dist_train.py

# SPDX-FileCopyrightText: Copyright (c) 2021-2022 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
# SPDX-License-Identifier: LicenseRef-NvidiaProprietary
#
# NVIDIA CORPORATION, its affiliates and licensors retain all intellectual
# property and proprietary rights in and to this material, related
# documentation and any modifications thereto. Any use, reproduction,
# disclosure or distribution of this material and related documentation
# without an express license agreement from NVIDIA CORPORATION or
# its affiliates is strictly prohibited.

"""Train a GAN using the techniques described in the paper
"Efficient Geometry-aware 3D Generative Adversarial Networks."
Code adapted from
"Alias-Free Generative Adversarial Networks"."""

import os
import click
import datetime
import re
import json
import tempfile
import torch
import torch.distributed as tdist
import torch.multiprocessing as tmp

import dnnlib
from training import training_loop
from metrics import metric_main
from torch_utils import training_stats
from torch_utils import custom_ops

#----------------------------------------------------------------------------

# def subprocess_fn(rank, c, temp_dir):
#     dnnlib.util.Logger(file_name=os.path.join(c.run_dir, 'log.txt'), file_mode='a', should_flush=True)

#     # Init torch.distributed.
#     if c.num_gpus > 1:
#         init_file = os.path.abspath(os.path.join(temp_dir, '.torch_distributed_init'))
#         if os.name == 'nt':
#             init_method = 'file:///' + init_file.replace('\\', '/')
#             torch.distributed.init_process_group(backend='gloo', init_method=init_method, rank=rank, world_size=c.num_gpus)
#         else:
#             init_method = f'file://{init_file}'
#             torch.distributed.init_process_group(backend='nccl', init_method=init_method, rank=rank, world_size=c.num_gpus)

#     # Init torch_utils.
#     sync_device = torch.device('cuda', rank) if c.num_gpus > 1 else None
#     training_stats.init_multiprocessing(rank=rank, sync_device=sync_device)
#     if rank != 0:
#         custom_ops.verbosity = 'none'

#     # Execute training loop.
#     training_loop.training_loop(rank=rank, **c)
def init_dist():
    """Initialize the distribution settings"""
    if tmp.get_start_method(allow_none=True) is None:
        tmp.set_start_method('spawn')
    rank = int(os.getenv('RANK', 0))
    torch.cuda.set_device(int(os.getenv('LOCAL_RANK', 0)))
    tdist.init_process_group(backend='nccl', init_method='env://',
                            timeout=datetime.timedelta(seconds=1800))
    training_stats.init_multiprocessing(rank, torch.cuda.current_device())

#----------------------------------------------------------------------------

def launch_training(c, desc, outdir, dry_run):
    rank = int(os.getenv('RANK', 0))

    dnnlib.util.Logger(should_flush=True)

    # Pick output directory.
    prev_run_dirs = []
    if os.path.isdir(outdir):
        prev_run_dirs = [x for x in os.listdir(outdir) if os.path.isdir(os.path.join(outdir, x))]
    prev_run_ids = [re.match(r'^\d+', x) for x in prev_run_dirs]
    prev_run_ids = [int(x.group()) for x in prev_run_ids if x is not None]
    cur_run_id = max(prev_run_ids, default=-1) + 1
    c.run_dir = os.path.join(outdir, f'{cur_run_id:05d}-{desc}')
    assert not os.path.exists(c.run_dir)


    # Print options.
    if rank == 0:
        print()
        print('Training options:')
        print(json.dumps(c, indent=2))
        print()
        print(f'Output directory:    {c.run_dir}')
        print(f'Number of GPUs:      {c.num_gpus}')
        print(f'Batch size:          {c.batch_size} images')
        print(f'Training duration:   {c.total_kimg} kimg')
        print(f'Dataset path:        {c.training_set_kwargs.path}')
        print(f'Dataset size:        {c.training_set_kwargs.max_size} images')
        print(f'Dataset resolution:  {c.training_set_kwargs.resolution}')
        print(f'Dataset labels:      {c.training_set_kwargs.use_labels}')
        print(f'Dataset x-flips:     {c.training_set_kwargs.xflip}')
        print()

