run_nerf_helpers.py

import paddle
import paddle.nn as nn
import paddle.nn.functional as F
import numpy as np


# Misc
img2mse = lambda x, y : paddle.mean((x - y) ** 2)
mse2psnr = lambda x : -10. * paddle.log(x) /paddle.log(paddle.to_tensor([10.]))
to8b = lambda x : (255*np.clip(x,0,1)).astype(np.uint8)

class Identity(nn.Layer):
    """A placeholder identity operator that accepts exactly one argument."""

    def __init__(self, *args, **kwargs):
        super(Identity, self).__init__()

    def forward(self, x):
        return x

# Positional encoding (section 5.1)
class Embedder:
    def __init__(self, **kwargs):
        self.kwargs = kwargs
        self.create_embedding_fn()
        
    def create_embedding_fn(self):
        embed_fns = []
        d = self.kwargs['input_dims']
        out_dim = 0
        if self.kwargs['include_input']:
            embed_fns.append(lambda x : x)
            out_dim += d
            
        max_freq = self.kwargs['max_freq_log2']
        N_freqs = self.kwargs['num_freqs']
        
        if self.kwargs['log_sampling']:
            freq_bands = 2.**paddle.linspace(0., max_freq, num=N_freqs)
        else:
            freq_bands = paddle.linspace(2.**0., 2.**max_freq,  num=N_freqs)
            
        for freq in freq_bands:
            for p_fn in self.kwargs['periodic_fns']:
                embed_fns.append(lambda x, p_fn=p_fn, freq=freq : p_fn(x * freq))
                out_dim += d
                    
        self.embed_fns = embed_fns
        self.out_dim = out_dim
        
    def embed(self, inputs):
        return paddle.concat([fn(inputs) for fn in self.embed_fns], axis=-1)


def get_embedder(multires, i=0):
    if i == -1:
        return Identity(), 3
    
    embed_kwargs = {
                'include_input' : True,
                'input_dims' : 3,
                'max_freq_log2' : multires-1,
                'num_freqs' : multires,
                'log_sampling' : True,
                'periodic_fns' : [paddle.sin, paddle.cos],
    }
    
    embedder_obj = Embedder(**embed_kwargs)
    embed = lambda x, eo=embedder_obj : eo.embed(x)
    return embed, embedder_obj.out_dim


# Model
class NeRF(nn.Layer):
    def __init__(self, D=8, W=256, input_ch=3, input_ch_views=3, output_ch=4, skips=[4], use_viewdirs=False):
        """ 
        """
        super(NeRF, self).__init__()
        self.D = D
        self.W = W
        self.input_ch = input_ch
        self.input_ch_views = input_ch_views
        self.skips = skips
        self.use_viewdirs = use_viewdirs
        
        self.pts_linears = nn.LayerList(
            [nn.Linear(input_ch, W)] + [nn.Linear(W, W) if i not in self.skips else nn.Linear(W + input_ch, W) for i in range(D-1)])
        
        ### Implementation according to the official code release (https://github.com/bmild/nerf/blob/master/run_nerf_helpers.py#L104-L105)
        self.views_linears = nn.LayerList([nn.Linear(input_ch_views + W, W//2)])

        ### Implementation according to the paper
        # self.views_linears = nn.ModuleList(
        #     [nn.Linear(input_ch_views + W, W//2)] + [nn.Linear(W//2, W//2) for i in range(D//2)])
        
        if use_viewdirs:
            self.feature_linear = nn.Linear(W, W)
            self.alpha_linear = nn.Linear(W, 1)
            self.rgb_linear = nn.Linear(W//2, 3)
        else:
            self.output_linear = nn.Linear(W, output_ch)

    def forward(self, x):
        input_pts, input_views = paddle.split(x, num_or_sections=[self.input_ch, self.input_ch_views], axis=-1)
        h = input_pts
        for i, l in enumerate(self.pts_linears):
            h = self.pts_linears[i](h)
            h = F.relu(h)
            if i in self.skips:
                h = paddle.concat([input_pts, h], axis=-1)

        if self.use_viewdirs:
            alpha = self.alpha_linear(h)
            feature = self.feature_linear(h)
            h = paddle.concat([feature, input_views], axis=-1)
        
            for i, l in enumerate(self.views_linears):
                h = self.views_linears[i](h)
                h = F.relu(h)

            rgb = self.rgb_linear(h)
            outputs = paddle.concat([rgb, alpha], -1)
        else:
            outputs = self.output_linear(h)

        return outputs    

    def load_weights_from_keras(self, weights):
        assert self.use_viewdirs, "Not implemented if use_viewdirs=False"
        
