DragGAN阅读笔记
一、简介问题合成符合用户需求的视觉内容通常需要对生成对象的姿势、形状、表情和布局进行灵活而精确的可控性。现有方法通过手动注释的训练数据或先前的3D模型获得了对生成对抗网络（GANs）的可控性，但往往缺乏灵活性、精确性和普适性。
本文贡献以用户交互的方式”拖动”图像中的任意点以精确到达目标点。D...
NeRF相关论文阅读
从零开始的论文阅读：
如何阅读一篇计算机领域的学术论文
一些整理得非常好的中文笔记：
NeRF阅读笔记
NeRF -> GRAF -> GIRAFFE
其他的：
由浅入深了解Diffusion Model
Image Primitives and Correspondence
Notes for An Invitation to 3-D Vision.
Correspondence of geometric featuresCorresponding points: projections of the same point in space.
Methods:...
Image Formation
Notes for An Invitation to 3-D Vision.
A geometric model of image formationGoal: project points in 3-D space onto images in a 2-D image plane.
Pr...
Representation of a Three-Dimensional Moving Scene
Notes for An Invitation to 3-D Vision.
3-D Euclidean spaceDefinition 1 Vector. Omitted.
Definition 2 Cross product. Given two vector $u,v \in \Bb...
Introduction - An Invitation to 3-D Vision
Notes for An Invitation to 3-D Vision.
Visual SystemA collection of devices that transform measurements of light into information about spatial a...
Preface
Today I set up my blog!