|
from .tensorBase import *
|
|
import torch.nn as nn
|
|
import itertools
|
|
|
|
class Density(nn.Module):
|
|
def __init__(self, params_init={}):
|
|
super().__init__()
|
|
for p in params_init:
|
|
param = nn.Parameter(torch.tensor(params_init[p]))
|
|
setattr(self, p, param)
|
|
|
|
|
|
|
|
|
|
|
|
def forward(self, sdf, beta=None):
|
|
return self.density_func(sdf, beta=beta)
|
|
|
|
|
|
class LaplaceDensity(Density):
|
|
|
|
def __init__(self, params_init={}, beta_min=0.0001):
|
|
super().__init__(params_init=params_init)
|
|
self.beta_min = torch.tensor(beta_min).cuda()
|
|
|
|
def density_func(self, sdf, beta=None):
|
|
if beta is None:
|
|
beta = self.get_beta()
|
|
|
|
alpha = 1 / beta
|
|
return alpha * (0.5 + 0.5 * sdf.sign() * torch.expm1(-sdf.abs() / beta))
|
|
|
|
def get_beta(self):
|
|
beta = self.beta.abs() + self.beta_min
|
|
return self.beta
|
|
|
|
|
|
def set_beta(self,t):
|
|
|
|
self.beta = self.beta0 * (1 + ((self.beta0 - self.beta1) / self.beta1) * (t**0.8)) ** -1
|
|
return self.beta
|
|
|
|
|
|
|
|
class TensorVMSplit_Mesh(TensorBase):
|
|
def __init__(self, aabb, gridSize, **kargs):
|
|
super(TensorVMSplit_Mesh, self).__init__(aabb, gridSize, **kargs)
|
|
|
|
hidden_dim = 64
|
|
num_layers = 5
|
|
activation = nn.ReLU
|
|
|
|
n_comp=self.density_n_comp+self.app_n_comp
|
|
|
|
self.decoder = nn.Sequential(
|
|
nn.Linear(n_comp*3, hidden_dim),
|
|
activation(),
|
|
*itertools.chain(*[[
|
|
nn.Linear(hidden_dim, hidden_dim),
|
|
activation(),
|
|
] for _ in range(num_layers - 2)]),
|
|
nn.Linear(hidden_dim, 7),
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
hidden_dim_min = 64
|
|
num_layers_min = 2
|
|
|
|
self.net_deformation = nn.Sequential(
|
|
nn.Linear(n_comp*3, hidden_dim_min),
|
|
activation(),
|
|
*itertools.chain(*[[
|
|
nn.Linear(hidden_dim_min, hidden_dim_min),
|
|
activation(),
|
|
] for _ in range(num_layers_min - 2)]),
|
|
nn.Linear(hidden_dim_min, 3),
|
|
)
|
|
|
|
self.net_weight = nn.Sequential(
|
|
nn.Linear(n_comp*3*8, hidden_dim_min),
|
|
activation(),
|
|
*itertools.chain(*[[
|
|
nn.Linear(hidden_dim_min, hidden_dim_min),
|
|
activation(),
|
|
] for _ in range(num_layers_min - 2)]),
|
|
nn.Linear(hidden_dim_min, 21),
|
|
)
|
|
|
|
|
|
for m in self.modules():
|
|
if isinstance(m, nn.Linear):
|
|
nn.init.zeros_(m.bias)
|
|
|
|
def init_render_func(self,shadingMode, pos_pe, view_pe, fea_pe, featureC):
|
|
pass
|
|
|
|
|
|
def compute_densityfeature(self, xyz_sampled):
|
|
|
|
B, N_point, _=xyz_sampled.shape
|
|
|
|
|
|
coordinate_plane = torch.stack((xyz_sampled[..., self.matMode[0]], xyz_sampled[..., self.matMode[1]], xyz_sampled[..., self.matMode[2]])).detach().view(3, B, -1, 1, 2)
|
|
coordinate_line = torch.stack((xyz_sampled[..., self.vecMode[0]], xyz_sampled[..., self.vecMode[1]], xyz_sampled[..., self.vecMode[2]]))
|
|
coordinate_line = torch.stack((torch.zeros_like(coordinate_line), coordinate_line), dim=-1).detach().view(3, B, -1, 1, 2)
|
|
|
|
plane_coef_point,line_coef_point = [],[]
|
|
for idx_plane in range(3):
|
|
|
|
density_plane=self.density_plane[:,idx_plane]
|
|
density_line=self.density_line[:,idx_plane]
