mirror of
https://github.com/nerfstudio-project/nerfstudio.git
synced 2026-11-16 21:03:46 +08:00
Code cleanup (#599)
This commit is contained in:
1 parent
531361a6aa
commit
61f04896f1
4 files changed
+27
-37
No files matched your search
+18
-37
@@ -12,13 +12,11 @@
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
# pylint: skip-file
|
||||
"""
|
||||
Camera transformation helper code.
|
||||
"""
|
||||
|
||||
import math
|
||||
from turtle import width
|
||||
from typing import List, Literal, Optional, Tuple
|
||||
|
||||
import numpy as np
|
||||
@@ -118,7 +116,7 @@ def quaternion_slerp(quat0, quat1, fraction: float, spin: int = 0, shortestpath:
|
||||
raise ValueError("Input quaternions invalid.")
|
||||
if fraction == 0.0:
|
||||
return q0
|
||||
elif fraction == 1.0:
|
||||
if fraction == 1.0:
|
||||
return q1
|
||||
d = np.dot(q0, q1)
|
||||
if abs(abs(d) - 1.0) < _EPS:
|
||||
@@ -159,7 +157,7 @@ def quaternion_matrix(quaternion) -> np.ndarray:
|
||||
)
|
||||
|
||||
|
||||
def get_interpolated_poses(poseA, poseB, steps: int = 10) -> List[float]:
|
||||
def get_interpolated_poses(pose_a, pose_b, steps: int = 10) -> List[float]:
|
||||
"""Return interpolation of poses with specified number of steps.
|
||||
Args:
|
||||
poseA: first pose
|
||||
@@ -167,23 +165,23 @@ def get_interpolated_poses(poseA, poseB, steps: int = 10) -> List[float]:
|
||||
steps: number of steps the interpolated pose path should contain
|
||||
"""
|
||||
|
||||
quatA = quaternion_from_matrix(poseA[:3, :3])
|
||||
quatB = quaternion_from_matrix(poseB[:3, :3])
|
||||
quat_a = quaternion_from_matrix(pose_a[:3, :3])
|
||||
quat_b = quaternion_from_matrix(pose_b[:3, :3])
|
||||
|
||||
ts = np.linspace(0, 1, steps)
|
||||
quats = [quaternion_slerp(quatA, quatB, t) for t in ts]
|
||||
trans = [(1 - t) * poseA[:3, 3] + t * poseB[:3, 3] for t in ts]
|
||||
quats = [quaternion_slerp(quat_a, quat_b, t) for t in ts]
|
||||
trans = [(1 - t) * pose_a[:3, 3] + t * pose_b[:3, 3] for t in ts]
|
||||
|
||||
posesAB = []
|
||||
poses_ab = []
|
||||
for quat, tran in zip(quats, trans):
|
||||
pose = np.identity(4)
|
||||
pose[:3, :3] = quaternion_matrix(quat)[:3, :3]
|
||||
pose[:3, 3] = tran
|
||||
posesAB.append(pose)
|
||||
return posesAB
|
||||
poses_ab.append(pose)
|
||||
return poses_ab
|
||||
|
||||
|
||||
def get_interpolated_K(KA, KB, steps: int = 10) -> TensorType[3, 4]:
|
||||
def get_interpolated_k(k_a, k_b, steps: int = 10) -> TensorType[3, 4]:
|
||||
"""
|
||||
Returns interpolated path between two camera poses with specified number of steps.
|
||||
|
||||
@@ -195,8 +193,8 @@ def get_interpolated_K(KA, KB, steps: int = 10) -> TensorType[3, 4]:
|
||||
Ks = []
|
||||
ts = np.linspace(0, 1, steps)
|
||||
for t in ts:
|
||||
newK = KA * (1.0 - t) + KB * t
|
||||
Ks.append(newK)
|
||||
new_k = k_a * (1.0 - t) + k_b * t
|
||||
Ks.append(new_k)
|
||||
return Ks
|
||||
|
||||
|
||||
@@ -218,31 +216,14 @@ def get_interpolated_poses_many(
|
||||
traj = []
|
||||
Ks = []
|
||||
for idx in range(poses.shape[0] - 1):
|
||||
poseA = poses[idx]
|
||||
poseB = poses[idx + 1]
|
||||
posesAB = get_interpolated_poses(poseA, poseB, steps=steps_per_transition)
|
||||
traj += posesAB
|
||||
Ks += get_interpolated_K(Ks[idx], Ks[idx + 1], steps_per_transition)
|
||||
pose_a = poses[idx]
|
||||
pose_b = poses[idx + 1]
|
||||
poses_ab = get_interpolated_poses(pose_a, pose_b, steps=steps_per_transition)
|
||||
traj += poses_ab
|
||||
Ks += get_interpolated_k(Ks[idx], Ks[idx + 1], steps_per_transition)
|
||||
return torch.stack(traj, dim=0), torch.stack(Ks, dim=0)
|
||||
|
||||
|
||||
def get_swirl_poses(pose):
|
||||
"""
|
||||
Returns a swirl around the camera poses.
