补帧 插帧 提高帧数 黑白相片上色 慢动作视频 照片优化 提高清晰度 模糊变清晰 人像美颜

超分辨率 super resolution realcugan bilibili official real cugan

avisynth的替代品：vapoursynth（in Python） FrameServer

vapoursynth is somehow installed on python 3.10(brew). do not know what depends on that.

mvtools for vapoursynth motion compensation

效果好 速度慢 dain

rife 速度快 画质会变差 another repo link with more stars

vapoursynth rife filter usage just a gist

vsrife using cuda

rife plugin for vapoursynth using vulkan

nvidia super slomo 比较吃显存 需要NVIDIA SDK 速度快

sepconv 看起来比较模糊 但是还是比直接overlay要好

pytorch sepconv slomo

memc-net比较清晰

FFmpeg自带插帧的filter：

1
2

ffmpeg -i input.60fps.hevc -filter "minterpolate='fps=120'" output.120fps.hevc

ai黑白上色可以把原视频洗稿：image colorization

coloring grayscale images

Coloring black and white images with deep learning

memc (motion estimation/motion compensation)

在CSDN上看到的算法名称和内容

内置超分辨率算法:

Waifu2x / SRMD / RealSR / Real-ESRGAN/ Real-CUGAN / Anime4K / ACNet

内置超分辨率引擎:

Waifu2x-caffe / Waifu2x-converter / Waifu2x-ncnn-vulkan

SRMD-ncnn-vulkan / RealSR-ncnn-vulkan / Anime4KCPP / SRMD-CUDA

RealESRGAN-NCNN-Vulkan / Real-CUGAN-ncnn-vulkan

内置插帧算法:

RIFE / CAIN / DAIN

内置插帧引擎:

rife-ncnn-vulkan / cain-ncnn-vulkan / dain-ncnn-vulkan

Image Restoration Upscaling Inpainting 图像修复 超分辨率

sota image inpainting: lama-cleaner still needs manual labeling on inpainting area

https://github.com/DmitryUlyanov/deep-image-prior

nas image prior

https://arxiv.org/abs/2008.11713

mmediting: OpenMMLab Image and Video Restoration, Editing and Generation Toolbox