|
|
|
|
plane_coef_point.append(F.grid_sample(density_plane, coordinate_plane[idx_plane],
|
|
align_corners=True).view(B, -1, N_point))
|
|
line_coef_point.append(F.grid_sample(density_line, coordinate_line[idx_plane],
|
|
align_corners=True).view(B, -1, N_point))
|
|
|
|
plane_coef_point, line_coef_point = torch.cat(plane_coef_point,dim=1), torch.cat(line_coef_point,dim=1)
|
|
plane_coef=plane_coef_point * line_coef_point
|
|
plane_coef=plane_coef.permute(0,2,1)
|
|
|
|
result = torch.matmul(plane_coef, self.d_basis_mat)
|
|
|
|
return result
|
|
|
|
|
|
def compute_appfeature(self, xyz_sampled):
|
|
|
|
B, N_point, _=xyz_sampled.shape
|
|
|
|
coordinate_plane = torch.stack((xyz_sampled[..., self.matMode[0]], xyz_sampled[..., self.matMode[1]], xyz_sampled[..., self.matMode[2]])).detach().view(3, B, -1, 1, 2)
|
|
coordinate_line = torch.stack((xyz_sampled[..., self.vecMode[0]], xyz_sampled[..., self.vecMode[1]], xyz_sampled[..., self.vecMode[2]]))
|
|
coordinate_line = torch.stack((torch.zeros_like(coordinate_line), coordinate_line), dim=-1).detach().view(3, B, -1, 1, 2)
|
|
|
|
plane_coef_point,line_coef_point = [],[]
|
|
for idx_plane in range(3):
|
|
|
|
app_plane=self.app_plane[:,idx_plane]
|
|
app_line=self.app_line[:,idx_plane]
|
|
|
|
plane_coef_point.append(F.grid_sample(app_plane, coordinate_plane[idx_plane],
|
|
align_corners=True).view(B, -1, N_point))
|
|
line_coef_point.append(F.grid_sample(app_line, coordinate_line[idx_plane],
|
|
align_corners=True).view(B, -1, N_point))
|
|
plane_coef_point, line_coef_point = torch.cat(plane_coef_point,dim=1), torch.cat(line_coef_point,dim=1)
|
|
plane_coef=plane_coef_point * line_coef_point
|
|
plane_coef=plane_coef.permute(0,2,1)
|
|
|
|
|
|
|
|
return plane_coef
|
|
|
|
|
|
def geometry_feature_decode(self, sampled_features, flexicubes_indices):
|
|
|
|
sdf = self.decoder(sampled_features)[...,-1:]
|
|
deformation = self.net_deformation(sampled_features)
|
|
|
|
grid_features = torch.index_select(input=sampled_features, index=flexicubes_indices.reshape(-1), dim=1)
|
|
grid_features = grid_features.reshape(
|
|
sampled_features.shape[0], flexicubes_indices.shape[0], flexicubes_indices.shape[1] * sampled_features.shape[-1])
|
|
weight = self.net_weight(grid_features) * 0.1
|
|
|
|
return sdf, deformation, weight
|
|
|
|
|
|
def get_geometry_prediction(self, svd_volume, sample_coordinates, flexicubes_indices):
|
|
|
|
self.svd_volume=svd_volume
|
|
self.app_plane=svd_volume['app_planes']
|
|
self.app_line=svd_volume['app_lines']
|
|
self.basis_mat=svd_volume['basis_mat']
|
|
self.density_plane=svd_volume['density_planes']
|
|
self.density_line=svd_volume['density_lines']
|
|
self.d_basis_mat=svd_volume['d_basis_mat']
|
|
|
|
self.app_plane=torch.cat([self.app_plane,self.density_plane],2)
|
|
self.app_line=torch.cat([self.app_line,self.density_line],2)
|
|
|
|
sampled_features = self.compute_appfeature(sample_coordinates)
|
|
|
|
sdf, deformation, weight = self.geometry_feature_decode(sampled_features, flexicubes_indices)
|
|
|
|
return sdf, deformation, weight
|
|
|
|
def get_texture_prediction(self,texture_pos, vsd_vome=None):\
