models.py

# coding=utf-8
# Copyright 2021 The Google Research Authors.
#
# 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.

# Lint as: python3
"""Different model implementation plus a general port for all the models."""
from typing import Any, Callable
from flax import linen as nn
from jax import random
import jax.numpy as jnp
import os

from nerf import model_utils
from nerf import utils


def get_model(key, example_batch, args):
  """A helper function that wraps around a 'model zoo'."""
  model_dict = {
      "nerf": construct_nerf,
  }
  return model_dict[args.model](key, example_batch, args)


class NerfModel(nn.Module):
  """Nerf NN Model with both coarse and fine MLPs."""
  num_coarse_samples: int  # The number of samples for the coarse nerf.
  num_fine_samples: int  # The number of samples for the fine nerf.
  use_viewdirs: bool  # If True, use viewdirs as an input.
  near: float  # The distance to the near plane
  far: float  # The distance to the far plane
  noise_std: float  # The std dev of noise added to raw sigma.
  net_depth: int  # The depth of the first part of MLP.
  net_width: int  # The width of the first part of MLP.
  net_depth_condition: int  # The depth of the second part of MLP.
  net_width_condition: int  # The width of the second part of MLP.
  net_activation: Callable[Ellipsis, Any]  # MLP activation
  skip_layer: int  # How often to add skip connections.
  num_rgb_channels: int  # The number of RGB channels.
  num_sigma_channels: int  # The number of density channels.
  white_bkgd: bool  # If True, use a white background.
  min_deg_point: int  # The minimum degree of positional encoding for positions.
  max_deg_point: int  # The maximum degree of positional encoding for positions.
  deg_view: int  # The degree of positional encoding for viewdirs.
  lindisp: bool  # If True, sample linearly in disparity rather than in depth.
  rgb_activation: Callable[Ellipsis, Any]  # Output RGB activation.
  sigma_activation: Callable[Ellipsis, Any]  # Output sigma activation.
  legacy_posenc_order: bool  # Keep the same ordering as the original tf code.
  use_vax: bool  # If True, use Vax.

  @nn.compact
  def __call__(self, rng_0, rng_1, rays, voxel, len_inpc, len_inpf, randomized):
    """Nerf Model.

    Args:
      rng_0: jnp.ndarray, random number generator for coarse model sampling.
      rng_1: jnp.ndarray, random number generator for fine model sampling.
      rays: util.Rays, a namedtuple of ray origins, directions, and viewdirs.
      randomized: bool, use randomized stratified sampling.

    Returns:
      ret: list, [(rgb_coarse, disp_coarse, acc_coarse), (rgb, disp, acc)]
    """
    # Stratified sampling along rays
    key, rng_0 = random.split(rng_0)
    z_vals, samples = model_utils.sample_along_rays(
        key,
        rays.origins,
        rays.directions,
        self.num_coarse_samples,
        self.near,
        self.far,
        randomized,
        self.lindisp,
    )

    batch_size, num_samples = samples.shape[:-1]
    samples = samples.reshape(-1, 3)

    if self.use_vax:
      pts= digitize(samples, self.near, self.far, voxel.shape[0])
      mask = voxel[pts[..., 0], pts[..., 1], pts[..., 2]].squeeze()
      len_c = jnp.sum(mask)
      ind_inp, ind_bak = jnp.split(jnp.argsort(mask)[::-1], [len_inpc])

    samples_enc = model_utils.posenc(
        samples,
        self.min_deg_point,
        self.max_deg_point,
        self.legacy_posenc_order,
    )

    # Construct the "coarse" MLP.
    coarse_mlp = model_utils.MLP(
        net_depth=self.net_depth,
        net_width=self.net_width,
        net_depth_condition=self.net_depth_condition,
        net_width_condition=self.net_width_condition,
        net_activation=self.net_activation,
        skip_layer=self.skip_layer,
        num_rgb_channels=self.num_rgb_channels,
        num_sigma_channels=self.num_sigma_channels)

