From 639af22b48b37d9eae6d64125d1a81f5a8f15c7e Mon Sep 17 00:00:00 2001 From: Hunor Laczko Date: Thu, 6 Jul 2023 16:34:15 +0200 Subject: [PATCH] config updates --- configs/archive/pyr_32.py | 4 +- configs/archive/pyr_64_cape_m.py | 4 +- configs/archive/pyr_64_cape_m_3.py | 4 +- configs/iccvw/ae_static_cape_m.py | 1 + configs/iccvw/pyr_32.py | 4 +- configs/iccvw/pyr_64.py | 4 +- configs/iccvw/pyr_64_128.py | 302 ++++++++++++++++++++++++ configs/iccvw/pyr_64_128_256_cape_f.py | 297 ++++++++++++++++++++++++ configs/iccvw/pyr_64_128_256_cape_m.py | 296 +++++++++++++++++++++++ configs/iccvw/pyr_64_128_cape_f.py | 297 ++++++++++++++++++++++++ configs/iccvw/pyr_64_128_cape_f_2.py | 297 ++++++++++++++++++++++++ configs/iccvw/pyr_64_128_cape_m.py | 296 +++++++++++++++++++++++ configs/iccvw/pyr_64_128_cape_m_2.py | 297 ++++++++++++++++++++++++ configs/iccvw/pyr_64_128_skirt.py | 302 ++++++++++++++++++++++++ configs/iccvw/pyr_64_2.py | 4 +- configs/iccvw/pyr_64_cape_f.py | 12 +- configs/iccvw/pyr_64_cape_m.py | 13 +- configs/iccvw/pyr_64_cape_m_skin.py | 297 ++++++++++++++++++++++++ configs/iccvw/pyr_64_norm.py | 4 +- configs/iccvw/pyr_64_norm_skirt.py | 288 +++++++++++++++++++++++ configs/iccvw/pyr_64_reg.py | 4 +- configs/iccvw/pyr_64_reg_skirt.py | 306 ++++++++++++++++++++++++ configs/iccvw/pyr_64_skin.py | 4 +- configs/iccvw/pyr_64_skin_skirt.py | 309 +++++++++++++++++++++++++ configs/iccvw/pyr_64_skirt.py | 4 +- 25 files changed, 3620 insertions(+), 30 deletions(-) create mode 100644 configs/iccvw/pyr_64_128.py create mode 100644 configs/iccvw/pyr_64_128_256_cape_f.py create mode 100644 configs/iccvw/pyr_64_128_256_cape_m.py create mode 100644 configs/iccvw/pyr_64_128_cape_f.py create mode 100644 configs/iccvw/pyr_64_128_cape_f_2.py create mode 100644 configs/iccvw/pyr_64_128_cape_m.py create mode 100644 configs/iccvw/pyr_64_128_cape_m_2.py create mode 100644 configs/iccvw/pyr_64_128_skirt.py create mode 100644 configs/iccvw/pyr_64_cape_m_skin.py create mode 100644 configs/iccvw/pyr_64_norm_skirt.py create mode 100644 configs/iccvw/pyr_64_reg_skirt.py create mode 100644 configs/iccvw/pyr_64_skin_skirt.py diff --git a/configs/archive/pyr_32.py b/configs/archive/pyr_32.py index 462fec2..deff7a4 100644 --- a/configs/archive/pyr_32.py +++ b/configs/archive/pyr_32.py @@ -134,7 +134,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -158,7 +158,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/archive/pyr_64_cape_m.py b/configs/archive/pyr_64_cape_m.py index cee4022..39b8b34 100644 --- a/configs/archive/pyr_64_cape_m.py +++ b/configs/archive/pyr_64_cape_m.py @@ -134,7 +134,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -158,7 +158,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/archive/pyr_64_cape_m_3.py b/configs/archive/pyr_64_cape_m_3.py index 9ff549b..e95432b 100644 --- a/configs/archive/pyr_64_cape_m_3.py +++ b/configs/archive/pyr_64_cape_m_3.py @@ -134,7 +134,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -158,7 +158,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/ae_static_cape_m.py b/configs/iccvw/ae_static_cape_m.py index 25e2662..90ed63b 100644 --- a/configs/iccvw/ae_static_cape_m.py +++ b/configs/iccvw/ae_static_cape_m.py @@ -16,6 +16,7 @@ config['valid_dir'] = '/data-local/127_2/cape_processed/test/male' config['test_dir'] = '/data-local/127_2/cape_processed/test/male' config['train_frames'] = None config['valid_frames'] = None +config['test_frames'] = None config['resolution'] = 512 config['is_skirt'] = False diff --git a/configs/iccvw/pyr_32.py b/configs/iccvw/pyr_32.py index c5e7f06..c43c96a 100644 --- a/configs/iccvw/pyr_32.py +++ b/configs/iccvw/pyr_32.py @@ -139,7 +139,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -163,7 +163,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/pyr_64.py b/configs/iccvw/pyr_64.py index fa6b943..2e51b80 100644 --- a/configs/iccvw/pyr_64.py +++ b/configs/iccvw/pyr_64.py @@ -139,7 +139,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -163,7 +163,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/pyr_64_128.py b/configs/iccvw/pyr_64_128.py new file mode 100644 index 0000000..71080e9 --- /dev/null +++ b/configs/iccvw/pyr_64_128.py @@ -0,0 +1,302 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '2' +config['name'] = 