|
|
|
|
app_features = self.compute_appfeature(texture_pos)
|
|
|
|
texture_rgb=self.decoder(app_features)[...,0:-1]
|
|
|
|
texture_rgb = torch.sigmoid(texture_rgb)*(1 + 2*0.001) - 0.001
|
|
|
|
return texture_rgb
|
|
|
|
|
|
|
|
def predict_color(self, svd_volume, xyz_sampled, white_bg=True, is_train=False, ndc_ray=False, N_samples=-1):
|
|
|
|
self.svd_volume=svd_volume
|
|
self.app_plane=svd_volume['app_planes']
|
|
self.app_line=svd_volume['app_lines']
|
|
self.basis_mat=svd_volume['basis_mat']
|
|
self.d_basis_mat=svd_volume['d_basis_mat']
|
|
self.density_plane=svd_volume['density_planes']
|
|
self.density_line=svd_volume['density_lines']
|
|
|
|
self.app_plane=torch.cat([self.app_plane,self.density_plane],2)
|
|
self.app_line=torch.cat([self.app_line,self.density_line],2)
|
|
|
|
|
|
|
|
chunk_size: int = 2**20
|
|
outs = []
|
|
for i in range(0, xyz_sampled.shape[2], chunk_size):
|
|
xyz_sampled_chunk = self.normalize_coord(xyz_sampled[:,i:i+chunk_size])
|
|
|
|
|
|
app_features = self.compute_appfeature(xyz_sampled_chunk)
|
|
|
|
chunk_out = self.decoder(app_features)[...,0:-1]
|
|
|
|
chunk_out = torch.sigmoid(chunk_out)*(1 + 2*0.001) - 0.001
|
|
|
|
rgbs = chunk_out.clamp(0,1)
|
|
outs.append(chunk_out)
|
|
|
|
rgbs=torch.cat(outs,1)
|
|
|
|
albedo=rgbs[:,:,3:6]
|
|
rgb=rgbs[:,:,0:3]
|
|
|
|
results = {
|
|
'shading':rgb,
|
|
'albedo':albedo,
|
|
'rgb':rgb*albedo,
|
|
}
|
|
return results
|
|
|
|
|
|
|
|
|
|
|
|
class TensorVMSplit_NeRF(TensorBase):
|
|
def __init__(self, aabb, gridSize, **kargs):
|
|
super(TensorVMSplit_NeRF, self).__init__(aabb, gridSize, **kargs)
|
|
|
|
hidden_dim = 64
|
|
num_layers = 4
|
|
activation = nn.ReLU
|
|
|
|
self.lap_density = LaplaceDensity(params_init={ 'beta' : 0.1})
|
|
|
|
n_comp=self.density_n_comp+self.app_n_comp
|
|
|
|
self.net_sdf = nn.Sequential(
|
|
nn.Linear(n_comp*3, hidden_dim),
|
|
activation(),
|
|
*itertools.chain(*[[
|
|
nn.Linear(hidden_dim, hidden_dim),
|
|
activation(),
|
|
] for _ in range(num_layers - 2)]),
|
|
nn.Linear(hidden_dim, 1),
|
|
)
|
|
|
|
self.decoder = nn.Sequential(
|
|
nn.Linear(n_comp*3, hidden_dim),
|
|
activation(),
|
|
*itertools.chain(*[[
|
|
nn.Linear(hidden_dim, hidden_dim),
|
|
activation(),
|
|
] for _ in range(num_layers - 2)]),
|
|
nn.Linear(hidden_dim, 6),
|
|
)
|
|
|
|
|
|
for m in self.modules():
|
|
if isinstance(m, nn.Linear):
|
|
nn.init.zeros_(m.bias)
|
|
|
|
def init_render_func(self,shadingMode, pos_pe, view_pe, fea_pe, featureC):
|
|
pass
|
|
|
|
|
|
def compute_densityfeature(self, xyz_sampled):
|
|
|
|
B, N_pixel, N_sample, _=xyz_sampled.shape
|
|
|
|
|
|
coordinate_plane = torch.stack((xyz_sampled[..., self.matMode[0]], xyz_sampled[..., self.matMode[1]], xyz_sampled[..., self.matMode[2]])).detach().view(3, B, -1, 1, 2)
|
|
coordinate_line = torch.stack((xyz_sampled[..., self.vecMode[0]], xyz_sampled[..., self.vecMode[1]], xyz_sampled[..., self.vecMode[2]]))
|
|
coordinate_line = torch.stack((torch.zeros_like(coordinate_line), coordinate_line), dim=-1).detach().view(3, B, -1, 1, 2)
|
|
|
|
plane_coef_point,line_coef_point = [],[]
|
|
