MCPcopy Create free account

hub / github.com/360CVGroup/FancyVideo / functions

Functions243 in github.com/360CVGroup/FancyVideo

⨍ Functions 243◇ Types & classes 62
↓ 16 callersMethodreshape_heads_to_batch_dim
(self, tensor, out_dim=3)
fancyvideo/models/ctgm/attention.py:150
↓ 13 callersFunctionassign_to_checkpoint
This does the final conversion step: take locally converted weights and apply a global renaming to them. It splits attention layers, and take
fancyvideo/utils/convert_from_ckpt.py:152
↓ 11 callersMethod__init__
( self, *, ch, out_ch, ch_mult=(1, 2, 4, 8), num_res_blocks,
fancyvideo/models/cv_vae/vae_models.py:721
↓ 8 callersFunctionNormalize
(in_channels, num_groups=32)
fancyvideo/models/cv_vae/vae_models.py:198
↓ 8 callersFunctionget_motion_module
( in_channels, motion_module_type: str, motion_module_kwargs: dict )
fancyvideo/models/motion_module.py:39
↓ 6 callersMethod__init__
(self, embed_dim: int, # vision image_resolution: int,
fancyvideo/models/long_clip/model_longclip.py:244
↓ 6 callersFunction_copy_linear
(hf_linear, pt_linear)
fancyvideo/utils/convert_from_ckpt.py:675
↓ 6 callersFunctionrenew_resnet_paths
Updates paths inside resnets to the new naming scheme (local renaming)
fancyvideo/utils/convert_from_ckpt.py:63
↓ 5 callersMethodforward
(self, x)
fancyvideo/models/cv_vae/vae_models.py:831
↓ 5 callersFunctionnonlinearity
(x)
fancyvideo/models/cv_vae/vae_models.py:193
↓ 5 callersFunctionshave_segments
Removes segments. Positive values shave the first segments, negative shave the last segments.
fancyvideo/utils/convert_from_ckpt.py:53
↓ 4 callersMethod__init__
( self, in_channels: int, prev_output_channel: int, out_channels: int,
fancyvideo/models/unet_blocks.py:769
↓ 4 callersMethod__init__
( self, d_model, dropout = 0., max_len = 24 )
fancyvideo/models/motion_module.py:229
↓ 4 callersMethod_make_layer
(self, planes, blocks, stride=1)
fancyvideo/models/long_clip/model_longclip.py:129
↓ 4 callersFunction_node_get
Gets attributes of a node which is polymorphic over return type. From https://github.com/pytorch/pytorch/pull/82628
fancyvideo/models/long_clip/longclip.py:79
↓ 4 callersFunction_transform
(n_px)
fancyvideo/models/long_clip/longclip.py:35
↓ 4 callersMethoddecode
(self, z: torch.Tensor, num_frames: int = None, return_dict: bool
fancyvideo/models/cv_vae/modeling_vae.py:288
↓ 4 callersMethodencode
( self, x: torch.Tensor, return_dict: bool = True )
fancyvideo/models/cv_vae/modeling_vae.py:201
↓ 4 callersFunctionpatch_device
(module)
fancyvideo/models/long_clip/longclip.py:87
↓ 4 callersFunctionpatch_float
(module)
fancyvideo/models/long_clip/longclip.py:111
↓ 4 callersFunctionrenew_vae_resnet_paths
Updates paths inside resnets to the new naming scheme (local renaming)
fancyvideo/utils/convert_from_ckpt.py:85
↓ 3 callersMethod__init__
( self, attention_mode = None, cross_frame_attention_m
fancyvideo/models/ctgm/temporal_cross_module.py:259
↓ 3 callersMethod_attention
(self, query, key, value,cross_frame_txtattn=False,block_status=None,trans_index=None,timesteps=None,attention
fancyvideo/models/ctgm/attention.py:267
↓ 3 callersMethod_memory_efficient_attention_xformers
(self, query, key, value, attention_mask)
fancyvideo/models/ctgm/attention.py:365
↓ 3 callersFunctionconvert_ldm_unet_checkpoint
Takes a state dict and a config, and returns a converted checkpoint.