    # Point attribute predictions
    if self.use_viewdirs:
      viewdirs_enc_ = model_utils.posenc(
          rays.viewdirs,
          0,
          self.deg_view,
          self.legacy_posenc_order,
      )
      viewdirs_enc = jnp.tile(viewdirs_enc_[:, None, :], (1, num_samples, 1))
      viewdirs_enc =  viewdirs_enc.reshape(batch_size * num_samples, -1)
      if self.use_vax:
        raw_rgb, raw_sigma = coarse_mlp(samples_enc[ind_inp], viewdirs_enc[ind_inp])
      else:
        raw_rgb, raw_sigma = coarse_mlp(samples_enc, viewdirs_enc)
    else:
      raise NotImplementedError("deleted")

    if self.use_vax:
      ind = jnp.argsort(jnp.concatenate([ind_inp, ind_bak]))
      len_pad = batch_size * num_samples - len_inpc
      raw_rgb = jnp.vstack([raw_rgb, jnp.zeros([len_pad, 3])])[ind] * mask[:, None]
      raw_sigma = jnp.vstack([raw_sigma, jnp.zeros([len_pad, 1])])[ind] * mask[:, None]

    raw_rgb = raw_rgb.reshape(batch_size, num_samples, 3)
    raw_sigma = raw_sigma.reshape(batch_size, num_samples, 1)

    # Add noises to regularize the density predictions if needed
    key, rng_0 = random.split(rng_0)
    raw_sigma = model_utils.add_gaussian_noise(
        key,
        raw_sigma,
        self.noise_std,
        randomized,
    )
    rgb = self.rgb_activation(raw_rgb)
    sigma = self.sigma_activation(raw_sigma)

    # Volumetric rendering.
    comp_rgb, disp, acc, weights = model_utils.volumetric_rendering(
        rgb,
        sigma,
        z_vals,
        rays.directions,
        white_bkgd=self.white_bkgd,
    )
    ret = [
        (comp_rgb, disp, acc),
    ]

    # Hierarchical sampling based on coarse predictions
    if self.num_fine_samples > 0:
      z_vals_mid = .5 * (z_vals[Ellipsis, 1:] + z_vals[Ellipsis, :-1])
      key, rng_1 = random.split(rng_1)
      z_vals, samples = model_utils.sample_pdf(
          key,
          z_vals_mid,
          weights[Ellipsis, 1:-1],
          rays.origins,
          rays.directions,
          z_vals,
          self.num_fine_samples,
          randomized,
      )

      batch_size, num_samples = samples.shape[:-1]
      samples = samples.reshape(-1, 3)

      if self.use_vax:
        pts= digitize(samples, self.near, self.far, voxel.shape[0])
        mask = voxel[pts[..., 0], pts[..., 1], pts[..., 2]].squeeze()
        len_f = jnp.sum(mask)
        ind_inp, ind_bak = jnp.split(jnp.argsort(mask)[::-1], [len_inpf])

      samples_enc = model_utils.posenc(
          samples,
          self.min_deg_point,
          self.max_deg_point,
          self.legacy_posenc_order,
      )

      # Construct the "fine" MLP.
      fine_mlp = model_utils.MLP(
          net_depth=self.net_depth,
          net_width=self.net_width,
          net_depth_condition=self.net_depth_condition,
          net_width_condition=self.net_width_condition,
          net_activation=self.net_activation,
          skip_layer=self.skip_layer,
          num_rgb_channels=self.num_rgb_channels,
          num_sigma_channels=self.num_sigma_channels)

      if self.use_viewdirs:
        viewdirs_enc = jnp.tile(viewdirs_enc_[:, None, :], (1, num_samples, 1))
        viewdirs_enc =  viewdirs_enc.reshape(batch_size * num_samples, -1)
        if self.use_vax:
          raw_rgb, raw_sigma = fine_mlp(samples_enc[ind_inp], viewdirs_enc[ind_inp])
        else:
          raw_rgb, raw_sigma = fine_mlp(samples_enc, viewdirs_enc)
      else:
        raise NotImplementedError("deleted")

      if self.use_vax:
        ind = jnp.argsort(jnp.concatenate([ind_inp, ind_bak]))
        len_pad = batch_size * num_samples - len_inpf
        raw_rgb = jnp.vstack([raw_rgb, jnp.zeros([len_pad, 3])])[ind] * mask[:, None]
        raw_sigma = jnp.vstack([raw_sigma, jnp.zeros([len_pad, 1])])[ind] * mask[:, None]