for idx_plane in range(3):
|
|
|
|
density_plane=self.density_plane[:,idx_plane]
|
|
density_line=self.density_line[:,idx_plane]
|
|
|
|
plane_coef_point.append(F.grid_sample(density_plane, coordinate_plane[idx_plane],
|
|
align_corners=True).view(B, -1, N_pixel, N_sample))
|
|
line_coef_point.append(F.grid_sample(density_line, coordinate_line[idx_plane],
|
|
align_corners=True).view(B, -1, N_pixel, N_sample))
|
|
|
|
plane_coef_point, line_coef_point = torch.cat(plane_coef_point,dim=1), torch.cat(line_coef_point,dim=1)
|
|
plane_coef=plane_coef_point * line_coef_point
|
|
plane_coef=plane_coef.permute(0,2,3,1)
|
|
|
|
result = torch.matmul(plane_coef, self.d_basis_mat.unsqueeze(1))
|
|
|
|
return result
|
|
|
|
|
|
def compute_appfeature(self, xyz_sampled):
|
|
|
|
B, N_pixel, N_sample, _=xyz_sampled.shape
|
|
|
|
coordinate_plane = torch.stack((xyz_sampled[..., self.matMode[0]], xyz_sampled[..., self.matMode[1]], xyz_sampled[..., self.matMode[2]])).detach().view(3, B, -1, 1, 2)
|
|
coordinate_line = torch.stack((xyz_sampled[..., self.vecMode[0]], xyz_sampled[..., self.vecMode[1]], xyz_sampled[..., self.vecMode[2]]))
|
|
coordinate_line = torch.stack((torch.zeros_like(coordinate_line), coordinate_line), dim=-1).detach().view(3, B, -1, 1, 2)
|
|
|
|
plane_coef_point,line_coef_point = [],[]
|
|
for idx_plane in range(3):
|
|
|
|
app_plane=self.app_plane[:,idx_plane]
|
|
app_line=self.app_line[:,idx_plane]
|
|
|
|
plane_coef_point.append(F.grid_sample(app_plane, coordinate_plane[idx_plane],
|
|
align_corners=True).view(B, -1, N_pixel, N_sample))
|
|
line_coef_point.append(F.grid_sample(app_line, coordinate_line[idx_plane],
|
|
align_corners=True).view(B, -1, N_pixel, N_sample))
|
|
plane_coef_point, line_coef_point = torch.cat(plane_coef_point,dim=1), torch.cat(line_coef_point,dim=1)
|
|
plane_coef=plane_coef_point * line_coef_point
|
|
plane_coef=plane_coef.permute(0,2,3,1)
|
|
|
|
return plane_coef
|
|
|
|
def forward(self, svd_volume, rays_o, rays_d, bg_color, white_bg=True, is_train=False, ndc_ray=False, N_samples=-1):
|
|
self.svd_volume=svd_volume
|
|
self.app_plane=svd_volume['app_planes']
|
|
self.app_line=svd_volume['app_lines']
|
|
self.basis_mat=svd_volume['basis_mat']
|
|
self.d_basis_mat=svd_volume['d_basis_mat']
|
|
self.density_plane=svd_volume['density_planes']
|
|
self.density_line=svd_volume['density_lines']
|
|
|
|
self.app_plane=torch.cat([self.app_plane,self.density_plane],2)
|
|
self.app_line=torch.cat([self.app_line,self.density_line],2)
|
|
|
|
B,V,H,W,_=rays_o.shape
|
|
rays_o=rays_o.reshape(B,-1, 3)
|
|
rays_d=rays_d.reshape(B,-1, 3)
|
|
if ndc_ray:
|
|
pass
|
|
else:
|
|
|
|
xyz_sampled, z_vals, ray_valid = self.sample_ray(rays_o, rays_d, is_train=is_train,N_samples=N_samples)
|
|
dists = torch.cat((z_vals[..., 1:] - z_vals[..., :-1], torch.zeros_like(z_vals[..., :1])), dim=-1)
|
|
rays_d = rays_d.unsqueeze(-2).expand(xyz_sampled.shape)
|
|
|
|
xyz_sampled = self.normalize_coord(xyz_sampled)
|
|
|
|
mix_feature = self.compute_appfeature(xyz_sampled)
|
|
|
|
sdf = self.net_sdf(mix_feature)
|
|
|
|
sigma= self.lap_density(sdf)
|
|
sigma=sigma[...,0]