    # Dry run?
    if dry_run and rank == 0:
        print('Dry run; exiting.')
        return

    # Create output directory.
    if rank == 0:
        print('Creating output directory...')
        os.makedirs(c.run_dir)
        with open(os.path.join(c.run_dir, 'training_options.json'), 'wt') as f:
            json.dump(c, f, indent=2)
        dnnlib.util.Logger(file_name=os.path.join(c.run_dir, 'log.txt'), file_mode='a', should_flush=True)
    tdist.barrier()

    # Launch processes.
    if rank == 0:
        print('Launching processes...')
    training_loop.training_loop(rank=rank, **c)
#----------------------------------------------------------------------------

def init_dataset_kwargs(data, rebalance):
    try:
        # dataset_kwargs = dnnlib.EasyDict(class_name='training.dataset.ImageFolderDataset', path=data, use_labels=True, max_size=None, xflip=False)
        if rebalance:
            dataset_kwargs = dnnlib.EasyDict(class_name='training.dataset.RebalancedSMPLLabeledDataset', path=data, use_labels=True, max_size=None, xflip=False)
        else:
            dataset_kwargs = dnnlib.EasyDict(class_name='training.dataset.SMPLLabeledDataset', path=data, use_labels=True, max_size=None, xflip=False)
        dataset_obj = dnnlib.util.construct_class_by_name(**dataset_kwargs) # Subclass of training.dataset.Dataset.
        dataset_kwargs.resolution = dataset_obj.resolution # Be explicit about resolution.
        dataset_kwargs.use_labels = dataset_obj.has_labels # Be explicit about labels.
        dataset_kwargs.max_size = len(dataset_obj) # Be explicit about dataset size.
        return dataset_kwargs, dataset_obj.name
    except IOError as err:
        raise click.ClickException(f'--data: {err}')

#----------------------------------------------------------------------------

def parse_comma_separated_list(s):
    if isinstance(s, list):
        return s
    if s is None or s.lower() == 'none' or s == '':
        return []
    return s.split(',')

#----------------------------------------------------------------------------

@click.command()

# Required.
@click.option('--outdir',       help='Where to save the results', metavar='DIR',                required=True)
@click.option('--cfg',          help='Base configuration',                                      type=str, required=True)
@click.option('--data',         help='Training data', metavar='[ZIP|DIR]',                      type=str, required=True)
@click.option('--gpus',         help='Number of GPUs to use', metavar='INT',                    type=click.IntRange(min=1), required=True)
@click.option('--batch',        help='Total batch size', metavar='INT',                         type=click.IntRange(min=1), required=True)
@click.option('--gamma',        help='R1 regularization weight', metavar='FLOAT',               type=click.FloatRange(min=0), required=True)

# Optional features.
@click.option('--cond',         help='Train conditional model', metavar='BOOL',                 type=bool, default=True, show_default=True)
@click.option('--mirror',       help='Enable dataset x-flips', metavar='BOOL',                  type=bool, default=False, show_default=True)
@click.option('--aug',          help='Augmentation mode',                                       type=click.Choice(['noaug', 'ada', 'fixed']), default='noaug', show_default=True)
@click.option('--resume',       help='Resume from given network pickle', metavar='[PATH|URL]',  type=str)
@click.option('--freezed',      help='Freeze first layers of D', metavar='INT',                 type=click.IntRange(min=0), default=0, show_default=True)

# Misc hyperparameters.
@click.option('--p',            help='Probability for --aug=fixed', metavar='FLOAT',            type=click.FloatRange(min=0, max=1), default=0.2, show_default=True)
@click.option('--target',       help='Target value for --aug=ada', metavar='FLOAT',             type=click.FloatRange(min=0, max=1), default=0.6, show_default=True)
@click.option('--batch-gpu',    help='Limit batch size per GPU', metavar='INT',                 type=click.IntRange(min=1))
@click.option('--cbase',        help='Capacity multiplier', metavar='INT',                      type=click.IntRange(min=1), default=32768, show_default=True)
@click.option('--cmax',         help='Max. feature maps', metavar='INT',                        type=click.IntRange(min=1), default=512, show_default=True)
@click.option('--glr',          help='G learning rate  [default: varies]', metavar='FLOAT',     type=click.FloatRange(min=0))
@click.option('--dlr',          help='D learning rate', metavar='FLOAT',                        type=click.FloatRange(min=0), default=0.002, show_default=True)
@click.option('--map-depth',    help='Mapping network depth  [default: varies]', metavar='INT', type=click.IntRange(min=1), default=2, show_default=True)
@click.option('--mbstd-group',  help='Minibatch std group size', metavar='INT',                 type=click.IntRange(min=1), default=4, show_default=True)

