aipixel / GPS-Gaussian

[CVPR 2024 Highlight] The official repo for “GPS-Gaussian: Generalizable Pixel-wise 3D Gaussian Splatting for Real-time Human Novel View Synthesis”
https://shunyuanzheng.github.io/GPS-Gaussian
MIT License
525 stars 34 forks source link

TypeError: GaussianRasterizationSettings.__new__() missing 1 required positional argument: 'antialiasing' #71

Open foooolish opened 2 weeks ago

foooolish commented 2 weeks ago

请问有知道怎么解决这个问题的吗?

foooolish commented 2 weeks ago

找到原因了,是因为gaussian-splatting更新了antialiasing,需要更改gaussian_renderer配置文件init.py

RAORAO611 commented 2 weeks ago

我也遇到了这个问题,请问具体你是怎么改的呢?

foooolish commented 2 weeks ago

我也遇到了这个问题,请问具体你是怎么改的呢?

这个需要在gaussian_renderer/int.py这里加上antialiasing=True就好了: raster_settings = GaussianRasterizationSettings( image_height=int(data['novel_view']['height'][idx]), image_width=int(data['novel_view']['width'][idx]), tanfovx=tanfovx, tanfovy=tanfovy, bg=bg_color, scale_modifier=1.0, viewmatrix=data['novel_view']['world_view_transform'][idx], projmatrix=data['novel_view']['full_proj_transform'][idx], sh_degree=3, campos=data['novel_view']['camera_center'][idx], prefiltered=False, debug=False, antialiasing=True

)

传输数据也需要改一下,需要多加一个_ : rasterizer = GaussianRasterizer(raster_settings=raster_settings)

# Rasterize visible Gaussians to image, obtain their radii (on screen).
rendered_image, _ ,_ = rasterizer(
RAORAO611 commented 2 weeks ago

我也遇到了这个问题,请问具体你是怎么改的呢?

这个需要在gaussian_renderer/int.py这里加上antialiasing=True就好了: raster_settings = GaussianRasterizationSettings( image_height=int(data['novel_view']['height'][idx]), image_width=int(data['novel_view']['width'][idx]), tanfovx=tanfovx, tanfovy=tanfovy, bg=bg_color, scale_modifier=1.0, viewmatrix=data['novel_view']['world_view_transform'][idx], projmatrix=data['novel_view']['full_proj_transform'][idx], sh_degree=3, campos=data['novel_view']['camera_center'][idx], prefiltered=False, debug=False, antialiasing=True

)

传输数据也需要改一下,需要多加一个_ : rasterizer = GaussianRasterizer(raster_settings=raster_settings)

# Rasterize visible Gaussians to image, obtain their radii (on screen).
rendered_image, _ ,_ = rasterizer(

好的,非常感谢!!成功解决了我的问题!!