Houdini 22.0 Machine Learning Trainable ML Solutions

Image nodes for trainable solutions

On this page

ML Computer Vision
ML GSplats
Neural Cellular Automata

Name	Type	Description
ML Train Style Transfer TOP	Model training	Trains a machine learning model for doing style transfer between two classes of images. The resulting model can be used with ONNX Interface SOP or ONNX Inference COP .
ML Train OIDN TOP	Model training	A wrapper around the OIDN training script used to train an OIDN denoising filter model on preprocessed training and validation data sets. Generally, you should use this node with Preprocess OIDN.
ML Preprocess OIDN TOP	Model training	A wrapper around the OIDN preprocessing script used to preprocess training and validation data sets compliant with the OIDN data set naming scheme.
PDG ML Training Monitor Panel		Can inspect the training progress and plots of the ML Train Style Transfer TOP and ML Train Regression TOP .

ML Computer Vision ¶

The ML Computer Vision set of nodes can train an ML model that detects and masks objects in an image sequence, using a small number of example images.

Name	Type	Description
ML Preprocess Computer Vision TOP	Model training	Prepares image data sets for computer vision training tasks.
ML Train Computer Vision TOP	Model training	Trains an object detection, masking, or keypoint tracking model using a data set prepared by the ML Preprocess Computer Vision TOP.
ML Computer Vision Inference COP	Inference	Applies an ONNX model trained using the ML Train Computer Vision TOP to detect and mask objects in an input image.

ML GSplats ¶

The set of GSplat nodes and recipes allows the creation of synthetic gaussian splats with support for custom AOVs. Karma generates the training data before preprocessing it into a COLMAP data set using ML Preprocess GSplats TOP. The training uses ML Train GSplats TOP.

Name	Type	Description
ML Preprocess GSplats TOP	Model training	Preprocesses rendered EXR images, cameras, and a point cloud into a COLMAP-like data set for Gaussian Splat training.
ML Train GSplats TOP	Model training	Trains a 3D Gaussian Splatting model from a set of posed images.
ML Train GSplats from Karma LOP (Recipe)	Inference \| This recipe automates the data generation and training process to create synthetic Gaussian Splats.

Neural Cellular Automata ¶

The NCA toolset allows you to train an ML based cellular automata system (ex. Reaction Diffusion) that can learn pattern based rules based on a given target image.

Name	Type	Description
ML Train Neural Cellular Automata TOP	Model training	Trains a Neural Cellular Automata model to synthesize tileable pattern based textures from a single target image.
ML Train Neural Cellular Automata COP (Recipe)	Inference	This recipe integrates the training process of neural cellular automata (NCA) models directly into COPs.
Neural Cellular Automata Core COP	Inference	Runs a Neural Cellular Automata (NCA) model on an input cell state, evolving it over a number of iterations. The model uses a learned neural network to simulate cellular automata rules that produce complex patterns and textures.
Neural Cellular Automata Decode COP	Inference	Takes the cell state output from a Neural Cellular Automata Core COP node and decodes it into a visible pattern using a decoder neural network. The cell state produced by the core node is a high-dimensional latent representation, so this node is needed to convert it into an RGB output.
Neural Cellular Automata (Recipe)	Inference	Drops down the NCA Core and Decode COP nodes to provide a convenient starting point.
Neural Cellular Automata Block (Recipe)	Inference	Configures the NCA Core and Decode COPs to be used inside of a COP block to simulate NCA updates over time.

Image nodes for trainable solutions

ML Computer Vision ¶

ML GSplats ¶

Neural Cellular Automata ¶

Trainable ML Solutions

ML recipes ¶

ML workflows ¶

ML nodes in other areas ¶