ExploreGS/render.py at main · minsu1206/ExploreGS · GitHub

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
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
#
# 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 numpy as np
import torch
from scene import Scene
import sys
import os
from tqdm import tqdm
from os import makedirs
from gaussian_renderer import render, modified_render
import torchvision
from utils.general_utils import safe_state, multi_yaml_parsing_render
from argparse import ArgumentParser
from arguments import ModelParams, PipelineParams, get_combined_args, OptimizationParams, MaskingParams
from gaussian_renderer import GaussianModel
from omegaconf import OmegaConf
from omegaconf.dictconfig import DictConfig
import contextlib
from train_stage1 import set_render_func

def render_set(model_path, name, iteration, views, gaussians, render_func, pipeline, background):
    render_path = os.path.join(model_path, name, "ours_{}".format(iteration), "renders")
    gts_path = os.path.join(model_path, name, "ours_{}".format(iteration), "gt")

    makedirs(render_path, exist_ok=True)
    makedirs(gts_path, exist_ok=True)

    kwargs = {}
    for idx, view in enumerate(tqdm(views, desc="Rendering progress")):
        render_dict = render_func(view, gaussians, pipeline, background, **kwargs)
        rendering = render_dict["render"]
        gt = view.original_image[0:3, :, :]
        torchvision.utils.save_image(rendering, os.path.join(render_path, '{0:05d}'.format(idx) + ".png"))
        torchvision.utils.save_image(gt, os.path.join(gts_path, '{0:05d}'.format(idx) + ".png"))

def render_sets(dataset : ModelParams,
                iteration : int,
                opt : OptimizationParams,
                pipeline : PipelineParams,
                mask_params : MaskingParams,
                skip_train : bool, skip_test : bool, skip_original : bool,
                args):
    # with torch.no_grad():
    render_func = set_render_func(opt)
    no_grad = not (mask_params.use_fisher or mask_params.use_visibility_mask)
    with torch.no_grad() if no_grad else contextlib.nullcontext():
        print("[DEBUG] : dataset = ", dataset)
        print("[DEBUG] : opt = ", opt)
        print("[DEBUG] : pipeline = ", pipeline)

        gaussians = GaussianModel(dataset, opt, mask_params)
        if iteration != -1:
            print("load iteration = ", iteration)
        else:
            print("load iteration = auto search of max iteration")

        if args.swap:
            dataset.dataset_distribution = "swap"

        scene = Scene(dataset, mask_params, gaussians, load_iteration=iteration, shuffle=False)

        bg_color = [1,1,1] if dataset.white_background else [0, 0, 0]
        background = torch.tensor(bg_color, dtype=torch.float32, device="cuda")

        if mask_params.use_fisher:    # needs another rasterization pipeline
            from masking.fisher_info import render_fisher_set
            print("[INFO] : render fisher-uncertainty")
            if not skip_train:
                render_fisher_set(dataset.model_path, "train", scene.loaded_iter, scene.getTrainCameras(), scene.getTestCameras(), gaussians, modified_render, pipeline, background)
            if not skip_test:
                render_fisher_set(dataset.model_path, "test", scene.loaded_iter, scene.getTrainCameras(), scene.getTestCameras(), gaussians, modified_render, pipeline, background)

        if mask_params.use_visibility:
            from masking.visibility import render_visibility
            print("[INFO] : render visibility custom")
            if not skip_train:
                render_visibility(dataset.model_path, "train", scene.loaded_iter, scene.getTrainCameras(), scene.getTestCameras(), gaussians, render_func, pipeline, background)
            if not skip_test:
                render_visibility(dataset.model_path, "test", scene.loaded_iter, scene.getTrainCameras(), scene.getTestCameras(), gaussians, render_func, pipeline, background)

        if mask_params.use_visibility_mask:
            from masking.visibility_mask import render_visibility_mask
            print("[INFO] : render visibility mask ; ExtraNeRF style")


            # if not skip_train:
            #     render_visibility_mask(dataset.model_path, "train", scene.loaded_iter, scene.getTestCameras(), scene.getTestCameras(), gaussians, render, pipeline, background)
            if not skip_test:

