MCPcopy Create free account

hub / github.com/Stable-X/ComfyUI-Hi3DGen / functions

Functions333 in github.com/Stable-X/ComfyUI-Hi3DGen

⨍ Functions 333◇ Types & classes 71
↓ 92 callersMethodreshape
(self, *shape)
trellis/modules/sparse/basic.py:243
↓ 50 callersMethodreplace
(self, feats: torch.Tensor, coords: Optional[torch.Tensor] = None)
trellis/modules/sparse/basic.py:250
↓ 33 callersMethodto
(self, device: torch.device)
trellis/pipelines/base.py:69
↓ 21 callersMethodtype
(self, dtype)
trellis/modules/sparse/basic.py:210
↓ 20 callersMethodunbind
(self, dim: int)
trellis/modules/sparse/basic.py:247
↓ 14 callersMethodfloat
(self)
trellis/modules/sparse/basic.py:228
↓ 12 callersFunctionget_debug_mode
Get current debug mode status
trellis/backend_config.py:98
↓ 12 callersMethodto
(self, dtype: torch.dtype)
trellis/modules/sparse/basic.py:184
↓ 11 callersMethodload
Load model configuration and checkpoints
trellis_model_manager.py:58
↓ 9 callersFunctionget_attention_backend
Get current attention backend
trellis/backend_config.py:88
↓ 8 callersMethod__elemwise__
(self, other: Union[torch.Tensor, 'SparseTensor'], op: callable)
trellis/modules/sparse/basic.py:312
↓ 8 callersMethodcpu
(self)
trellis/pipelines/base.py:76
↓ 7 callersMethoddim
(self)
trellis/modules/sparse/basic.py:140
↓ 7 callersMethodget_spatial_cache
Get a spatial cache.
trellis/modules/sparse/basic.py:388
↓ 7 callersMethodregister_spatial_cache
Register a spatial cache. The spatial cache can be any thing you want to cache. The registery and retrieval of the cache i
trellis/modules/sparse/basic.py:377
↓ 6 callersFunctionget_available_backends
Return dict of available attention backends and their status
trellis/backend_config.py:71
↓ 6 callersMethodload_models
Load specific models back to GPU
trellis/pipelines/trellis_image_to_3d.py:256
↓ 5 callersMethoddetach
(self)
trellis/modules/sparse/basic.py:232
↓ 5 callersMethodfrom_pretrained
Load a pretrained model.
trellis/pipelines/base.py:22
↓ 5 callersFunctionget_sparse_backend
Get current sparse backend
trellis/backend_config.py:93
↓ 5 callersMethodunload_models
Unload specific models from GPU memory
trellis/pipelines/trellis_image_to_3d.py:247
↓ 4 callersMethod__init__
( self, channels: int, out_channels: Optional[int] = None, norm_type: Lite
trellis/models/sparse_structure_vae.py:23
↓ 4 callersMethod_inference_model
(self, model, x_t, t, cond=None, **kwargs)
trellis/pipelines/samplers/flow_euler.py:38
↓ 4 callersMethod_linear
(module: nn.Linear, x: Union[SparseTensor, torch.Tensor])
trellis/modules/sparse/attention/modules.py:78
↓ 4 callersFunctionget_spconv_algo
Get current spconv algorithm.
trellis/backend_config.py:13
↓ 4 callersMethodhalf
(self)
trellis/modules/sparse/basic.py:224
↓ 4 callersFunctionnorm_layer
Return a normalization layer.
trellis/models/sparse_structure_vae.py:10
↓ 4 callersFunctionscaled_dot_product_attention
Apply scaled dot product attention. Args: qkv (torch.Tensor): A [N, L, 3, H, C] tensor containing Qs, Ks, and Vs.
trellis/modules/attention/full_attn.py:57
↓ 3 callersMethod__cal_layout
(self, coords, batch_size)
trellis/modules/sparse/basic.py:130
↓ 3 callersMethod__cal_shape
(self, feats, coords)
trellis/modules/sparse/basic.py:124
↓ 3 callersMethod__init__
(self, channels: int, mlp_ratio: float = 4.0)
trellis/modules/transformer/blocks.py:50
↓ 3 callersMethod_linear_interp
Computes the location of zero-crossings on 'edges_x' using linear interpolation with 'edges_weight'.
trellis/representations/mesh/flexicube.py:194
↓ 3 callersMethoddevice
(self)
trellis/pipelines/base.py:60
↓ 3 callersMethodforward
(self, input: SparseTensor)
trellis/modules/sparse/norm.py:18
↓ 3 callersFunctionzero_module
Zero out the parameters of a module and return it.