# Misc settings.
@click.option('--desc',         help='String to include in result dir name', metavar='STR',     type=str)
@click.option('--metrics',      help='Quality metrics', metavar='[NAME|A,B,C|none]',            type=parse_comma_separated_list, default='fid50k_full', show_default=True)
@click.option('--kimg',         help='Total training duration', metavar='KIMG',                 type=click.IntRange(min=1), default=25000, show_default=True)
@click.option('--tick',         help='How often to print progress', metavar='KIMG',             type=click.IntRange(min=1), default=4, show_default=True)
@click.option('--snap',         help='How often to save snapshots', metavar='TICKS',            type=click.IntRange(min=1), default=50, show_default=True)
@click.option('--seed',         help='Random seed', metavar='INT',                              type=click.IntRange(min=0), default=0, show_default=True)
# @click.option('--fp32',         help='Disable mixed-precision', metavar='BOOL',                 type=bool, default=False, show_default=True)
@click.option('--nobench',      help='Disable cuDNN benchmarking', metavar='BOOL',              type=bool, default=False, show_default=True)
@click.option('--workers',      help='DataLoader worker processes', metavar='INT',              type=click.IntRange(min=1), default=3, show_default=True)
@click.option('-n','--dry-run', help='Print training options and exit',                         is_flag=True)

# @click.option('--sr_module',    help='Superresolution module', metavar='STR',  type=str, required=True)
@click.option('--neural_rendering_resolution_initial', help='Resolution to render at', metavar='INT',  type=click.IntRange(min=1), default=64, required=False)
@click.option('--neural_rendering_resolution_final', help='Final resolution to render at, if blending', metavar='INT',  type=click.IntRange(min=1), required=False, default=None)
@click.option('--neural_rendering_resolution_fade_kimg', help='Kimg to blend resolution over', metavar='INT',  type=click.IntRange(min=0), required=False, default=1000, show_default=True)