'pyr_iccvw_64_128' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = False +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_2' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127/cloth3d_processed' +config['valid_dir'] = '/data-local/127/cloth3d_processed' +config['test_dir'] = '/data-local/127_2/cloth3d_processed_test' + +if config['is_skirt']: + config['train_frames'] = 'frames_train_skirt.txt' + config['valid_frames'] = 'frames_valid_skirt.txt' + config['test_frames'] = 'frames_test_skirt.txt' +else: + config['train_frames'] = 'frames_train.txt' + config['valid_frames'] = 'frames_valid.txt' + config['test_frames'] = 'frames_test.txt' + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_4' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_1' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_1' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_256_cape_f.py b/configs/iccvw/pyr_64_128_256_cape_f.py new file mode 100644 index 0000000..16d453c --- /dev/null +++ b/configs/iccvw/pyr_64_128_256_cape_f.py @@ -0,0 +1,297 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '2' +config['name'] = 'pyr_iccvw_64_128_256_cape_f' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128, 256] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_128_cape_f' +config['pyramid']['pretrained_pyramid_iteration'] = '100000' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/female' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/female' +config['test_dir'] = '/data-local/127_2/cape_processed/test/female' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_f' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_cape_f' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_f' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_256_cape_m.py b/configs/iccvw/pyr_64_128_256_cape_m.py new file mode 100644 index 0000000..9164cf4 --- /dev/null +++ b/configs/iccvw/pyr_64_128_256_cape_m.py @@ -0,0 +1,296 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '0' +config['name'] = 'pyr_iccvw_64_128_256_cape_m' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128, 256] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_128_cape_m' +config['pyramid']['pretrained_pyramid_iteration'] = '125000' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/male' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/male' +config['test_dir'] = '/data-local/127_2/cape_processed/test/male' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_m' +config['body_model'] = 'checkpoints/iccvw/body_torch_cape_m' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_m' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_cape_f.py b/configs/iccvw/pyr_64_128_cape_f.py new file mode 100644 index 0000000..fed085a --- /dev/null +++ b/configs/iccvw/pyr_64_128_cape_f.py @@ -0,0 +1,297 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '2' +config['name'] = 'pyr_iccvw_64_128_cape_f' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_cape_f' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/female' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/female' +config['test_dir'] = '/data-local/127_2/cape_processed/test/female' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_f' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_cape_f' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_f' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_cape_f_2.py b/configs/iccvw/pyr_64_128_cape_f_2.py new file mode 100644 index 0000000..b04777d --- /dev/null +++ b/configs/iccvw/pyr_64_128_cape_f_2.py @@ -0,0 +1,297 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '2' +config['name'] = 'pyr_iccvw_64_128_cape_f_2' +config['notes'] = 'add a scaling to the new level' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_cape_f' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/female' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/female' +config['test_dir'] = '/data-local/127_2/cape_processed/test/female' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_f' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_cape_f' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_f' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['scaling'] = 0.01 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_f' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_cape_m.py