|
|
alpha, weight, bg_weight = raw2alpha(sigma, dists)
|
|
|
|
rgbs = self.decoder(mix_feature)
|
|
rgbs = torch.sigmoid(rgbs)*(1 + 2*0.001) - 0.001
|
|
|
|
|
|
acc_map = torch.sum(weight, -1)
|
|
rgb_map = torch.sum(weight[..., None] * rgbs, -2)
|
|
|
|
if white_bg or (is_train and torch.rand((1,))<0.5):
|
|
rgb_map = rgb_map + (1. - acc_map[..., None])
|
|
|
|
|
|
rgb_map = rgb_map.clamp(0,1)
|
|
rgb_map=rgb_map.view(B,V,H,W,6).permute(0,1,4,2,3)
|
|
|
|
albedo_map=rgb_map[:,:,3:6,:,:]
|
|
rgb_map=rgb_map[:,:,0:3,:,:]
|
|
|
|
with torch.no_grad():
|
|
depth_map = torch.sum(weight * z_vals, -1)
|
|
depth_map=depth_map.view(B,V,H,W,1).permute(0,1,4,2,3)
|
|
acc_map=acc_map.view(B,V,H,W,1).permute(0,1,4,2,3)
|
|
|
|
results = {
|
|
'image':rgb_map,
|
|
'albedo':albedo_map,
|
|
'alpha':acc_map,
|
|
'depth_map':depth_map
|
|
}
|
|
|
|
return results
|
|
|
|
|
|
def predict_sdf(self, svd_volume, xyz_sampled, white_bg=True, is_train=False, ndc_ray=False, N_samples=-1):
|
|
|
|
self.svd_volume=svd_volume
|
|
self.app_plane=svd_volume['app_planes']
|
|
self.app_line=svd_volume['app_lines']
|
|
self.basis_mat=svd_volume['basis_mat']
|
|
self.d_basis_mat=svd_volume['d_basis_mat']
|
|
self.density_plane=svd_volume['density_planes']
|
|
self.density_line=svd_volume['density_lines']
|
|
|
|
self.app_plane=torch.cat([self.app_plane,self.density_plane],2)
|
|
self.app_line=torch.cat([self.app_line,self.density_line],2)
|
|
|
|
chunk_size: int = 2**20
|
|
outs = []
|
|
for i in range(0, xyz_sampled.shape[1], chunk_size):
|
|
xyz_sampled_chunk = self.normalize_coord(xyz_sampled[:,i:i+chunk_size]).half()
|
|
|
|
sigma_feature = self.compute_appfeature(xyz_sampled_chunk)
|
|
chunk_out = self.net_sdf(sigma_feature)
|
|
|
|
outs.append(chunk_out)
|
|
sdf=torch.cat(outs,1)
|
|
results = {
|
|
'sigma':sdf
|
|
}
|
|
return results
|
|
|
|
|
|
def predict_color(self, svd_volume, xyz_sampled, white_bg=True, is_train=False, ndc_ray=False, N_samples=-1):
|
|
|
|
self.svd_volume=svd_volume
|
|
self.app_plane=svd_volume['app_planes']
|
|
self.app_line=svd_volume['app_lines']
|
|
self.basis_mat=svd_volume['basis_mat']
|
|
self.d_basis_mat=svd_volume['d_basis_mat']
|
|
self.density_plane=svd_volume['density_planes']
|
|
self.density_line=svd_volume['density_lines']
|
|
|
|
self.app_plane=torch.cat([self.app_plane,self.density_plane],2)
|
|
self.app_line=torch.cat([self.app_line,self.density_line],2)
|
|
|
|
xyz_sampled=xyz_sampled.unsqueeze(2)
|
|
|
|
chunk_size: int = 2**20
|
|
outs = []
|
|
for i in range(0, xyz_sampled.shape[2], chunk_size):
|
|
xyz_sampled_chunk = self.normalize_coord(xyz_sampled[:,i:i+chunk_size]).half()
|
|
|
|
|
|
app_features = self.compute_appfeature(xyz_sampled_chunk)
|
|
|
|
chunk_out = self.decoder(app_features)
|
|
|
|
chunk_out = torch.sigmoid(chunk_out)*(1 + 2*0.001) - 0.001
|
|
|
|
rgbs = chunk_out.clamp(0,1)
|
|
outs.append(chunk_out)
|
|
|
|
rgbs=torch.cat(outs,1)
|
|
rgbs=rgbs[:,:,0,:]
|
|
|
|
albedo=rgbs[:,:,3:6]
|
|
rgb=rgbs[:,:,0:3]
|
|
|
|
results = {
|
|
'shading':rgb,
|
|
'albedo':albedo,
|
|
'rgb':rgb*albedo,
|
|
}
|
|
return results
|
|
|
|
|
|
|