|
||||
# TODO(ethan): add parameters and make this work + add docstrings + return type
|
||||
"""
|
||||
N_frames = 30 * 4
|
||||
dx = np.linspace(0, 0.03, N_frames)
|
||||
dy = np.linspace(0, -0.1, N_frames)
|
||||
dz = np.linspace(0, 0.5, N_frames)
|
||||
poses = np.tile(pose, (N_frames, 1, 1))
|
||||
for i in range(N_frames):
|
||||
poses[i, 0, 3] += dx[i]
|
||||
poses[i, 1, 3] += dy[i]
|
||||
poses[i, 2, 3] += dz[i]
|
||||
return poses
|
||||
|
||||
|
||||
def normalize(x) -> TensorType[...]:
|
||||
"""Returns a normalized vector."""
|
||||
return x / torch.linalg.norm(x)
|
||||
@@ -418,7 +399,7 @@ def rotation_matrix(a: TensorType[3], b: TensorType[3]) -> TensorType[3, 3]:
|
||||
# If vectors are exactly opposite, we add a little noise to one of them
|
||||
if c < -1 + 1e-8:
|
||||
eps = (torch.rand(3) - 0.5) * 0.01
|
||||
return rotmat(a + eps, b)
|
||||
return rotation_matrix(a + eps, b)
|
||||
s = torch.linalg.norm(v)
|
||||
skew_sym_mat = torch.Tensor(
|
||||
[
|
||||
|
||||
@@ -50,6 +50,7 @@ def distortion_loss(
|
||||
starts = ray_samples.spacing_starts
|
||||
ends = ray_samples.spacing_ends
|
||||
|
||||
assert starts is not None and ends is not None, "Ray samples must have spacing starts and ends"
|
||||
midpoints = (starts + ends) / 2.0 # (..., num_samples, 1)
|
||||
|
||||
loss = (
|
||||
|
||||
@@ -502,6 +502,9 @@ class ProposalNetworkSampler(Sampler):
|
||||
self.num_nerf_samples_per_ray = num_nerf_samples_per_ray
|
||||
self.num_proposal_network_iterations = num_proposal_network_iterations
|
||||
|
||||
if self.num_proposal_network_iterations < 1:
|
||||
raise ValueError("num_proposal_network_iterations must be >= 1")
|
||||
|
||||
# samplers
|
||||
self.initial_sampler = UniformLinDispPiecewiseSampler(single_jitter=single_jitter)
|
||||
self.pdf_sampler = PDFSampler(include_original=False)
|
||||
@@ -524,6 +527,8 @@ class ProposalNetworkSampler(Sampler):
|
||||
ray_samples_list = []
|
||||
|
||||
n = self.num_proposal_network_iterations
|
||||
weights = None
|
||||
ray_samples = None
|
||||
for i_level in range(n + 1):
|
||||
is_prop = i_level < n
|
||||
num_samples = self.num_proposal_samples_per_ray[i_level] if is_prop else self.num_nerf_samples_per_ray
|
||||
@@ -533,6 +538,7 @@ class ProposalNetworkSampler(Sampler):
|
||||
else:
|
||||
# PDF sampling based on the last samples and their weights
|
||||
# Perform annealing to the weights. This will be a no-op if self._anneal is 1.0.
|
||||
assert weights is not None
|
||||
annealed_weights = torch.pow(weights, self._anneal)
|
||||
ray_samples = self.pdf_sampler(ray_bundle, ray_samples, annealed_weights, num_samples=num_samples)
|
||||
if is_prop:
|
||||
@@ -541,4 +547,5 @@ class ProposalNetworkSampler(Sampler):
|
||||
weights_list.append(weights) # (num_rays, num_samples)
|
||||
ray_samples_list.append(ray_samples)
|
||||
|
||||
assert ray_samples is not None
|
||||
return ray_samples, weights_list, ray_samples_list
|
||||
@@ -214,6 +214,7 @@ class NerfactoModel(Model):
|
||||
loss_dict["interlevel_loss"] = self.config.interlevel_loss_mult * interlevel_loss(
|
||||
outputs["weights_list"], outputs["ray_samples_list"]
|
||||
)
|
||||
assert metrics_dict is not None and "distortion" in metrics_dict
|
||||
loss_dict["distortion_loss"] = self.config.distortion_loss_mult * metrics_dict["distortion"]
|
||||
return loss_dict
|
||||
|
||||
|
||||
Reference in new issue
Block a user