fancyvideo/utils/convert_from_ckpt.py:328
↓ 3 callersFunctionrenew_attention_paths
Updates paths inside attentions to the new naming scheme (local renaming)
fancyvideo/utils/convert_from_ckpt.py:101
↓ 2 callersMethod__init__
(self, channels, use_conv=False, use_conv_transpose=False, out_channels=None, name="conv")
fancyvideo/models/resnet.py:33
↓ 2 callersMethod__init__
( self, query_dim: int, cross_attention_dim: Optional[int] = None, heads: int
fancyvideo/models/ctgm/attention.py:48
↓ 2 callersMethodblend_h
( self, a: torch.Tensor, b: torch.Tensor, overlap_size: int )
fancyvideo/models/cv_vae/modeling_vae.py:308
↓ 2 callersMethodblend_v
( self, a: torch.Tensor, b: torch.Tensor, overlap_size: int )
fancyvideo/models/cv_vae/modeling_vae.py:319
↓ 2 callersFunctionbuild_model
(state_dict: dict, load_from_clip: bool)
fancyvideo/models/long_clip/model_longclip.py:512
↓ 2 callersFunctionbytes_to_unicode
Returns list of utf-8 byte and a corresponding list of unicode strings. The reversible bpe codes work on unicode strings. This means you
fancyvideo/models/long_clip/simple_tokenizer.py:16
↓ 2 callersFunctionconv_attn_to_linear
(checkpoint)
fancyvideo/utils/convert_from_ckpt.py:203
↓ 2 callersFunctionconvert_ldm_clip_checkpoint
(checkpoint)
fancyvideo/utils/convert_from_ckpt.py:716
↓ 2 callersFunctionconvert_ldm_vae_checkpoint
(checkpoint, config)
fancyvideo/utils/convert_from_ckpt.py:559
↓ 2 callersFunctiondefault
(val, d)
fancyvideo/models/cv_vae/vae_models.py:37
↓ 2 callersMethodencode_text
(self, text)
fancyvideo/models/long_clip/model_longclip.py:356
↓ 2 callersMethodencode_text_full
(self, text)
fancyvideo/models/long_clip/model_longclip.py:372
↓ 2 callersFunctionexists
(val)
fancyvideo/models/cv_vae/vae_models.py:34
↓ 2 callersMethodforward
(self, hidden_states)
fancyvideo/models/resnet.py:219
↓ 2 callersFunctionget_pairs
Return set of symbol pairs in a word. Word is represented as tuple of symbols (symbols being variable-length strings).
fancyvideo/models/long_clip/simple_tokenizer.py:38
↓ 2 callersFunctionmake_attn
(in_channels, attn_type="vanilla", attn_kwargs=None)
fancyvideo/models/cv_vae/vae_models.py:682
↓ 2 callersFunctionprocessing_reference_image
(reference_image, vae, vae_type, device, video_length, resolution)
fancyvideo/pipelines/fancyvideo_infer_pipeline.py:31
↓ 2 callersFunctionrenew_vae_attention_paths
Updates paths inside attentions to the new naming scheme (local renaming)
fancyvideo/utils/convert_from_ckpt.py:122
↓ 2 callersMethodreshape_batch_dim_to_heads
(self, tensor)
fancyvideo/models/ctgm/attention.py:162
↓ 2 callersMethodspatial_tiled_decode
(self, z: torch.Tensor, **kwargs)
fancyvideo/models/cv_vae/modeling_vae.py:220
↓ 2 callersMethodspatial_tiled_encode
(self, x)
fancyvideo/models/cv_vae/modeling_vae.py:133
↓ 1 callersMethodPCA
(self, input_tensor, PCA_dim)
fancyvideo/models/long_clip/model_longclip.py:389
↓ 1 callersMethod__init__