        # Load pts_linears
        for i in range(self.D):
            idx_pts_linears = 2 * i
            self.pts_linears[i].weight.data =paddle.to_tensor(np.transpose(weights[idx_pts_linears]))    
            self.pts_linears[i].bias.data = paddle.to_tensor(np.transpose(weights[idx_pts_linears+1]))
        
        # Load feature_linear
        idx_feature_linear = 2 * self.D
        self.feature_linear.weight.data = paddle.to_tensor(np.transpose(weights[idx_feature_linear]))
        self.feature_linear.bias.data = paddle.to_tensor(np.transpose(weights[idx_feature_linear+1]))

        # Load views_linears
        idx_views_linears = 2 * self.D + 2
        self.views_linears[0].weight.data = paddle.to_tensor(np.transpose(weights[idx_views_linears]))
        self.views_linears[0].bias.data =  paddle.to_tensor(np.transpose(weights[idx_views_linears+1]))

        # Load rgb_linear
        idx_rbg_linear = 2 * self.D + 4
        self.rgb_linear.weight.data =  paddle.to_tensor(np.transpose(weights[idx_rbg_linear]))
        self.rgb_linear.bias.data =  paddle.to_tensor(np.transpose(weights[idx_rbg_linear+1]))

        # Load alpha_linear
        idx_alpha_linear = 2 * self.D + 6
        self.alpha_linear.weight.data =  paddle.to_tensor(np.transpose(weights[idx_alpha_linear]))
        self.alpha_linear.bias.data =  paddle.to_tensor(np.transpose(weights[idx_alpha_linear+1]))



# Ray helpers
def get_rays(H, W, K, c2w):
    i, j = paddle.meshgrid(paddle.linspace(0, W-1, W), paddle.linspace(0, H-1, H))  # pytorch's meshgrid has indexing='ij'
    i = i.t()
    j = j.t()
    dirs = paddle.stack([(i-K[0][2])/K[0][0], -(j-K[1][2])/K[1][1], -paddle.ones_like(i)], axis=-1)
    # Rotate ray directions from camera frame to the world frame
    rays_d = paddle.sum(dirs[..., np.newaxis, :] * c2w[:3,:3], axis=-1)  # dot product, equals to: [c2w.dot(dir) for dir in dirs]
    # Translate camera frame's origin to the world frame. It is the origin of all rays.
    rays_o = c2w[:3,-1].expand(rays_d.shape)
    return rays_o, rays_d


def get_rays_np(H, W, K, c2w):
    i, j = np.meshgrid(np.arange(W, dtype=np.float32), np.arange(H, dtype=np.float32), indexing='xy')
    dirs = np.stack([(i-K[0][2])/K[0][0], -(j-K[1][2])/K[1][1], -np.ones_like(i)], -1)
    # Rotate ray directions from camera frame to the world frame
    rays_d = np.sum(dirs[..., np.newaxis, :] * c2w[:3,:3], -1)  # dot product, equals to: [c2w.dot(dir) for dir in dirs]
    # Translate camera frame's origin to the world frame. It is the origin of all rays.
    rays_o = np.broadcast_to(c2w[:3,-1], np.shape(rays_d))
    return rays_o, rays_d


def ndc_rays(H, W, focal, near, rays_o, rays_d):
    # Shift ray origins to near plane
    t = -(near + rays_o[...,2]) / rays_d[...,2]
    rays_o = rays_o + t[...,None] * rays_d
    