trellis/modules/utils.py:35
↓ 2 callersMethod__init__
(self, dim: int, heads: int)
trellis/modules/attention/modules.py:9
↓ 2 callersMethod__init__
(self, factor: Union[int, Tuple[int, int, int], List[int]])
trellis/modules/sparse/spatial.py:64
↓ 2 callersMethod__init__
(self, channels: int, mlp_ratio: float = 4.0)
trellis/modules/sparse/transformer/blocks.py:12
↓ 2 callersMethod_fused_pre
(self, x: Union[SparseTensor, torch.Tensor], num_fused: int)
trellis/modules/sparse/attention/modules.py:91
↓ 2 callersMethod_init_image_cond_model
Initialize the image conditioning model.
trellis/pipelines/trellis_image_to_3d.py:91
↓ 2 callersMethod_reshape_chs
(x: Union[SparseTensor, torch.Tensor], shape: Tuple[int, ...])
trellis/modules/sparse/attention/modules.py:85
↓ 2 callersMethod_rotary_embedding
(self, x: torch.Tensor, phases: torch.Tensor)
trellis/modules/attention/modules.py:34
↓ 2 callersFunction_try_import_flash_attn
()
trellis/backend_config.py:40
↓ 2 callersFunction_try_import_sageattention
()
trellis/backend_config.py:47
↓ 2 callersFunction_try_import_xformers
()
trellis/backend_config.py:33
↓ 2 callersFunctioncreate_model_from_config
Helper function to create model from config
trellis/models/__init__.py:91
↓ 2 callersMethoddecode_slat
Decode the structured latent. Args: slat (sp.SparseTensor): The structured latent. formats (List[str]):
trellis/pipelines/trellis_image_to_3d.py:300
↓ 2 callersMethodforward
(self, x: torch.Tensor)
trellis/modules/norm.py:6
↓ 2 callersMethodforward
(self, input: SparseTensor)
trellis/modules/sparse/nonlinearity.py:14
↓ 2 callersFunctionget_available_sparse_backends
Return dict of available sparse backends and their status
trellis/backend_config.py:81
↓ 2 callersMethodget_checkpoint_path
Returns the full path to a checkpoint file.
trellis_model_manager.py:97
↓ 2 callersMethodget_cond
Get the conditioning information for the model. Args: image (Union[torch.Tensor, list[Image.Image]]): The image prom
trellis/pipelines/trellis_image_to_3d.py:191
↓ 2 callersFunctionget_defomed_verts
(v_pos : torch.Tensor, deform : torch.Tensor, res)
trellis/representations/mesh/utils_cube.py:59
↓ 2 callersFunctionget_dense_attrs
(coords : torch.Tensor, feats : torch.Tensor, res : int, sdf_init=True)
trellis/representations/mesh/utils_cube.py:50
↓ 2 callersMethodinject_sampler_multi_image
Inject a sampler with multiple images as condition. Args: sampler_name (str): The name of the sampler to inj
trellis/pipelines/trellis_image_to_3d.py:429
↓ 2 callersMethodnormalize_normals
(normals: torch.Tensor, eps: float = 1e-6)
stablex/pipeline_yoso.py:571
↓ 2 callersMethodpreprocess_image
Preprocess the input image.
trellis/pipelines/trellis_image_to_3d.py:123
↓ 2 callersMethodsample
Sample from a model.
trellis/pipelines/samplers/base.py:11
↓ 2 callersMethodsample
Generate samples from the model using Euler method. Args: model: The model to sample from. nois
trellis/pipelines/samplers/flow_euler.py:78
↓ 2 callersMethodsample_slat
Sample structured latent with the given conditioning. Args: cond (dict): The conditioning information.
trellis/pipelines/trellis_image_to_3d.py:339
↓ 2 callersMethodsample_sparse_structure
Sample sparse structures with the given conditioning. Args: cond (dict): The conditioning information.
