We introduce a multi-level framework that infers 3D geometry of clothed humans at an unprecedentedly high 1k image resolution in a pixel-aligned manner, retaining the details in the original inputs without any post-processing. Recent advances in image-based 3D human shape estimation have been driven by the significant improvement in representation power afforded by deep neural networks. Although c
![PIFuHD](https://cdn-ak-scissors.b.st-hatena.com/image/square/80679710802991878d09666efe64e83af47feb01/height=288;version=1;width=512/https%3A%2F%2Fshunsukesaito.github.io%2FPIFuHD%2Fresources%2Fimages%2Fteaser.png)