(self, temporal_position_encoding_max_len=32, model_path="",
fancyvideo/pipelines/fancyvideo_infer_pipeline.py:98
↓ 1 callersMethod__init__
(self, in_channels: int, cond_embed_dim: int, act_fn: str = "silu")
fancyvideo/models/unet.py:45
↓ 1 callersFunction_copy_attn_layer
(hf_attn_layer, pt_attn_layer)
fancyvideo/utils/convert_from_ckpt.py:667
↓ 1 callersFunction_copy_layer
(hf_layer, pt_layer)
fancyvideo/utils/convert_from_ckpt.py:679
↓ 1 callersFunction_copy_layers
(hf_layers, pt_layers)
fancyvideo/utils/convert_from_ckpt.py:692
↓ 1 callersFunction_download
(url: str, root: str)
fancyvideo/models/long_clip/longclip.py:178
↓ 1 callersMethod_encode_prompt
(self, prompt, device, num_videos_per_prompt, do_classifier_free_guidance, negative_prompt)
fancyvideo/pipelines/pipeline_animation_vae_2d.py:168
↓ 1 callersMethod_encode_prompt
(self, prompt, device, num_videos_per_prompt, do_classifier_free_guidance, negative_prompt)
fancyvideo/pipelines/pipeline_animation_vae_3d.py:169
↓ 1 callersMethod_make_attn
(self)
fancyvideo/models/cv_vae/vae_models.py:987
↓ 1 callersMethod_make_conv
(self)
fancyvideo/models/cv_vae/vae_models.py:993
↓ 1 callersMethod_make_resblock
(self)
fancyvideo/models/cv_vae/vae_models.py:990
↓ 1 callersMethodattention
(self, h_: torch.Tensor)
fancyvideo/models/cv_vae/vae_models.py:452
↓ 1 callersMethodattention
(self, h_: torch.Tensor)
fancyvideo/models/cv_vae/vae_models.py:506
↓ 1 callersMethodattention
(self, h_: torch.Tensor)
fancyvideo/models/cv_vae/vae_models.py:618
↓ 1 callersMethodattention
(self, x: torch.Tensor)
fancyvideo/models/long_clip/model_longclip.py:185
↓ 1 callersMethodattention_t
(self, h_: torch.Tensor)
fancyvideo/models/cv_vae/vae_models.py:591
↓ 1 callersFunctionavailable_models
Returns the names of available CLIP models
fancyvideo/models/long_clip/longclip.py:174
↓ 1 callersFunctionbasic_clean
(text)
fancyvideo/models/long_clip/simple_tokenizer.py:50
↓ 1 callersMethodbpe
(self, token)
fancyvideo/models/long_clip/simple_tokenizer.py:80
↓ 1 callersMethodbuild_attention_mask
(self)
fancyvideo/models/long_clip/model_longclip.py:341
↓ 1 callersMethodcheck_inputs
(self, prompt, height, width, callback_steps)
fancyvideo/pipelines/pipeline_animation_vae_2d.py:297
↓ 1 callersMethodcheck_inputs
(self, prompt, height, width, callback_steps)
fancyvideo/pipelines/pipeline_animation_vae_3d.py:301
↓ 1 callersFunctionconvert_lora
(pipeline, state_dict, LORA_PREFIX_UNET="lora_unet", LORA_PREFIX_TEXT_ENCODER="lora_te", alpha=0.6)
fancyvideo/utils/convert_lora_safetensor_to_diffusers.py:52
↓ 1 callersFunctionconvert_motion_lora_ckpt_to_diffusers
(pipeline, state_dict, alpha=1.0)
fancyvideo/utils/convert_lora_safetensor_to_diffusers.py:28
↓ 1 callersFunctionconvert_weights
Convert applicable model parameters to fp16
fancyvideo/models/long_clip/model_longclip.py:488
↓ 1 callersFunctioncreate_unet_diffusers_config
Creates a config for the diffusers based on the config of the LDM model.