                # refer utils/virtual_cam_utils.py - prepare_diffusion_cameras
                K = -1
                pair_cameras = []
                test_cameras = sorted(scene.getTestCameras(), key=lambda cam: cam.image_name)
                train_cameras = sorted(scene.getTrainCameras(), key=lambda cam: cam.image_name)
                pair_cameras.append(train_cameras)

                render_visibility_mask(dataset.model_path, "test", scene.loaded_iter, test_cameras, pair_cameras,
                                       gaussians, render_func, pipeline, background)

        if mask_params.use_scale_mask:
            from masking.scale_mask import render_scale_mask
            print("[INFO] : render scale mask ; 3DGS-Enhancer style")

            if not skip_train:
                render_scale_mask(dataset.model_path, "train", scene.loaded_iter, scene.getTrainCameras(), gaussians, render_func, pipeline, background)
            if not skip_test:
                render_scale_mask(dataset.model_path, "test", scene.loaded_iter, scene.getTestCameras(), gaussians, render_func, pipeline, background)

        if mask_params.use_viewdirection_mask:
            from masking.viewdirection_mask import render_viewdirection_mask
            print("[INFO] : render viewdirection mask ; 3DGS-Enhancer style")
            if not skip_test:
                render_viewdirection_mask(dataset.model_path, "test", scene.loaded_iter, scene.getTestCameras(), scene.getTrainCameras(), gaussians, render_func, pipeline, background)
            # if not skip_train:
            #     render_viewdirection_mask(dataset.model_path, "train", scene.loaded_iter, scene.getTrainCameras(), scene.getTrainCameras(), gaussians, render_func, pipeline, background)

        if not skip_original:
            print("[INFO] : render original RGB")
            if not skip_train:
                render_set(dataset.model_path, "train", scene.loaded_iter, scene.getTrainCameras(), gaussians, render_func, pipeline, background)
            if not skip_test:
                render_set(dataset.model_path, "test", scene.loaded_iter, scene.getTestCameras(), gaussians, render_func, pipeline, background)

if __name__ == "__main__":
    # Set up command line argument parser
    parser = ArgumentParser(description="Testing script parameters")
    model = ModelParams(parser)
    optimization = OptimizationParams(parser)
    pipeline = PipelineParams(parser)
    mask_params = MaskingParams(parser)
    parser.add_argument("--iteration", default=-1, type=int)
    parser.add_argument("--skip_train", action="store_true")
    parser.add_argument("--skip_test", action="store_true")
    parser.add_argument("--skip_original", action="store_true",
                        help="skip original rendering or not")
    parser.add_argument("--quiet", action="store_true")
    parser.add_argument("--config", "-c", type=str, default="")
    parser.add_argument("--expname", type=str, default="")
    parser.add_argument("--logname", type=str, default="") # only for logging at monitoring server
    # FIXME: Current config codes ignore cfg_args
    parser.add_argument("--override_fisher", action="store_true")
    parser.add_argument("--override_visibility", action="store_true")
    parser.add_argument("--override_visibility_mask", action="store_true")
    parser.add_argument("--override_scale_mask", action="store_true")
    parser.add_argument("--override_viewdirection_mask", action="store_true")
    parser.add_argument("--config_system", "-cs", type=str, default="")
    parser.add_argument("--config_gs", "-cg", type=str, default="")
    parser.add_argument("--config_diffusion", "-cd", type=str, default="")
    parser.add_argument("--config_uncertainty", "-cu", type=str, default="")
    parser.add_argument("--config_virtualcam", "-cv", type=str, default="")
    parser.add_argument("--start_checkpoint", type=str, default="")
    parser.add_argument("--stage1", action="store_true")
    parser.add_argument("--swap", action="store_true")
    args = parser.parse_args(sys.argv[1:])

    # args override
    args.use_fisher = True if args.override_fisher else False
    args.use_visiblity = True if args.override_visibility else False
    args.use_visibility_mask = True if args.override_visibility_mask else False
    args.use_scale_mask = True if args.override_scale_mask else False
    args.use_viewdirection_mask = True if args.override_viewdirection_mask else False
    # args.isotropic = True

    if not args.stage1:
        scene_name = os.path.basename(args.source_path)
        save_expname = multi_yaml_parsing_render(args)
        save_expname = f"stage2_{args.expname}/{save_expname}" # HARDCODED
        args.model_path = os.path.join("output", save_expname, scene_name)

    if args.swap:
        args.model_path = args.model_path + "_swap"

    print("Rendering " + args.model_path)
    # Initialize system state (RNG)
    safe_state(args.quiet)

    render_sets(model.extract(args), args.iteration, optimization.extract(args), pipeline.extract(args), mask_params.extract(args),
                args.skip_train, args.skip_test, args.skip_original, args)