@click.option('--resume_kimg', help='Resume over how many', metavar='INT',  type=click.IntRange(min=0), required=False, default=0)
@click.option('--blur_fade_kimg', help='Blur over how many', metavar='INT',  type=click.IntRange(min=1), required=False, default=200)
@click.option('--gen_pose_cond', help='If true, enable generator pose conditioning.', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--c-scale', help='Scale factor for generator pose conditioning.', metavar='FLOAT',  type=click.FloatRange(min=0), required=False, default=1)
@click.option('--c-noise', help='Add noise for generator pose conditioning.', metavar='FLOAT',  type=click.FloatRange(min=0), required=False, default=0)
@click.option('--gpc_reg_prob', help='Strength of swapping regularization. None means no generator pose conditioning, i.e. condition with zeros.', metavar='FLOAT',  type=click.FloatRange(min=0), required=False, default=0.5)
@click.option('--gpc_reg_fade_kimg', help='Length of swapping prob fade', metavar='INT',  type=click.IntRange(min=0), required=False, default=1000)
@click.option('--disc_c_noise', help='Strength of discriminator pose conditioning regularization, in standard deviations.', metavar='FLOAT',  type=click.FloatRange(min=0), required=False, default=0)
@click.option('--sr_noise_mode', help='Type of noise for superresolution', metavar='STR',  type=click.Choice(['random', 'none']), required=False, default='none')
@click.option('--enable_blur', help='Enable to blur even on resume', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--sr_num_fp16_res',    help='Number of fp16 layers in superresolution', metavar='INT', type=click.IntRange(min=0), default=4, required=False, show_default=True)
@click.option('--g_num_fp16_res',    help='Number of fp16 layers in generator', metavar='INT', type=click.IntRange(min=0), default=0, required=False, show_default=True)
@click.option('--d_num_fp16_res',    help='Number of fp16 layers in discriminator', metavar='INT', type=click.IntRange(min=0), default=4, required=False, show_default=True)
@click.option('--sr_first_cutoff',    help='First cutoff for AF superresolution', metavar='INT', type=click.IntRange(min=2), default=2, required=False, show_default=True)
@click.option('--sr_first_stopband',    help='First cutoff for AF superresolution', metavar='FLOAT', type=click.FloatRange(min=2), default=2**2.1, required=False, show_default=True)
@click.option('--style_mixing_prob',    help='Style-mixing regularization probability for training.', metavar='FLOAT', type=click.FloatRange(min=0, max=1), default=0, required=False, show_default=True)
@click.option('--sr-module',    help='Superresolution module override', metavar='STR',  type=str, required=False, default=None)
@click.option('--density_reg',    help='Density regularization strength.', metavar='FLOAT', type=click.FloatRange(min=0), default=0.25, required=False, show_default=True)
@click.option('--density_reg_every',    help='lazy density reg', metavar='int', type=click.FloatRange(min=1), default=4, required=False, show_default=True)
@click.option('--density_reg_p_dist',    help='density regularization strength.', metavar='FLOAT', type=click.FloatRange(min=0), default=0.004, required=False, show_default=True)
@click.option('--reg_type', help='Type of regularization', metavar='STR',  type=click.Choice(['l1', 'l1-alt', 'monotonic-detach', 'monotonic-fixed', 'total-variation']), required=False, default='l1')
@click.option('--decoder_lr_mul',    help='decoder learning rate multiplier.', metavar='FLOAT', type=click.FloatRange(min=0), default=1, required=False, show_default=True)
# avatargen related args
@click.option('--depth_resolution', help='depth resolution for coarse sampling', metavar='INT',  type=click.IntRange(min=2), required=False, default=24)
@click.option('--depth_resolution_importance', help='depth resolution for fine sampling', metavar='INT',  type=click.IntRange(min=0), required=False, default=0)
@click.option('--calc_eikonal_coarse', help='If true, enable eikonal loss for coarse stage.', metavar='BOOL',  type=bool, required=False, default=True)
@click.option('--rebalance', help='If true, enable dataset balancing for large pose.', metavar='BOOL',  type=bool, required=False, default=True)
@click.option('--face_disc_weight',          help='face disc loss weight  [default: varies]', metavar='FLOAT', type=click.FloatRange(min=0), default=1)
@click.option('--hand_disc_weight',          help='hand disc loss weight  [default: varies]', metavar='FLOAT', type=click.FloatRange(min=0), default=1)
@click.option('--eik_weight',          help='eikonal loss weight  [default: varies]', metavar='FLOAT',     type=click.FloatRange(min=0), default=1e-3)
@click.option('--minsurf_weight',          help='minsurf loss weight  [default: varies]', metavar='FLOAT',     type=click.FloatRange(min=0), default=5e-2)
@click.option('--smplx_reg_weight',          help='smpl regularization weight  [default: varies]', metavar='FLOAT',     type=click.FloatRange(min=0), default=0.1)
@click.option('--deform_reg_weight',          help='deformation regularization weight  [default: varies]', metavar='FLOAT',     type=click.FloatRange(min=0), default=1.0)
@click.option('--sigmoid_beta',          help='sigmoid beta initialization for sdf to density funciton', metavar='FLOAT',     type=click.FloatRange(min=0), default=3.0)
@click.option('--use_deformation', help='Enable to blur even on resume', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--backbone_face_pkl', help='Face backbone checkpoint', metavar='STR',  type=str, required=False, default=None)
@click.option('--backbone_hand_pkl', help='Face backbone checkpoint', metavar='STR',  type=str, required=False, default=None)
@click.option('--backbone_body_pkl', help='Body backbone checkpoint', metavar='STR',  type=str, required=False, default=None)
@click.option('--canonical_reg', help='Enable to blur even on resume', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--representation',      help='Triplane or TriVolume or Hyperplane representation',       type=click.Choice(['plane', 'volume', 'hplane', 'density','canoplane']), default='plane', show_default=True)
@click.option('--part_disc', help='Enable part based discrimination', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--body_sdf_from_obs', help='Enable to blur even on resume', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--use_pose_cond', help='Enable to blur even on resume', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--smplx_reg_full_space', help='Enable to blur even on resume', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--raw_sdf', help='Predict raw sdf', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--no_sdf_prior', help='Disable the use of body sdf prior', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--positional_encoding', help='If use positional encoding on Triplane', metavar='BOOL',  type=bool, required=False, default=False)
@click.option('--embed_face_cond', help='Condition mapping network on face latent code', metavar='BOOL', type=bool, default=False, show_default=True)

