Sawyer

Paper name/title: GPS-Gaussian: Generalizable Pixel-wise 3D Gaussian Splatting for Real-time Human Novel View Synthesis Paper link: https://arxiv.org/abs/2312.02155 Code link: https://github.com/ShunyuanZheng/GPS-Gaussian Project link: https://shunyuanzheng.github.io/GPS-Gaussian

> Hi @Miaosheng1, in our code, we assume the image shapes are even (see [here](https://github.com/donydchen/mvsplat/blob/378ff818c0151719bbc052ac2797a2c769766320/src/dataset/shims/patch_shim.py#L8)) and mainly perform the center crop operation, which is symmetric and should not change the...

> Hi @ShunyuanZheng, you are correct. The original principal point of DTU is _slightly_ off-center; we force it to be in the center for simplicity. > > Take the scene...

I encountered a similar problem. I have tried a series of images in the shape of 518x518, and the result of VGGT is perfect. However, when I downsample the images...

> Hi [@ShunyuanZheng](https://github.com/ShunyuanZheng) can you try first resizing to 336x336 and then black padding it to 336x518? Under this implementation, the result looks reasonable without distortion though its performance is...