fancyvideo/utils/convert_from_ckpt.py:215
↓ 1 callersFunctionddim_loop
(pipeline, ddim_scheduler, latent, num_inv_steps, prompt)
fancyvideo/utils/util.py:76
↓ 1 callersMethoddecode
(self, tokens)
fancyvideo/models/long_clip/simple_tokenizer.py:129
↓ 1 callersMethoddecode_latents
(self, latents)
fancyvideo/pipelines/pipeline_animation_vae_2d.py:265
↓ 1 callersMethoddecode_latents_3d
(self, latents)
fancyvideo/pipelines/pipeline_animation_vae_3d.py:273
↓ 1 callersFunctiondefault_bpe
()
fancyvideo/models/long_clip/simple_tokenizer.py:11
↓ 1 callersMethodencode
(self, text)
fancyvideo/models/long_clip/simple_tokenizer.py:121
↓ 1 callersMethodencode_image
(self, image)
fancyvideo/models/long_clip/model_longclip.py:353
↓ 1 callersFunctionenforce_zero_terminal_snr
(betas)
fancyvideo/pipelines/fancyvideo_infer_pipeline.py:71
↓ 1 callersMethodforward
(self, image, text_long,text_short,rank)
fancyvideo/models/long_clip/model_longclip.py:442
↓ 1 callersMethodfrom_pretrained_2d
(cls, pretrained_model_path, subfolder=None, unet_additional_kwargs=None,cross_frame_txtattn=False,part_module
fancyvideo/models/unet.py:570
↓ 1 callersFunctionget_down_block
( down_block_type, num_layers, in_channels, out_channels, temb_channels, add_downsampl
fancyvideo/models/unet_blocks.py:12
↓ 1 callersFunctionget_noise_pred_single
(latents, t, context, unet)
fancyvideo/utils/util.py:70
↓ 1 callersFunctionget_temporal_text_module
( in_channels, temporal_module_type: str, temporal_module_kwargs: dict, if_mid: bool )
fancyvideo/models/ctgm/temporal_cross_module.py:40
↓ 1 callersFunctionget_up_block
( up_block_type, num_layers, in_channels, out_channels, prev_output_channel, temb_chan
fancyvideo/models/unet_blocks.py:101
↓ 1 callersFunctioninit_prompt
(prompt, pipeline)
fancyvideo/utils/util.py:38
↓ 1 callersMethodinitialize_parameters
(self)
fancyvideo/models/long_clip/model_longclip.py:312
↓ 1 callersMethodload_unet
(self, unet, motion_module_path, trained_keys)
fancyvideo/pipelines/fancyvideo_infer_pipeline.py:341
↓ 1 callersFunctionmain
(args)
scripts/demo.py:10
↓ 1 callersFunctionnext_step
(model_output: Union[torch.FloatTensor, np.ndarray], timestep: int, sample: Union[torch.FloatTen
fancyvideo/utils/util.py:57
↓ 1 callersMethodprepare_extra_step_kwargs
(self, generator, eta)
fancyvideo/pipelines/pipeline_animation_vae_2d.py:280
↓ 1 callersMethodprepare_extra_step_kwargs
(self, generator, eta)
fancyvideo/pipelines/pipeline_animation_vae_3d.py:284
↓ 1 callersMethodprepare_latents
(self, batch_size, num_channels_latents, video_length, height, width, dtype, device, generator, latents=None)
fancyvideo/pipelines/pipeline_animation_vae_2d.py:312
↓ 1 callersMethodprepare_latents
(self, batch_size, num_channels_latents, video_length, height, width, dtype, device, generator, latents=None)
fancyvideo/pipelines/pipeline_animation_vae_3d.py:316
↓ 1 callersMethodsave_video
(self, frame_list, fps, dst_path)
fancyvideo/pipelines/fancyvideo_infer_pipeline.py:364
↓ 1 callersMethodset_attention_slice
r""" Enable sliced attention computation. When this option is enabled, the attention module will split the input tensor in slices, to
fancyvideo/models/unet.py:307
↓ 1 callersMethodt2v_process_one_prompt
(self, prompt, reference_image_path, seed=None, video_length=16, resolution=(512, 512), use_noise_scheduler_sn
fancyvideo/pipelines/fancyvideo_infer_pipeline.py:371
next →1–100 of 243, ranked by callers