def main(**kwargs):
    """Train a GAN using the techniques described in the paper
    "Alias-Free Generative Adversarial Networks".
    Examples:
    \b
    # Train StyleGAN3-T for AFHQv2 using 8 GPUs.
    python train.py --outdir=~/training-runs --cfg=stylegan3-t --data=~/datasets/afhqv2-512x512.zip \\
        --gpus=8 --batch=32 --gamma=8.2 --mirror=1
    \b
    # Fine-tune StyleGAN3-R for MetFaces-U using 1 GPU, starting from the pre-trained FFHQ-U pickle.
    python train.py --outdir=~/training-runs --cfg=stylegan3-r --data=~/datasets/metfacesu-1024x1024.zip \\
        --gpus=8 --batch=32 --gamma=6.6 --mirror=1 --kimg=5000 --snap=5 \\
        --resume=https://api.ngc.nvidia.com/v2/models/nvidia/research/stylegan3/versions/1/files/stylegan3-r-ffhqu-1024x1024.pkl
    \b
    # Train StyleGAN2 for FFHQ at 1024x1024 resolution using 8 GPUs.
    python train.py --outdir=~/training-runs --cfg=stylegan2 --data=~/datasets/ffhq-1024x1024.zip \\
        --gpus=8 --batch=32 --gamma=10 --mirror=1 --aug=noaug
    """
    
    init_dist()
    
    # Initialize config.
    opts = dnnlib.EasyDict(kwargs) # Command line arguments.
    c = dnnlib.EasyDict() # Main config dict.
    c.G_kwargs = dnnlib.EasyDict(class_name=None, z_dim=512, w_dim=512, mapping_kwargs=dnnlib.EasyDict())
    c.D_kwargs = dnnlib.EasyDict(class_name='training.networks_stylegan2.Discriminator', block_kwargs=dnnlib.EasyDict(), mapping_kwargs=dnnlib.EasyDict(), epilogue_kwargs=dnnlib.EasyDict())
    c.G_opt_kwargs = dnnlib.EasyDict(class_name='torch.optim.Adam', betas=[0,0.99], eps=1e-8)
    c.D_opt_kwargs = dnnlib.EasyDict(class_name='torch.optim.Adam', betas=[0,0.99], eps=1e-8)
    if opts.representation=='density':
        # Density fields
        c.loss_kwargs = dnnlib.EasyDict(class_name='training.loss.StyleGAN2DensityLoss')
    else:
        # Signed distance fields
        c.loss_kwargs = dnnlib.EasyDict(class_name='training.loss.StyleGAN2Loss')
    c.data_loader_kwargs = dnnlib.EasyDict(pin_memory=True, prefetch_factor=2)

    # Training set.
    c.training_set_kwargs, dataset_name = init_dataset_kwargs(data=opts.data, rebalance=opts.rebalance)
    if opts.cond and not c.training_set_kwargs.use_labels:
        raise click.ClickException('--cond=True requires labels specified in dataset.json')
    c.training_set_kwargs.use_labels = opts.cond
    c.training_set_kwargs.xflip = opts.mirror

