Files
nerfstudio/nerfstudio/models/instant_ngp.py
T
Nicholas PfaffandJonas Kulhanek 9f487a464f Implement nerfacto alpha transparency training (#2165)
* Implement nerfacto alpha transparency training

* Stop adding all ones alpha channel and clean up blend_background

* Make background_color a required argument in blend_background

* Enable alpha transparent training by default

* Improve background_color typing

* Blend uniform background color for metric computation

* Fix typing error

* Clean up typing

* Fix background_color to outputs assignment

* Model outputs returns unblended RGB (black bg) for random bg

* Update renderers

* Fix comments

* Add support for transparent BG for other methods

* Minor refac

* Fix get_background_color

* Fix typing error for blend_background

* Fix comments

* Export utils support transparency

* Pylint fix

* Fix pcd export masking

* Filter pcd view directions based on opacity

* Clean up get_rgba_image

---------

Co-authored-by: Jonas Kulhanek <jonas.kulhanek@live.com>
2023-07-18 12:54:52 +02:00

274 lines
10 KiB
Python

# Copyright 2022 the Regents of the University of California, Nerfstudio Team and contributors. All rights reserved.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Implementation of Instant NGP.
"""
from __future__ import annotations
from dataclasses import dataclass, field
from typing import Dict, List, Literal, Optional, Tuple, Type
import nerfacc
import torch
from torch.nn import Parameter
from torchmetrics.functional import structural_similarity_index_measure
from torchmetrics.image import PeakSignalNoiseRatio
from torchmetrics.image.lpip import LearnedPerceptualImagePatchSimilarity
from nerfstudio.cameras.rays import RayBundle
from nerfstudio.engine.callbacks import (
TrainingCallback,
TrainingCallbackAttributes,
TrainingCallbackLocation,
)
from nerfstudio.field_components.field_heads import FieldHeadNames
from nerfstudio.field_components.spatial_distortions import SceneContraction
from nerfstudio.fields.nerfacto_field import NerfactoField
from nerfstudio.model_components.losses import MSELoss
from nerfstudio.model_components.ray_samplers import VolumetricSampler
from nerfstudio.model_components.renderers import (
AccumulationRenderer,
DepthRenderer,
RGBRenderer,
)
from nerfstudio.models.base_model import Model, ModelConfig
from nerfstudio.utils import colormaps
@dataclass
class InstantNGPModelConfig(ModelConfig):
"""Instant NGP Model Config"""
_target: Type = field(
default_factory=lambda: NGPModel
) # We can't write `NGPModel` directly, because `NGPModel` doesn't exist yet
"""target class to instantiate"""
enable_collider: bool = False
"""Whether to create a scene collider to filter rays."""
collider_params: Optional[Dict[str, float]] = None
"""Instant NGP doesn't use a collider."""
grid_resolution: int = 128
"""Resolution of the grid used for the field."""
grid_levels: int = 4
"""Levels of the grid used for the field."""
max_res: int = 2048
"""Maximum resolution of the hashmap for the base mlp."""
log2_hashmap_size: int = 19
"""Size of the hashmap for the base mlp"""
alpha_thre: float = 0.01
"""Threshold for opacity skipping."""
cone_angle: float = 0.004
"""Should be set to 0.0 for blender scenes but 1./256 for real scenes."""
render_step_size: Optional[float] = None
"""Minimum step size for rendering."""
near_plane: float = 0.05
"""How far along ray to start sampling."""
far_plane: float = 1e3
"""How far along ray to stop sampling."""
use_appearance_embedding: bool = False
"""Whether to use an appearance embedding."""
background_color: Literal["random", "black", "white"] = "random"
"""The color that is given to untrained areas."""
disable_scene_contraction: bool = False
"""Whether to disable scene contraction or not."""
class NGPModel(Model):
"""Instant NGP model
Args:
config: instant NGP configuration to instantiate model
"""
config: InstantNGPModelConfig
field: NerfactoField
def __init__(self, config: InstantNGPModelConfig, **kwargs) -> None:
super().__init__(config=config, **kwargs)
def populate_modules(self):
"""Set the fields and modules."""
super().populate_modules()
if self.config.disable_scene_contraction:
scene_contraction = None
else:
scene_contraction = SceneContraction(order=float("inf"))
self.field = NerfactoField(
aabb=self.scene_box.aabb,
appearance_embedding_dim=0 if self.config.use_appearance_embedding else 32,
num_images=self.num_train_data,
log2_hashmap_size=self.config.log2_hashmap_size,
max_res=self.config.max_res,
spatial_distortion=scene_contraction,
)
self.scene_aabb = Parameter(self.scene_box.aabb.flatten(), requires_grad=False)