b/configs/iccvw/pyr_64_128_cape_m.py new file mode 100644 index 0000000..5250821 --- /dev/null +++ b/configs/iccvw/pyr_64_128_cape_m.py @@ -0,0 +1,296 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '0' +config['name'] = 'pyr_iccvw_64_128_cape_m' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_cape_m' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/male' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/male' +config['test_dir'] = '/data-local/127_2/cape_processed/test/male' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_m' +config['body_model'] = 'checkpoints/iccvw/body_torch_cape_m' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_m' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_cape_m_2.py b/configs/iccvw/pyr_64_128_cape_m_2.py new file mode 100644 index 0000000..3257190 --- /dev/null +++ b/configs/iccvw/pyr_64_128_cape_m_2.py @@ -0,0 +1,297 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '3' +config['name'] = 'pyr_iccvw_64_128_cape_m_2' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_cape_m' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/male' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/male' +config['test_dir'] = '/data-local/127_2/cape_processed/test/male' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_m' +config['body_model'] = 'checkpoints/iccvw/body_torch_cape_m' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_m' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['scaling'] = 0.01 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_128_skirt.py b/configs/iccvw/pyr_64_128_skirt.py new file mode 100644 index 0000000..26008d0 --- /dev/null +++ b/configs/iccvw/pyr_64_128_skirt.py @@ -0,0 +1,302 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '4' +config['name'] = 'pyr_iccvw_64_128_skirt' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = True +config['is_cape'] = False +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64, 128] +config['pyramid']['pretrained_pyramid'] = 'checkpoints/iccvw/pyr_iccvw_64_skirt' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127/cloth3d_processed' +config['valid_dir'] = '/data-local/127/cloth3d_processed' +config['test_dir'] = '/data-local/127_2/cloth3d_processed_test' + +if config['is_skirt']: + config['train_frames'] = 'frames_train_skirt.txt' + config['valid_frames'] = 'frames_valid_skirt.txt' + config['test_frames'] = 'frames_test_skirt.txt' +else: + config['train_frames'] = 'frames_train.txt' + config['valid_frames'] = 'frames_valid.txt' + config['test_frames'] = 'frames_test.txt' + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'mesh_unposed', + 'uv_unposed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_1_skirt' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_1_skirt' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_1_skirt' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_skirt' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_skirt' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_skirt' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32_skirt' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_2.py b/configs/iccvw/pyr_64_2.py index cf7affb..4bfa113 100644 --- a/configs/iccvw/pyr_64_2.py +++ b/configs/iccvw/pyr_64_2.py @@ -139,7 +139,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -163,7 +163,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/pyr_64_cape_f.py b/configs/iccvw/pyr_64_cape_f.py index 6c58a1f..7f8fb47 100644 --- a/configs/iccvw/pyr_64_cape_f.py +++ b/configs/iccvw/pyr_64_cape_f.py @@ -33,6 +33,8 @@ config['valid_dir'] = '/data-local/127_2/cape_processed/test/female' config['test_dir'] = '/data-local/127_2/cape_processed/test/female' config['train_frames'] = None config['valid_frames'] = None +config['test_frames'] = None +config['test_frames'] = None config['dataset_in_memory'] = False config['resolution'] = None @@ -117,7 +119,7 @@ for res in config['pyramid']['resolutions']: level = dict() level['resolution'] = 256 level['kernel_size'] = 5 - level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256' + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_f' level['upscale_enhancer_freeze'] = True level['bottleneck_size'] = 64 * config['bottleneck_scale'] level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 @@ -132,7 +134,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -143,7 +145,7 @@ for res in config['pyramid']['resolutions']: level = dict() level['resolution'] = 128 level['kernel_size'] = 5 - level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128' + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_f' level['upscale_enhancer_freeze'] = True level['bottleneck_size'] = 64 * config['bottleneck_scale'] level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 @@ -156,7 +158,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] @@ -167,7 +169,7 @@ for res in config['pyramid']['resolutions']: level = dict() level['resolution'] = 64 level['kernel_size'] = 3 - level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64' + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_f' level['upscale_enhancer_freeze'] = True level['bottleneck_size'] = 64 * config['bottleneck_scale'] level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 diff --git a/configs/iccvw/pyr_64_cape_m.py b/configs/iccvw/pyr_64_cape_m.py index e639ff8..e83543b 100644 --- a/configs/iccvw/pyr_64_cape_m.py +++ b/configs/iccvw/pyr_64_cape_m.py @@ -33,6 +33,9 @@ config['valid_dir'] = '/data-local/127_2/cape_processed/test/male' config['test_dir'] = '/data-local/127_2/cape_processed/test/male' config['train_frames'] = None config['valid_frames'] = None +config['test_frames'] = None +config['test_frames'] = None +config['test_frames'] = None config['dataset_in_memory'] = False config['resolution'] = None @@ -117,7 +120,7 @@ for res in config['pyramid']['resolutions']: level = dict() level['resolution'] = 256 level['kernel_size'] = 5 - level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256' + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_m' level['upscale_enhancer_freeze'] = True level['bottleneck_size'] = 64 * config['bottleneck_scale'] level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 @@ -132,7 +135,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -143,7 +146,7 @@ for res in config['pyramid']['resolutions']: level = dict() level['resolution'] = 128 level['kernel_size'] = 5 - level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128' + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_m' level['upscale_enhancer_freeze'] = True level['bottleneck_size'] = 64 * config['bottleneck_scale'] level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 @@ -156,7 +159,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] @@ -167,7 +170,7 @@ for res in config['pyramid']['resolutions']: level = dict() level['resolution'] = 64 level['kernel_size'] = 3 - level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64' + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_m' level['upscale_enhancer_freeze'] = True level['bottleneck_size'] = 64 * config['bottleneck_scale'] level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 diff --git a/configs/iccvw/pyr_64_cape_m_skin.py b/configs/iccvw/pyr_64_cape_m_skin.py new file mode 100644 index 0000000..a32af82 --- /dev/null +++ b/configs/iccvw/pyr_64_cape_m_skin.py @@ -0,0 +1,297 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '5' +config['name'] = 'pyr_iccvw_64_cape_m_skin' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = True +config['device'] = 'cuda' + + +config['use_skinning'] = False # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64] +config['pyramid']['pretrained_pyramid'] = '' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127_2/cape_processed/train/male' +config['valid_dir'] = '/data-local/127_2/cape_processed/test/male' +config['test_dir'] = '/data-local/127_2/cape_processed/test/male' +config['train_frames'] = None +config['valid_frames'] = None +config['test_frames'] = None +config['test_frames'] = None +config['test_frames'] = None + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'uv_posed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_posed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_cape_m' +config['body_model'] = 'checkpoints/iccvw/body_torch_cape_m' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_cape_m' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64_cape_m' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_norm.py b/configs/iccvw/pyr_64_norm.py index a64fafa..44da9f3 100644 --- a/configs/iccvw/pyr_64_norm.py +++ b/configs/iccvw/pyr_64_norm.py @@ -143,7 +143,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2 - 40*2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -167,7 +167,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2 - 40*2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/pyr_64_norm_skirt.py b/configs/iccvw/pyr_64_norm_skirt.py new file mode 100644 index 0000000..653e3d6 --- /dev/null +++ b/configs/iccvw/pyr_64_norm_skirt.py @@ -0,0 +1,288 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '3' +config['name'] = 'pyr_iccvw_64_norm_skirt_2' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = True +config['is_cape'] = False +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = False + +config['pretrained_model'] = 'checkpoints/iccvw/pyr_iccvw_64_norm_skirt' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64] +config['pyramid']['pretrained_pyramid'] = '' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127/cloth3d_processed' +config['valid_dir'] = '/data-local/127/cloth3d_processed' +config['test_dir'] = '/data-local/127_2/cloth3d_processed_test' + +if config['is_skirt']: + config['train_frames'] = 'frames_train_skirt.txt' + config['valid_frames'] = 'frames_valid_skirt.txt' + config['test_frames'] = 'frames_test_skirt.txt' +else: + config['train_frames'] = 'frames_train.txt' + config['valid_frames'] = 'frames_valid.txt' + config['test_frames'] = 'frames_test.txt' + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +if config['use_skinning']: + config['signature'].add('uv_unposed') + config['signature'].add('mesh_unposed') +else: + config['signature'].add('uv_posed') + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_1_skirt' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_1_skirt' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_1_skirt' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2 - 40*2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2 - 40*2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2 - 40*2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2 - 40*2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + raise NotImplementedError + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_reg.py b/configs/iccvw/pyr_64_reg.py index 65b00d0..4132eac 100644 --- a/configs/iccvw/pyr_64_reg.py +++ b/configs/iccvw/pyr_64_reg.py @@ -143,7 +143,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -167,7 +167,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/pyr_64_reg_skirt.py b/configs/iccvw/pyr_64_reg_skirt.py new file mode 100644 index 0000000..beb3a7b --- /dev/null +++ b/configs/iccvw/pyr_64_reg_skirt.py @@ -0,0 +1,306 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '3' +config['name'] = 'pyr_iccvw_64_reg_skirt_2' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = False +config['device'] = 'cuda' + + +config['use_skinning'] = True # and deshaping +config['use_extra_losses'] = False # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = 'checkpoints/iccvw/pyr_iccvw_64_reg_skirt' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64] +config['pyramid']['pretrained_pyramid'] = '' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127/cloth3d_processed' +config['valid_dir'] = '/data-local/127/cloth3d_processed' +config['test_dir'] = '/data-local/127_2/cloth3d_processed_test' + +if config['is_skirt']: + config['train_frames'] = 'frames_train_skirt.txt' + config['valid_frames'] = 'frames_valid_skirt.txt' + config['test_frames'] = 'frames_test_skirt.txt' +else: + config['train_frames'] = 'frames_train.txt' + config['valid_frames'] = 'frames_valid.txt' + config['test_frames'] = 'frames_test.txt' + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +if config['use_skinning']: + config['signature'].add('uv_unposed') + config['signature'].add('mesh_unposed') +else: + config['signature'].add('uv_posed') + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + config['signature'].add('uv_unposed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_1_skirt' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_1_skirt' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_1_skirt' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_skin.py b/configs/iccvw/pyr_64_skin.py index 8d2e6a3..c208026 100644 --- a/configs/iccvw/pyr_64_skin.py +++ b/configs/iccvw/pyr_64_skin.py @@ -146,7 +146,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -170,7 +170,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ] diff --git a/configs/iccvw/pyr_64_skin_skirt.py b/configs/iccvw/pyr_64_skin_skirt.py new file mode 100644 index 0000000..6d58c3d --- /dev/null +++ b/configs/iccvw/pyr_64_skin_skirt.py @@ -0,0 +1,309 @@ +from utils.convnext_utils import CNBlockConfig + +config = dict() + +config['gpu'] = '4' +config['name'] = 'pyr_iccvw_64_skin_skirt' +config['notes'] = '' +config['tags'] = ['pyr', 'iccv', 'torch'] +config['script'] = 'scripts/cvae_pyramid.py' + +config['batch_size'] = 4 +config['is_skirt'] = False +config['is_cape'] = False +config['device'] = 'cuda' + + +config['use_skinning'] = False # and deshaping +config['use_extra_losses'] = True # edge, collision, 3d mesh +config['use_norm_loss'] = True +config['use_normals'] = True + +config['pretrained_model'] = '' +config['pretrained_model_iteration'] = '' + +config['pyramid'] = dict() +config['pyramid']['resolutions'] = [64] +config['pyramid']['pretrained_pyramid'] = '' +config['pyramid']['pretrained_pyramid_iteration'] = '' +config['pyramid']['final_upscale_enhancer'] = '' + +config['train_dir'] = '/data-local/127/cloth3d_processed' +config['valid_dir'] = '/data-local/127/cloth3d_processed' +config['test_dir'] = '/data-local/127_2/cloth3d_processed_test' + +if config['is_skirt']: + config['train_frames'] = 'frames_train_skirt.txt' + config['valid_frames'] = 'frames_valid_skirt.txt' + config['test_frames'] = 'frames_test_skirt.txt' +else: + config['train_frames'] = 'frames_train.txt' + config['valid_frames'] = 'frames_valid.txt' + config['test_frames'] = 'frames_test.txt' + +config['dataset_in_memory'] = False +config['resolution'] = None + +config['smpl_res'] = 14475 +if config['is_cape']: + config['smpl_res'] = 3638 + +config['lr'] = 0.0001 +config['starting_balancing_coeff'] = 0.001 +config['balancing_activation_limit'] = 0.012 +config['reg_loss_coeff'] = 0.1 # set to 0 if don't want to use edge+collision +config['avg_garment_path'] = '' +config['bottleneck_scale'] = 2 + +config['signature']: set = { 'id', + 'frame_nr', + 'pose_shape', + 'uv_body_posed', + 'mesh_body_posed', + 'mesh_posed', + 'uv_normals', + 'uv_static', + 'mesh_mask', + 'uv_mask', + } + +config['pyramid']['is_label'] = False +# config['pyramid']['add_noise'] = False +# config['pyramid']['noise_amount'] = 0.0 + +config['pyramid']['loss_weights'] = [1,1,1,1,1] + +if config['use_skinning']: + config['signature'].add('uv_unposed') + config['signature'].add('mesh_unposed') +else: + config['signature'].add('uv_posed') + +for res in config['pyramid']['resolutions']: + if res != 512: + res_modifier = '_' + str(res) + config['signature'].add('uv_body_posed' + res_modifier) + if config['use_skinning']: + config['signature'].add('uv_unposed' + res_modifier) + else: + config['signature'].add('uv_posed' + res_modifier) + config['signature'].add('uv_normals' + res_modifier) + config['signature'].add('uv_static' + res_modifier) + config['signature'].add('uv_mask' + res_modifier) + + +config['static_model'] = 'checkpoints/iccvw/static_iccvw_1_skirt' +config['body_model'] = 'checkpoints/iccvw/body_iccvw_1_skirt' +config['normals_model'] = 'checkpoints/iccvw/norm_iccvw_1_skirt' +config['freeze_encoders'] = True + + +config['pyramid']['levels'] = [] +for res in config['pyramid']['resolutions']: + if res == 512: + level = dict() + level['resolution'] = 512 + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['kernel_size'] = 7 + level['upscale_enhancer'] = '' + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 3), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 256: + level = dict() + level['resolution'] = 256 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_256' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 3), + CNBlockConfig(128, 128, 3), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 3), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), + CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 3), + CNBlockConfig(64, 3, 3), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 128: + level = dict() + level['resolution'] = 128 + level['kernel_size'] = 5 