File size: 4,040 Bytes
24f9881
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
#
# Copyright (C) 2023, Inria
# GRAPHDECO research group, https://team.inria.fr/graphdeco
# All rights reserved.
#
# This software is free for non-commercial, research and evaluation use 
# under the terms of the LICENSE.md file.
#
# For inquiries contact  george.drettakis@inria.fr
#
import json

import numpy as np
import torch

from scene.cameras import Camera, MiniCam
from utils.general import PILtoTorch
from utils.graphics import fov2focal, focal2fov, getWorld2View, getProjectionMatrix


WARNED = False


def load_json(path, H, W):
    cams = []
    with open(path) as json_file:
        contents = json.load(json_file)
        FoVx = contents["camera_angle_x"]
        FoVy = focal2fov(fov2focal(FoVx, W), H)
        zfar = 100.0
        znear = 0.01

        frames = contents["frames"]
        for idx, frame in enumerate(frames):
            # NeRF 'transform_matrix' is a camera-to-world transform
            c2w = np.array(frame["transform_matrix"])
            # change from OpenGL/Blender camera axes (Y up, Z back) to COLMAP (Y down, Z forward)
            c2w[:3, 1:3] *= -1
            if c2w.shape[0] == 3:
                one = np.zeros((1, 4))
                one[0, -1] = 1
                c2w = np.concatenate((c2w, one), axis=0)

            # get the world-to-camera transform and set R, T
            w2c = np.linalg.inv(c2w)
            R = np.transpose(w2c[:3, :3])  # R is stored transposed due to 'glm' in CUDA code
            T = w2c[:3, 3]

            w2c = torch.as_tensor(getWorld2View(R, T)).T.cuda()
            proj = getProjectionMatrix(znear, zfar, FoVx, FoVy).T.cuda()
            cams.append(MiniCam(W, H, FoVx, FoVy, znear, zfar, w2c, w2c @ proj))
    return cams


def loadCam(args, id, cam_info, resolution_scale):
    orig_w, orig_h = cam_info.image.size

    if args.resolution in [1, 2, 4, 8]:
        resolution = round(orig_w/(resolution_scale * args.resolution)), round(orig_h/(resolution_scale * args.resolution))
    else:  # should be a type that converts to float
        if args.resolution == -1:
            if orig_w > 1600:
                global WARNED
                if not WARNED:
                    print("[ INFO ] Encountered quite large input images (>1.6K pixels width), rescaling to 1.6K.\n "
                        "If this is not desired, please explicitly specify '--resolution/-r' as 1")
                    WARNED = True
                global_down = orig_w / 1600
            else:
                global_down = 1
        else:
            global_down = orig_w / args.resolution

        scale = float(global_down) * float(resolution_scale)
        resolution = (int(orig_w / scale), int(orig_h / scale))

    resized_image_rgb = PILtoTorch(cam_info.image, resolution)

    gt_image = resized_image_rgb[:3, ...]
    loaded_mask = None

    if resized_image_rgb.shape[1] == 4:
        loaded_mask = resized_image_rgb[3:4, ...]

    return Camera(colmap_id=cam_info.uid, R=cam_info.R, T=cam_info.T, 
                  FoVx=cam_info.FovX, FoVy=cam_info.FovY, 
                  image=gt_image, gt_alpha_mask=loaded_mask,
                  image_name=cam_info.image_name, uid=id, data_device=args.data_device)


def cameraList_from_camInfos(cam_infos, resolution_scale, args):
    camera_list = []

    for id, c in enumerate(cam_infos):
        camera_list.append(loadCam(args, id, c, resolution_scale))

    return camera_list


def camera_to_JSON(id, camera : Camera):
    Rt = np.zeros((4, 4))
    Rt[:3, :3] = camera.R.transpose()
    Rt[:3, 3] = camera.T
    Rt[3, 3] = 1.0

    W2C = np.linalg.inv(Rt)
    pos = W2C[:3, 3]
    rot = W2C[:3, :3]
    serializable_array_2d = [x.tolist() for x in rot]
    camera_entry = {
        'id' : id,
        'img_name' : camera.image_name,
        'width' : camera.width,
        'height' : camera.height,
        'position': pos.tolist(),
        'rotation': serializable_array_2d,
        'fy' : fov2focal(camera.FovY, camera.height),
        'fx' : fov2focal(camera.FovX, camera.width)
    }
    return camera_entry