    # Projection
    o0 = -1./(W/(2.*focal)) * rays_o[...,0] / rays_o[...,2]
    o1 = -1./(H/(2.*focal)) * rays_o[...,1] / rays_o[...,2]
    o2 = 1. + 2. * near / rays_o[...,2]

    d0 = -1./(W/(2.*focal)) * (rays_d[...,0]/rays_d[...,2] - rays_o[...,0]/rays_o[...,2])
    d1 = -1./(H/(2.*focal)) * (rays_d[...,1]/rays_d[...,2] - rays_o[...,1]/rays_o[...,2])
    d2 = -2. * near / rays_o[...,2]
    
    rays_o = paddle.stack([o0,o1,o2], axis=-1)
    rays_d = paddle.stack([d0,d1,d2], axis=-1)
    
    return rays_o, rays_d

# 组合实现torch的gether
def paddle_gather(x, dim, index):
    index_shape = index.shape
    index_flatten = index.flatten()
    if dim < 0:
        dim = len(x.shape) + dim
    nd_index = []
    for k in range(len(x.shape)):
        if k == dim:
            nd_index.append(index_flatten)
        else:
            reshape_shape = [1] * len(x.shape)
            reshape_shape[k] = x.shape[k]
            x_arange = paddle.arange(x.shape[k], dtype=index.dtype)
            x_arange = x_arange.reshape(reshape_shape)
            dim_index = paddle.expand(x_arange, index_shape).flatten()
            nd_index.append(dim_index)
    ind2 = paddle.transpose(paddle.stack(nd_index), [1, 0]).astype("int64")
    paddle_out = paddle.gather_nd(x, ind2).reshape(index_shape)
    return paddle_out

# Hierarchical sampling (section 5.2)
def sample_pdf(bins, weights, N_samples, det=False, pytest=False):
    # Get pdf
    weights = weights + 1e-5 # prevent nans
    pdf = weights / paddle.sum(weights, axis=-1, keepdim=True)
    cdf = paddle.cumsum(pdf, axis=-1)
    cdf = paddle.concat([paddle.zeros_like(cdf[...,:1]), cdf], axis=-1)  # (batch, len(bins))

    # Take uniform samples
    if det:
        u = paddle.linspace(0., 1., num=N_samples)
        u = u.expand(list(cdf.shape[:-1]) + [N_samples])
    else:
        u = paddle.rand(list(cdf.shape[:-1]) + [N_samples])

    # Pytest, overwrite u with numpy's fixed random numbers
    if pytest:
        np.random.seed(0)
        new_shape = list(cdf.shape[:-1]) + [N_samples]
        if det:
            u = np.linspace(0., 1., N_samples)
            u = np.broadcast_to(u, new_shape)
        else:
            u = np.random.rand(*new_shape)
        u = paddle.to_tensor(u)

    # Invert CDF
    # u = u.contiguous() torch需要这步，但paddle不用，因为torch执行transpose等操作时，并不会创建新的、转置后的tensor，两个tensor内存共享
    inds = paddle.searchsorted(cdf, u, right=True)
    below = paddle.maximum(paddle.zeros_like(inds-1), inds-1)
    above = paddle.minimum((cdf.shape[-1]-1) * paddle.ones_like(inds), inds)
    inds_g = paddle.stack([below, above], axis=-1)  # (batch, N_samples, 2)

    # cdf_g = tf.gather(cdf, inds_g, axis=-1, batch_dims=len(inds_g.shape)-2)
    # bins_g = tf.gather(bins, inds_g, axis=-1, batch_dims=len(inds_g.shape)-2)
    matched_shape = [inds_g.shape[0], inds_g.shape[1], cdf.shape[-1]]
   
    cdf_g = paddle_gather(cdf.unsqueeze(1).expand(matched_shape), 2, inds_g)
    bins_g = paddle_gather(bins.unsqueeze(1).expand(matched_shape), 2, inds_g)

    denom = (cdf_g[...,1]-cdf_g[...,0])
    denom = paddle.where(denom<1e-5, paddle.ones_like(denom), denom)
    t = (u-cdf_g[...,0])/denom
    samples = bins_g[...,0] + t * (bins_g[...,1]-bins_g[...,0])

    return samples