tsumugii24 / pifu-multiview

Multiview Reconstruction of PIFu

PIFu-multiview

Multiview Implementation of PIFu

detect

Instead of having to use the original Shell Command for inference sh ./scripts/test.sh.

You can easily convey the same process by running python dectect.py in a Terminal. This change may facilitates researchers of the windows operating system.

gradio_demo

You can run this script to provide a simple and intuitive web demonstration for Multiview-PIFu results using the API provided by Gradio. Then, you can directly view the 3D model on a web page without the need to additionally download MeshLab which is suggested by the official. This makes the entire process more convenient and efficient.

However, please remember to install the necessary dependencies for Gradio by using the following command line

pip install gradio

Now, try to run

python gradio_demo

and then click the link to go to the web page

PIFu reconstruction from an arbitrary number of viewpoints

The project theoretically supports reconstruction from an arbitrary number of viewpoints. One can achieve this by modifying the relevant parameters and adjusting the number of corresponding input images.

References

The origin paper: PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitization

The origin repository: shunsukesaito/PIFu: This repository contains the code for the paper "PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitization" (github.com)

pifu-multiview's Issues

How to use?

How do I get my own images?
The images I have given to the sample_images/rp_Man folder are not good enough for the model.
Is there a requirement for four images?