    # Hyperparameters & settings.
    c.num_gpus = opts.gpus
    c.batch_size = opts.batch
    c.batch_gpu = opts.batch_gpu or opts.batch // opts.gpus
    c.G_kwargs.channel_base = c.D_kwargs.channel_base = opts.cbase
    c.G_kwargs.channel_max = c.D_kwargs.channel_max = opts.cmax
    c.G_kwargs.mapping_kwargs.num_layers = opts.map_depth
    c.G_kwargs.mapping_kwargs.w_face_dim = 512 if opts.embed_face_cond else 0
    c.D_kwargs.block_kwargs.freeze_layers = opts.freezed
    c.D_kwargs.epilogue_kwargs.mbstd_group_size = opts.mbstd_group
    c.loss_kwargs.r1_gamma = opts.gamma
    c.G_opt_kwargs.lr = (0.002 if opts.cfg == 'stylegan2' else 0.0025) if opts.glr is None else opts.glr
    c.D_opt_kwargs.lr = opts.dlr
    c.metrics = opts.metrics
    c.total_kimg = opts.kimg
    c.kimg_per_tick = opts.tick
    c.image_snapshot_ticks = c.network_snapshot_ticks = opts.snap
    c.random_seed = c.training_set_kwargs.random_seed = opts.seed
    c.data_loader_kwargs.num_workers = opts.workers
    c.resume_kimg = opts.resume_kimg

    # Sanity checks.
    if c.batch_size % c.num_gpus != 0:
        raise click.ClickException('--batch must be a multiple of --gpus')
    if c.batch_size % (c.num_gpus * c.batch_gpu) != 0:
        raise click.ClickException('--batch must be a multiple of --gpus times --batch-gpu')
    if c.batch_gpu < c.D_kwargs.epilogue_kwargs.mbstd_group_size:
        raise click.ClickException('--batch-gpu cannot be smaller than --mbstd')
    if any(not metric_main.is_valid_metric(metric) for metric in c.metrics):
        raise click.ClickException('\n'.join(['--metrics can only contain the following values:'] + metric_main.list_valid_metrics()))

    # Base configuration.
    c.ema_kimg = c.batch_size * 10 / 32
    if opts.representation=='plane':
        c.G_kwargs.class_name = 'training.triplane.TriPlaneGenerator'
    elif opts.representation=='canoplane':
        c.G_kwargs.class_name = 'training.triplane_cano.CanoTriPlaneGenerator'
    elif opts.representation=='volume':
        c.G_kwargs.class_name = 'training.trivolume.TriVolumeGenerator'
    elif opts.representation=='hplane':
        c.G_kwargs.class_name = 'training.hyperplane.HyperPlaneGenerator'
    elif opts.representation=='density':
        c.G_kwargs.class_name = 'training.triplane_density.TriPlaneDensityGenerator'
    else:
        raise NotImplementedError
    
    c.D_kwargs.class_name = 'training.dual_discriminator.JointDualDiscriminator'
    
    c.G_kwargs.fused_modconv_default = 'inference_only' # Speed up training by using regular convolutions instead of grouped convolutions.
    c.loss_kwargs.filter_mode = 'antialiased' # Filter mode for raw images ['antialiased', 'none', float [0-1]]
    c.D_kwargs.disc_c_noise = opts.disc_c_noise # Regularization for discriminator pose conditioning

    if c.training_set_kwargs.resolution[0] == 1024:
        sr_module = 'training.superresolution.SuperresolutionHybrid4X'
    elif c.training_set_kwargs.resolution[0] == 512:
        sr_module = 'training.superresolution.SuperresolutionHybrid2X'
    elif c.training_set_kwargs.resolution[0] == 256:
        sr_module = 'training.superresolution.SuperresolutionHybrid1X'
    else:
        assert False, f"Unsupported resolution {c.training_set_kwargs.resolution}; make a new superresolution module"
    
    if opts.sr_module != None:
        sr_module = opts.sr_module
    
    rendering_options = {
        'image_resolution': c.training_set_kwargs.resolution,
        'disparity_space_sampling': False,
        'clamp_mode': 'softplus',
        'superresolution_module': sr_module,
        'c_gen_conditioning_zero': not opts.gen_pose_cond, # if true, fill generator pose conditioning label with dummy zero vector
        'gpc_reg_prob': opts.gpc_reg_prob if opts.gen_pose_cond else None,
        'c_scale': opts.c_scale, # mutliplier for generator pose conditioning label
        'superresolution_noise_mode': opts.sr_noise_mode, # [random or none], whether to inject pixel noise into super-resolution layers
        'density_reg': opts.density_reg, # strength of density regularization
        'density_reg_p_dist': opts.density_reg_p_dist, # distance at which to sample perturbed points for density regularization
        'reg_type': opts.reg_type, # for experimenting with variations on density regularization
        'decoder_lr_mul': opts.decoder_lr_mul, # learning rate multiplier for decoder
        'sr_antialias': True,
    }

