Hole Image Inpainting". arXiv:2203.15270 [cs.CV].
- ↑ Ma, Cheng; Rao, Yongming; Cheng, Yean; Chen, Ce; Lu, Jiwen; Zhou, Jie (2020-03-29). "Structure-Preserving Super Resolution with Gradient Guidance". arXiv:2003.13081 [eess.IV].
- ↑ Chen, Xiangyu; Wang, Xintao; Zhou, Jiantao; Dong, Chao (2022-05-16). "Activating More Pixels in Image Super-Resolution Transformer". arXiv:2205.04437 [eess.IV].
- ↑ Zhou, Shangchen (2023-03-17), sczhou/CodeFormer, retrieved 2023-03-17
- ↑ Zhou, Shangchen; Chan, Kelvin C. K.; Li, Chongyi; Loy, Chen Change (2022-10-31). "Towards Robust Blind Face Restoration with Codebook Lookup Transformer". arXiv:2206.11253 [cs.CV].
- ↑ Krishnan, Koushik Sivarama; Krishnan, Karthik Sivarama (2021-12-01). "SwiftSRGAN - Rethinking Super-Resolution for Efficient and Real-time Inference". 2021 International Conference on Intelligent Cybernetics Technology & Applications (ICICyTA). pp. 46–51. arXiv:2111.14320. doi:10.1109/ICICyTA53712.2021.9689188. ISBN 978-1-6654-1777-8. Unknown parameter
|s2cid=ignored (help) Search this book on
External Links[edit]
This article "ChaiNNer (software)" is from Wikipedia. The list of its authors can be seen in its historical and/or the page Edithistory:ChaiNNer (software). Articles copied from Draft Namespace on Wikipedia could be seen on the Draft Namespace of Wikipedia and not main one.