trellis/pipelines/trellis_image_to_3d.py:264
↓ 2 callersFunctionsparse_batch_broadcast
Broadcast a 1D tensor to a sparse tensor along the batch dimension then perform an operation. Args: input (torch.Tensor): 1D
trellis/modules/sparse/basic.py:399
↓ 2 callersFunctionsparse_scaled_dot_product_attention
Apply scaled dot product attention to a sparse tensor. Args: qkv (SparseTensor): A [N, *, 3, H, C] sparse tensor containing Qs,
trellis/modules/sparse/attention/full_attn.py:41
↓ 2 callersMethodtorch_to_pil_batch
(self, images: Union[torch.Tensor, List[torch.Tensor]], masks: Optional[torch.Tenso
IF_Trellis.py:77
↓ 1 callersFunction__from_env
Initialize settings from environment variables
trellis/backend_config.py:103
↓ 1 callersFunction__from_env
()
trellis/modules/sparse/conv/__init__.py:8
↓ 1 callersFunction__getattr__
(name)
trellis/models/__init__.py:20
↓ 1 callersMethod__init__
(self, num_groups, num_channels, eps=1e-5, affine=True)
trellis/modules/sparse/norm.py:15
↓ 1 callersMethod__init__
(self, dim: int, heads: int)
trellis/modules/sparse/attention/modules.py:13
↓ 1 callersMethod__init__
(self, in_channels, out_channels, kernel_size, stride=1, dilation=1, bias=True, indice_key=None)
trellis/modules/sparse/conv/conv_torchsparse.py:12
↓ 1 callersMethod__init__
(self, in_channels, out_channels, kernel_size, stride=1, dilation=1, padding=None, bias=True, indice_key=None)
trellis/modules/sparse/conv/conv_spconv.py:13
↓ 1 callersMethod__init__
( self, channels: int, num_heads: int, mlp_ratio: float = 4.0, at
trellis/modules/sparse/transformer/modulated.py:14
↓ 1 callersMethod__init__
( self, channels: int, num_heads: int, mlp_ratio: float = 4.0, at
trellis/modules/transformer/modulated.py:13
↓ 1 callersMethod__init__
(self, hidden_size, frequency_embedding_size=256)
trellis/models/sparse_structure_flow.py:15
↓ 1 callersMethod__init__
( self, resolution: int, in_channels: int, model_channels: int, c
trellis/models/structured_latent_flow.py:69
↓ 1 callersMethod__init__
( self, resolution: int, model_channels: int, latent_channels: int,
trellis/models/structured_latent_vae/decoder_mesh.py:72
↓ 1 callersMethod__init__
(self, vertices, faces, vertex_attrs=None, res=64 )
trellis/representations/mesh/cube2mesh.py:9
↓ 1 callersMethod__merge_sparse_cache
(self, other: 'SparseTensor')
trellis/modules/sparse/basic.py:297
↓ 1 callersMethod_calc_layout
(self)
trellis/representations/mesh/cube2mesh.py:74
↓ 1 callersMethod_compute_reg_loss
Regularizer L_dev as in Equation 8
trellis/representations/mesh/flexicube.py:81
↓ 1 callersMethod_compute_vd
Computes the location of dual vertices as described in Section 4.2
trellis/representations/mesh/flexicube.py:221
↓ 1 callersFunction_dict_merge
Merge two dictionaries.
trellis/utils/general_utils.py:7
↓ 1 callersMethod_forward
(self, x: SparseTensor)
trellis/modules/sparse/transformer/blocks.py:65
↓ 1 callersMethod_forward
(self, x: SparseTensor, mod: torch.Tensor, context: torch.Tensor)
trellis/modules/sparse/transformer/blocks.py:135
↓ 1 callersMethod_forward
(self, x: SparseTensor, mod: torch.Tensor)
trellis/modules/sparse/transformer/modulated.py:57
↓ 1 callersMethod_forward
(self, x: SparseTensor, mod: torch.Tensor, context: torch.Tensor)
trellis/modules/sparse/transformer/modulated.py:142
↓ 1 callersMethod_forward
(self, x: torch.Tensor)
trellis/modules/transformer/blocks.py:99
↓ 1 callersMethod_forward
(self, x: torch.Tensor, context: torch.Tensor)
trellis/modules/transformer/blocks.py:165
↓ 1 callersMethod_forward
(self, x: torch.Tensor, mod: torch.Tensor)
trellis/modules/transformer/modulated.py:52
↓ 1 callersMethod_forward
(self, x: torch.Tensor, mod: torch.Tensor, context: torch.Tensor)
trellis/modules/transformer/modulated.py:132
↓ 1 callersMethod_get_case_id
Obtains the ID of topology cases based on cell corner occupancy. This function resolves the ambiguity in the Dual Marching Cubes (
trellis/representations/mesh/flexicube.py:115
↓ 1 callersMethod_get_model_prediction
(self, model, x_t, t, cond=None, **kwargs)
trellis/pipelines/samplers/flow_euler.py:42
↓ 1 callersMethod_get_phases
(self, indices: torch.Tensor)
trellis/modules/attention/modules.py:28
↓ 1 callersMethod_identify_surf_cubes
Identifies grid cubes that intersect with the underlying surface by checking if the signs at all corners are not identical.
trellis/representations/mesh/flexicube.py:183
↓ 1 callersMethod_identify_surf_edges
Identifies grid edges that intersect with the underlying surface by checking for opposite signs. As each edge can be shared by mul
trellis/representations/mesh/flexicube.py:158
↓ 1 callersMethod_load_config
Load model configuration from pipeline.json
trellis_model_manager.py:110
↓ 1 callersFunction_naive_sdpa
Naive implementation of scaled dot product attention.
trellis/modules/attention/full_attn.py:41
↓ 1 callersMethod_normalize_weights
Normalizes the given weights to be non-negative. If input weights are None, it creates and returns a set of weights of ones.
trellis/representations/mesh/flexicube.py:91
↓ 1 callersMethod_optimize_pipeline
Apply typical optimizations, half-precision, etc.
IF_TrellisCheckpointLoader.py:107
↓ 1 callersMethod_rope
(self, qkv: SparseTensor)
trellis/modules/sparse/attention/modules.py:99
next →1–100 of 333, ranked by callers