    if opts.cfg == 'ffhq':
        rendering_options.update({
            'depth_resolution': 48, # number of uniform samples to take per ray.
            'depth_resolution_importance': 48, # number of importance samples to take per ray.
            'ray_start': 2.25, # near point along each ray to start taking samples.
            'ray_end': 3.3, # far point along each ray to stop taking samples. 
            'box_warp': 1, # the side-length of the bounding box spanned by the tri-planes; box_warp=1 means [-0.5, -0.5, -0.5] -> [0.5, 0.5, 0.5].
            'avg_camera_radius': 2.7, # used only in the visualizer to specify camera orbit radius.
            'avg_camera_pivot': [0, 0, 0.2], # used only in the visualizer to control center of camera rotation.
        })
    elif opts.cfg == 'afhq':
        rendering_options.update({
            'depth_resolution': 48,
            'depth_resolution_importance': 48,
            'ray_start': 2.25,
            'ray_end': 3.3,
            'box_warp': 1,
            'avg_camera_radius': 2.7,
            'avg_camera_pivot': [0, 0, -0.06],
        })
    elif opts.cfg == 'shapenet':
        rendering_options.update({
            'depth_resolution': 64,
            'depth_resolution_importance': 64,
            'ray_start': 0.1,
            'ray_end': 2.6,
            'box_warp': 1.6,
            'white_back': True,
            'avg_camera_radius': 1.7,
            'avg_camera_pivot': [0, 0, 0],
        })
    elif opts.cfg == 'renderpeople':
        rendering_options.update({
            'depth_resolution': 24,
            'depth_resolution_importance': 24,
            'ray_start': 1.2,
            'ray_end': 2.8,
            'box_warp': 2.4,
            'white_back': True,
            'avg_camera_radius': 2.0,
            'avg_camera_pivot': [0, 0, 0],
        })
    elif opts.cfg in ["mpv", "mpv_smplx"]:
        rendering_options.update({
            'depth_resolution': opts.depth_resolution,
            'depth_resolution_importance': opts.depth_resolution_importance,
            'ray_start': 7.6,
            'ray_end': 11.4,
            'box_warp': 5,
            'white_back': True,
            'use_cam_dist': True,
            'avg_camera_radius': 10.0,
            'avg_camera_pivot': [0, 0, 0],
            'calc_eikonal_coarse': opts.calc_eikonal_coarse,
            'sigmoid_beta': opts.sigmoid_beta,
            "use_deformation": opts.use_deformation,
            "backbone_face_pkl": opts.backbone_face_pkl,
            "backbone_hand_pkl": opts.backbone_hand_pkl,
            "backbone_body_pkl": opts.backbone_body_pkl,
            "canonical_reg": opts.canonical_reg,
            "body_sdf_from_obs": opts.body_sdf_from_obs,
            "part_disc": opts.part_disc,
            "use_pose_cond": opts.use_pose_cond,
            "smplx_reg_full_space": opts.smplx_reg_full_space
        })
    else:
        assert False, "Need to specify config"

    if opts.density_reg > 0:
        c.G_reg_interval = opts.density_reg_every
    c.G_kwargs.rendering_kwargs = rendering_options
    c.G_kwargs.num_fp16_res = 0
    c.loss_kwargs.blur_init_sigma = 10 # Blur the images seen by the discriminator.
    c.loss_kwargs.blur_fade_kimg = c.batch_size * opts.blur_fade_kimg / 32 # Fade out the blur during the first N kimg.

