unreal.MLDeformerAsset¶
- class unreal.MLDeformerAsset(outer=None, name='None')¶
Bases:
unreal.ObjectThe machine learning deformer asset. At runtime this contains only the data needed to run the neural network inference. In the editor it contains the skeletal mesh and geometry cache that are required to calculate vertex position deltas.
C++ Source:
Plugin: MLDeformer
Module: MLDeformer
File: MLDeformerAsset.h
Editor Properties: (see get_editor_property/set_editor_property)
activation_function(ActivationFunction): [Read-Write] Activation Function: The activation function to use in the neural network.alignment_transform(Transform): [Read-Write] Alignment Transform: The transform that aligns the Geometry Cache to the SkeletalMesh. This will mostly apply some scale and a rotation, but no translation.anim_sequence(AnimSequence): [Read-Write] Anim Sequence: The animation sequence to apply to the base mesh. This has to match the animation of the target mesh’s geometry cache. Internally we force the Interpolation property for this motion to be “Step”.batch_size(int32): [Read-Write] Batch Size: The number of frames per batch when training the model.bone_include_list(Array(BoneReference)): [Read-Write] Bone Include List: The bones to include during training. When none are provided, all bones of the Skeleton will be included.cache_size_in_megabytes(int32): [Read-Write] Cache Size in Megabytes: The maximum allowed size of the training cache in memory, in megabytes. So a value of 1024 would be one gigabyte. The larger the cache size the faster the training.curve_include_list(Array(CurveReference)): [Read-Write] Curve Include List: The curves to include during training. When none are provided, all curves of the Skeleton will be included.decay_function(DecayFunction): [Read-Write] Decay Function: The decay function to adapt the learning rate.decay_rate(float): [Read-Write] Decay Rate: The decay rate to apply to the learning rate once non-decay epochs have been reached. Higher values give less decay.delta_cutoff_length(float): [Read-Write] Delta Cutoff Length: Sometimes there can be some vertices that cause some issues that cause deltas to be very long. We can ignore these deltas by setting a cutoff value. Deltas that are longer than the cutoff value (in units), will be ignored and set to zero length.device_type(DeviceType): [Read-Write] Device Type: The loss function to use during model training.epochs(int32): [Read-Write] Epochs: The number of epochs to process without any decay.epochs_with_decay(int32): [Read-Write] Epochs with Decay: The number of epochs to process that include a decay.geometry_cache(GeometryCache): [Read-Write] Geometry Cache: The geometry cache that represents the complex mesh deformations.learning_rate(float): [Read-Write] Learning Rate: The learning rate used during the model training.loss_function(LossFunction): [Read-Write] Loss Function: The loss function to use during model training.max_training_frames(int32): [Read-Write] Max Training Frames: The maximum numer of training frames (samples) to train on. Use this to train on a sub-section of your full training data.noise_amount(float): [Read-Write] Noise Amount: The percentage of noise to add. Adding some noise can make the system generalize better. Generally you want a low percentage, such as 0.5. Currently using noise will disable the caching system, which will slow down training a lot. // A percentage.num_hidden_layers(int32): [Read-Write] Num Hidden Layers: The number of hidden layers that the neural network model will have.nHigher numbers will slow down performance but can deal with more complex deformations.num_neurons_per_layer(int32): [Read-Write] Num Neurons Per Layer: The number of units/neurons per hidden layer. Higher numbers will slow down performance but allow for more complex mesh deformations.shrinkage_speed(float): [Read-Write] Shrinkage Speed: Shrinkage speed. Only if the shrinkage loss is used.shrinkage_threshold(float): [Read-Write] Shrinkage Threshold: Shrinkage threshold. Only if the shrinkage loss is used.skeletal_mesh(SkeletalMesh): [Read-Write] Skeletal Mesh: The skeletal mesh that represents the linear skinned mesh.training_inputs(TrainingInputs): [Read-Write] Training Inputs: Describes what inputs we should train the neural network on.
- property activation_function¶
[Read-Write] Activation Function: The activation function to use in the neural network.
- Type
- property alignment_transform¶
[Read-Write] Alignment Transform: The transform that aligns the Geometry Cache to the SkeletalMesh. This will mostly apply some scale and a rotation, but no translation.
- Type
- property batch_size¶
[Read-Write] Batch Size: The number of frames per batch when training the model.
- Type
(int32)
- property cache_size_in_megabytes¶
[Read-Write] Cache Size in Megabytes: The maximum allowed size of the training cache in memory, in megabytes. So a value of 1024 would be one gigabyte. The larger the cache size the faster the training.
- Type
(int32)
- property decay_function¶
[Read-Write] Decay Function: The decay function to adapt the learning rate.
- Type
- property decay_rate¶
[Read-Write] Decay Rate: The decay rate to apply to the learning rate once non-decay epochs have been reached. Higher values give less decay.
- Type
(float)
- property delta_cutoff_length¶
[Read-Write] Delta Cutoff Length: Sometimes there can be some vertices that cause some issues that cause deltas to be very long. We can ignore these deltas by setting a cutoff value. Deltas that are longer than the cutoff value (in units), will be ignored and set to zero length.
- Type
(float)
- property device_type¶
[Read-Write] Device Type: The loss function to use during model training.
- Type
- property epochs¶
[Read-Write] Epochs: The number of epochs to process without any decay.
- Type
(int32)
- property epochs_with_decay¶
[Read-Write] Epochs with Decay: The number of epochs to process that include a decay.
- Type
(int32)
- property learning_rate¶
[Read-Write] Learning Rate: The learning rate used during the model training.
- Type
(float)
- property loss_function¶
[Read-Write] Loss Function: The loss function to use during model training.
- Type
- property max_training_frames¶
[Read-Write] Max Training Frames: The maximum numer of training frames (samples) to train on. Use this to train on a sub-section of your full training data.
- Type
(int32)
- property noise_amount¶
[Read-Write] Noise Amount: The percentage of noise to add. Adding some noise can make the system generalize better. Generally you want a low percentage, such as 0.5. Currently using noise will disable the caching system, which will slow down training a lot. // A percentage.
- Type
(float)
[Read-Write] Num Hidden Layers: The number of hidden layers that the neural network model will have.nHigher numbers will slow down performance but can deal with more complex deformations.
- Type
(int32)
- property num_neurons_per_layer¶
[Read-Write] Num Neurons Per Layer: The number of units/neurons per hidden layer. Higher numbers will slow down performance but allow for more complex mesh deformations.
- Type
(int32)