      raw_rgb = raw_rgb.reshape(batch_size, num_samples, 3)
      raw_sigma = raw_sigma.reshape(batch_size, num_samples, 1)

      key, rng_1 = random.split(rng_1)
      raw_sigma = model_utils.add_gaussian_noise(
          key,
          raw_sigma,
          self.noise_std,
          randomized,
      )
      rgb = self.rgb_activation(raw_rgb)
      sigma = self.sigma_activation(raw_sigma)

      comp_rgb, disp, acc, unused_weights = model_utils.volumetric_rendering(
          rgb,
          sigma,
          z_vals,
          rays.directions,
          white_bkgd=self.white_bkgd,
      )
      ret.append((comp_rgb, disp, acc))
    else:
      len_f = 0

    if self.use_vax:
      aux = (len_c, len_f)
    else:
      aux = (0, 0)
    return ret, aux


def digitize(p, near, far, size):
  return jnp.digitize(p + (near + far) / 2., jnp.linspace(near, far, size-1))


def construct_nerf(key, example_batch, args):
  """Construct a Neural Radiance Field.

  Args:
    key: jnp.ndarray. Random number generator.
    example_batch: dict, an example of a batch of data.
    args: FLAGS class. Hyperparameters of nerf.

  Returns:
    model: nn.Model. Nerf model with parameters.
    state: flax.Module.state. Nerf model state for stateful parameters.
  """
  net_activation = getattr(nn, str(args.net_activation))
  rgb_activation = getattr(nn, str(args.rgb_activation))
  sigma_activation = getattr(nn, str(args.sigma_activation))

  # Assert that rgb_activation always produces outputs in [0, 1], and
  # sigma_activation always produce non-negative outputs.
  x = jnp.exp(jnp.linspace(-90, 90, 1024))
  x = jnp.concatenate([-x[::-1], x], 0)

  rgb = rgb_activation(x)
  if jnp.any(rgb < 0) or jnp.any(rgb > 1):
    raise NotImplementedError(
        "Choice of rgb_activation `{}` produces colors outside of [0, 1]"
        .format(args.rgb_activation))

  sigma = sigma_activation(x)
  if jnp.any(sigma < 0):
    raise NotImplementedError(
        "Choice of sigma_activation `{}` produces negative densities".format(
            args.sigma_activation))

  model = NerfModel(
      min_deg_point=args.min_deg_point,
      max_deg_point=args.max_deg_point,
      deg_view=args.deg_view,
      num_coarse_samples=args.num_coarse_samples,
      num_fine_samples=args.num_fine_samples,
      use_viewdirs=args.use_viewdirs,
      near=args.near,
      far=args.far,
      noise_std=args.noise_std,
      white_bkgd=args.white_bkgd,
      net_depth=args.net_depth,
      net_width=args.net_width,
      net_depth_condition=args.net_depth_condition,
      net_width_condition=args.net_width_condition,
      skip_layer=args.skip_layer,
      num_rgb_channels=args.num_rgb_channels,
      num_sigma_channels=args.num_sigma_channels,
      lindisp=args.lindisp,
      net_activation=net_activation,
      rgb_activation=rgb_activation,
      sigma_activation=sigma_activation,
      legacy_posenc_order=args.legacy_posenc_order,
      use_vax=args.voxel_dir != "")
  rays = example_batch["rays"]
  key1, key2, key3 = random.split(key, num=3)

  if not args.voxel_dir == "":
    voxel = jnp.load(os.path.join(args.voxel_dir, "voxel.npy"))
  else:
    voxel = None

  init_variables = model.init(
      key1,
      rng_0=key2,
      rng_1=key3,
      rays=utils.namedtuple_map(lambda x: x[0], rays),
      voxel=voxel,
      # small len_inp is recommended for model.init (small peek batch data)
      len_inpc=1 if args.num_coarse_samples > 0 else 0,
      len_inpf=1 if args.num_fine_samples > 0 else 1,
      randomized=args.randomized)

  return model, init_variables