if self.config.render_step_size is None:
# auto step size: ~1000 samples in the base level grid
self.config.render_step_size = ((self.scene_aabb[3:] - self.scene_aabb[:3]) ** 2).sum().sqrt().item() / 1000
# Occupancy Grid.
self.occupancy_grid = nerfacc.OccGridEstimator(
roi_aabb=self.scene_aabb,
resolution=self.config.grid_resolution,
levels=self.config.grid_levels,
)
# Sampler
self.sampler = VolumetricSampler(
occupancy_grid=self.occupancy_grid,
density_fn=self.field.density_fn,
)
# renderers
self.renderer_rgb = RGBRenderer(background_color=self.config.background_color)
self.renderer_accumulation = AccumulationRenderer()
self.renderer_depth = DepthRenderer(method="expected")
# losses
self.rgb_loss = MSELoss()
# metrics
self.psnr = PeakSignalNoiseRatio(data_range=1.0)
self.ssim = structural_similarity_index_measure
self.lpips = LearnedPerceptualImagePatchSimilarity(normalize=True)
def get_training_callbacks(
self, training_callback_attributes: TrainingCallbackAttributes
) -> List[TrainingCallback]:
def update_occupancy_grid(step: int):
self.occupancy_grid.update_every_n_steps(
step=step,
occ_eval_fn=lambda x: self.field.density_fn(x) * self.config.render_step_size,
)
return [
TrainingCallback(
where_to_run=[TrainingCallbackLocation.BEFORE_TRAIN_ITERATION],
update_every_num_iters=1,
func=update_occupancy_grid,
),
]
def get_param_groups(self) -> Dict[str, List[Parameter]]:
param_groups = {}
if self.field is None:
raise ValueError("populate_fields() must be called before get_param_groups")
param_groups["fields"] = list(self.field.parameters())
return param_groups
def get_outputs(self, ray_bundle: RayBundle):
assert self.field is not None
num_rays = len(ray_bundle)
with torch.no_grad():
ray_samples, ray_indices = self.sampler(
ray_bundle=ray_bundle,
near_plane=self.config.near_plane,
far_plane=self.config.far_plane,
render_step_size=self.config.render_step_size,
alpha_thre=self.config.alpha_thre,
cone_angle=self.config.cone_angle,
)
field_outputs = self.field(ray_samples)
# accumulation
packed_info = nerfacc.pack_info(ray_indices, num_rays)
weights = nerfacc.render_weight_from_density(
t_starts=ray_samples.frustums.starts[..., 0],
t_ends=ray_samples.frustums.ends[..., 0],
sigmas=field_outputs[FieldHeadNames.DENSITY][..., 0],
packed_info=packed_info,
)[0]
weights = weights[..., None]
rgb = self.renderer_rgb(
rgb=field_outputs[FieldHeadNames.RGB],
weights=weights,
ray_indices=ray_indices,
num_rays=num_rays,
)
depth = self.renderer_depth(
weights=weights, ray_samples=ray_samples, ray_indices=ray_indices, num_rays=num_rays
)
accumulation = self.renderer_accumulation(weights=weights, ray_indices=ray_indices, num_rays=num_rays)
outputs = {
"rgb": rgb,
"accumulation": accumulation,
"depth": depth,
"num_samples_per_ray": packed_info[:, 1],
}
return outputs
def get_metrics_dict(self, outputs, batch):
image = batch["image"].to(self.device)
image = self.renderer_rgb.blend_background(image)
metrics_dict = {}
metrics_dict["psnr"] = self.psnr(outputs["rgb"], image)
metrics_dict["num_samples_per_batch"] = outputs["num_samples_per_ray"].sum()
return metrics_dict
def get_loss_dict(self, outputs, batch, metrics_dict=None):
image = batch["image"][..., :3].to(self.device)
pred_rgb, image = self.renderer_rgb.blend_background_for_loss_computation(
pred_image=outputs["rgb"],
pred_accumulation=outputs["accumulation"],
gt_image=image,
)
rgb_loss = self.rgb_loss(image, pred_rgb)
loss_dict = {"rgb_loss": rgb_loss}
return loss_dict
def get_image_metrics_and_images(
self, outputs: Dict[str, torch.Tensor], batch: Dict[str, torch.Tensor]
) -> Tuple[Dict[str, float], Dict[str, torch.Tensor]]:
image = batch["image"].to(self.device)
image = self.renderer_rgb.blend_background(image)
rgb = outputs["rgb"]
acc = colormaps.apply_colormap(outputs["accumulation"])
depth = colormaps.apply_depth_colormap(
outputs["depth"],
accumulation=outputs["accumulation"],
)
combined_rgb = torch.cat([image, rgb], dim=1)
combined_acc = torch.cat([acc], dim=1)
combined_depth = torch.cat([depth], dim=1)
# Switch images from [H, W, C] to [1, C, H, W] for metrics computations
image = torch.moveaxis(image, -1, 0)[None, ...]
rgb = torch.moveaxis(rgb, -1, 0)[None, ...]
psnr = self.psnr(image, rgb)
ssim = self.ssim(image, rgb)
lpips = self.lpips(image, rgb)
# all of these metrics will be logged as scalars
metrics_dict = {"psnr": float(psnr.item()), "ssim": float(ssim), "lpips": float(lpips)} # type: ignore
# TODO(ethan): return an image dictionary
images_dict = {
"img": combined_rgb,
"accumulation": combined_acc,
"depth": combined_depth,
}
return metrics_dict, images_dict