waifu2x

waifu2x is an image scaling and noise reduction program for anime-style art and other types of photos.[1]

waifu2x
Original author(s)nagadomi
Initial releaseOctober 11, 2015 (2015-10-11)
Stable release
v0.13.2 / November 18, 2018 (2018-11-18)
Repository
Written inLua
Operating systemLinux with CUDA support
LicenseMIT License
Websitewaifu2x.udp.jp

waifu2x was inspired by Super-Resolution Convolutional Neural Network (SRCNN).[2][3] It uses Nvidia CUDA for computing,[4] although alternative implementations that allow for OpenCL[5] and Vulkan[6] have been created.

Etymology

Waifu (from the Japanese pronunciation of "wife") is anime slang for a female character to whom one is attracted. 2x means two-times magnification.

Example
　
  • Left: 512×512px, PNG lossless, original source Wikipe-tan face.svg
  • Middle: 256×256px, JPG quality 30%, then nearest-neighbor interpolation
  • Right: 512×512px, waifu2x upscaling & noise reduction

See also

References
  1. "Amplía la resolución de tus imágenes con este portal web". TekCrispy (in Spanish). 2019-01-17. Retrieved 2019-01-21.
  2. "GitHub - nagadomi/Waifu2x: Image Super-Resolution for Anime-Style Art". GitHub. April 2020.
  3. Dong C, Loy C C, He K, et al. Image super-resolution using deep convolutional networks[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2016, 38(2): 295-307.
  4. "Even better image upscaling with Waifu2x". Fedora Magazine. 2018-10-02. Retrieved 2019-01-21.
  5. "GitHub - marcan/Cl-waifu2x: OpenCL implementation of waifu2x image upscaling". GitHub. 25 March 2020.
  6. "Waifu2x converter NCNN version, runs fast on intel / Amd / Nvidia GPU with vulkan: Nihui/Waifu2x-NCNN-vulkan". GitHub. April 2020.


This article is issued from Wikipedia. The text is licensed under Creative Commons - Attribution - Sharealike. Additional terms may apply for the media files.