imaginaire.model_utils.wc_vid2vid package¶
Submodules¶
imaginaire.model_utils.wc_vid2vid.render module¶
-
class
imaginaire.model_utils.wc_vid2vid.render.
SplatRenderer
[source]¶ Bases:
object
Splatting 3D point cloud into image using precomputed mapping.
-
render_image
(point_info, w, h, return_mask=False)[source]¶ Creates image of (h, w) and fills in colors. :param point_info: (i, j, 3D point idx) per row containing :type point_info: N x 3 :param mapping of image pixel to 3D point in point cloud.: :param w: Width of output image. :type w: int :param h: Height of output image. :type h: int :param return_mask: Return binary mask of coloring. :type return_mask: bool
- Returns
output (H x W x 3, uint8): Image formed with mapping and colors.
mask (H x W x 1, uint8): Binary (255 or 0) mask of colorization.
- Return type
(tuple)
-
update_point_cloud
(image, point_info)[source]¶ Updates point cloud with new points and colors. :param image: Select colors from this image to assign to :type image: H x W x 3, uint8 :param 3D points which do not have previously assigned colors.: :param point_info: (i, j, 3D point idx) per row containing :type point_info: N x 3 :param mapping of image pixel to 3D point in point cloud.:
-
-
imaginaire.model_utils.wc_vid2vid.render.
decode_unprojections
(data)[source]¶ Unpickle unprojections and make array. :param data: Each pickled string has keypoint mapping :type data: array of pickled info :param info.:
- Returns
Keys are the different resolutions, and values are padded mapping information.
- Return type
output (dict)