shunsukesaito / PIFu

This repository contains the code for the paper "PIFu: Pixel-Aligned Implicit Function for High-Resolution Clothed Human Digitization"

Home Page:https://shunsukesaito.github.io/PIFu/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Performance on multiview setting

sta105 opened this issue · comments

Hi shunsuke,

I trained PIFU under multiview setting(4 random views) on twindom dataset, and tested its performance on renderpeople rigged data(I used your data generation code with the same rendering setting to render twindom and renderpeople models). However, multiview-pifu seems only work on standing poses. like:

test_eval_epoch0_000001

image

and:

image

image

if the pose changes, the results will be very bad:

image

image

Does Multiview PIFU overfit on standing poses?

hello, can you share how to train by multi-view?

Hi shunsuke,

I trained PIFU under multiview setting(4 random views) on twindom dataset, and tested its performance on renderpeople rigged data(I used your data generation code with the same rendering setting to render twindom and renderpeople models). However, multiview-pifu seems only work on standing poses. like:

test_eval_epoch0_000001

image

and:

image

image

if the pose changes, the results will be very bad:

image

image

Does Multiview PIFU overfit on standing poses?

hello, can you share how to train by multi-view?