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_128' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 64: + level = dict() + level['resolution'] = 64 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_64' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(64, level['bottleneck_size'], 9), + CNBlockConfig(level['bottleneck_size'], None, 9), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 64, 9), + CNBlockConfig(64, 3, 9), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + elif res == 32: + level = dict() + level['resolution'] = 32 + level['kernel_size'] = 3 + level['upscale_enhancer'] = 'checkpoints/iccvw/iccv_32' + level['upscale_enhancer_freeze'] = True + level['bottleneck_size'] = 64 * config['bottleneck_scale'] + level['input_features'] = 3+3+3+3 if config['use_normals'] else 3+3+3 + + block_setting = [ + CNBlockConfig(level['bottleneck_size'], None, 16), + ] + + block_setting_reverse = [ + CNBlockConfig(level['bottleneck_size'] * 2, 3, 16), + ] + + level['encoder_block_settings'] = block_setting + level['decoder_block_settings'] = block_setting_reverse + config['pyramid']['levels'].append(level) + else: + raise NotImplementedError + + +####################### cond_ae_norm ############################ +config['cond_ae_norm'] = dict() +config['cond_ae_norm']['device'] = config['device'] +config['cond_ae_norm']['is_label'] = False +# config['cond_ae_norm']['latent_dim'] = config['latent_dim'] +config['cond_ae_norm']['kernel_size'] = 7 +config['cond_ae_norm']['bottleneck_size'] = 40 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_norm']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_norm']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_norm']['encoder_block_settings'] = block_setting +config['cond_ae_norm']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_norm']['activation'] = 'tanh' + + +####################### cond_ae_body ############################ +config['cond_ae_body'] = dict() +config['cond_ae_body']['device'] = config['device'] +config['cond_ae_body']['is_label'] = False +# config['cond_ae_body']['latent_dim'] = config['latent_dim'] +config['cond_ae_body']['kernel_size'] = 7 +config['cond_ae_body']['bottleneck_size'] = 16 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, 128, 9), + CNBlockConfig(128, config['cond_ae_body']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_body']['bottleneck_size'], 128, 3), + CNBlockConfig(128, 64, 9), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 3, 3), +] + +config['cond_ae_body']['encoder_block_settings'] = block_setting +config['cond_ae_body']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_body']['activation'] = 'tanh' + +####################### cond_ae_static ############################ +config['cond_ae_static'] = dict() +config['cond_ae_static']['device'] = config['device'] +config['cond_ae_static']['is_label'] = False +# config['cond_ae_static']['latent_dim'] = config['latent_dim'] +config['cond_ae_static']['kernel_size'] = 7 +config['cond_ae_static']['bottleneck_size'] = 8 * config['bottleneck_scale'] + +block_setting = [ + CNBlockConfig(8, 16, 3), + CNBlockConfig(16, 32, 3), + CNBlockConfig(32, 64, 3), + CNBlockConfig(64, config['cond_ae_static']['bottleneck_size'], 9), + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], None, 3), +] + +block_setting_reverse = [ + CNBlockConfig(config['cond_ae_static']['bottleneck_size'], 64, 3), + CNBlockConfig(64, 32, 9), + CNBlockConfig(32, 16, 3), + CNBlockConfig(16, 8, 3), + CNBlockConfig(8, 3, 3), +] + +config['cond_ae_static']['encoder_block_settings'] = block_setting +config['cond_ae_static']['decoder_block_settings'] = block_setting_reverse +config['cond_ae_static']['activation'] = 'tanh' \ No newline at end of file diff --git a/configs/iccvw/pyr_64_skirt.py b/configs/iccvw/pyr_64_skirt.py index 6c1549e..9ab2c3c 100644 --- a/configs/iccvw/pyr_64_skirt.py +++ b/configs/iccvw/pyr_64_skirt.py @@ -139,7 +139,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 3), CNBlockConfig(128, 128, 9), - CNBlockConfig(128, 65, 3), + CNBlockConfig(128, 64, 3), CNBlockConfig(64, 3, 3), ] @@ -163,7 +163,7 @@ for res in config['pyramid']['resolutions']: block_setting_reverse = [ CNBlockConfig(level['bottleneck_size'] * 2, 128, 9), - CNBlockConfig(128, 128, 9), + CNBlockConfig(128, 64, 9), CNBlockConfig(64, 3, 9), ]