    c.loss_kwargs.gpc_reg_prob = opts.gpc_reg_prob if opts.gen_pose_cond else None
    c.loss_kwargs.gpc_reg_fade_kimg = opts.gpc_reg_fade_kimg
    c.loss_kwargs.dual_discrimination = True
    c.loss_kwargs.neural_rendering_resolution_initial = opts.neural_rendering_resolution_initial
    c.loss_kwargs.neural_rendering_resolution_final = opts.neural_rendering_resolution_final
    c.loss_kwargs.neural_rendering_resolution_fade_kimg = opts.neural_rendering_resolution_fade_kimg
    c.G_kwargs.sr_num_fp16_res = opts.sr_num_fp16_res

    c.G_kwargs.sr_kwargs = dnnlib.EasyDict(channel_base=opts.cbase, channel_max=opts.cmax, fused_modconv_default='inference_only')

    c.loss_kwargs.style_mixing_prob = opts.style_mixing_prob
    # loss args for avatargen
    c.loss_kwargs.face_disc_weight = opts.face_disc_weight
    c.loss_kwargs.hand_disc_weight = opts.hand_disc_weight
    c.loss_kwargs.eik_weight = opts.eik_weight
    c.loss_kwargs.minsurf_weight = opts.minsurf_weight
    c.loss_kwargs.smplx_reg_weight = opts.smplx_reg_weight
    c.loss_kwargs.deform_reg_weight = opts.deform_reg_weight
    # Augmentation.
    if opts.aug != 'noaug':
        c.augment_kwargs = dnnlib.EasyDict(class_name='training.augment.AugmentPipe', xflip=1, rotate90=1, xint=1, scale=1, rotate=1, aniso=1, xfrac=1, brightness=1, contrast=1, lumaflip=1, hue=1, saturation=1)
        if opts.aug == 'ada':
            c.ada_target = opts.target
        if opts.aug == 'fixed':
            c.augment_p = opts.p

    # Resume.
    if opts.resume is not None:
        c.resume_pkl = opts.resume
        c.ada_kimg = 100 # Make ADA react faster at the beginning.
        c.ema_rampup = None # Disable EMA rampup.
    
    if opts.backbone_face_pkl is not None:
        c.ada_kimg = 100
        c.ema_rampup = None
        c.loss_kwargs.blur_init_sigma = 0
        c.loss_kwargs.blur_fade_kimg = 100
        c.loss_kwargs.gpc_reg_fade_kimg = 0
    # if not opts.enable_blur:
    #     c.loss_kwargs.blur_init_sigma = 0 # Disable blur rampup.
    #     c.loss_kwargs.blur_fade_kimg = 0
        # c.loss_kwargs.blur_raw_target = False
        # c.loss_kwargs.gpc_reg_fade_kimg = 0 # Disable swapping rampup
        # c.loss_kwargs.style_mixing_prob = 0
        # c.loss_kwargs.gpc_reg_prob = 0.5

    # Performance-related toggles.
    # if opts.fp32:
    #     c.G_kwargs.num_fp16_res = c.D_kwargs.num_fp16_res = 0
    #     c.G_kwargs.conv_clamp = c.D_kwargs.conv_clamp = None
    c.G_kwargs.num_fp16_res = opts.g_num_fp16_res
    c.G_kwargs.conv_clamp = 256 if opts.g_num_fp16_res > 0 else None
    c.D_kwargs.num_fp16_res = opts.d_num_fp16_res
    c.D_kwargs.conv_clamp = 256 if opts.d_num_fp16_res > 0 else None
    c.D_kwargs.backbone_body_pkl = opts.backbone_body_pkl
    c.D_kwargs.backbone_face_pkl = opts.backbone_face_pkl
    c.D_kwargs.backbone_hand_pkl = opts.backbone_hand_pkl

    if opts.nobench:
        c.cudnn_benchmark = False

    # Description string.
    desc = f'{opts.cfg:s}-{dataset_name:s}-gpus{c.num_gpus:d}-batch{c.batch_size:d}-gamma{c.loss_kwargs.r1_gamma:g}'
    if opts.desc is not None:
        desc += f'-{opts.desc}'

    # Launch.
    launch_training(c=c, desc=desc, outdir=opts.outdir, dry_run=opts.dry_run)

#----------------------------------------------------------------------------

if __name__ == "__main__":
    main() # pylint: disable=no-value-for-parameter

#----------------------------------------------------------------------------