Video Helper Suite πŸŽ₯πŸ…₯πŸ…—πŸ…’/batched nodesgenerated

VAE Decode Batched πŸŽ₯πŸ…₯πŸ…—πŸ…’(VAEDecodeBatched)

This node technically receives a batch containing multiple latent samples and decodes them using the specified VAE model.

VAE Decode Batched

samples
vae
IMAGE
per_batch
16
VideoHelperSuite

The VAE Decode Batched πŸŽ₯πŸ…₯πŸ…—πŸ…’ node technically receives a batch containing multiple latent samples and decodes them using the specified VAE model.

The VAE Decode Batched node is a specialized tool for batch-decoding latent space representations into visual images. Its core function is to efficiently convert a set of compressed "latent" data back into standard image format through a Variational Autoencoder model.

Node Functionality

This node technically receives a batch containing multiple latent samples and decodes them using the specified VAE model. Its core mechanism involves splitting the entire batch decoding task into multiple smaller sub-batches for sequential processing, thereby optimizing memory usage. It supports standard LATENT input and IMAGE output formats, internally completing the entire decoding process through iterative loops and concatenation operations.

Node Parameter Description - VAE Decode Batched πŸŽ₯πŸ…₯πŸ…—πŸ…’

Connection Inputs (Inputs)

Parameter NameData TypeRequiredDefault ValueValue Range/OptionsDescription
samplesLATENTYes--samples (input parameter): Input containing the latent representation data to be decoded. Type is LATENT.
vaeVAEYes--vae (input parameter): The Variational Autoencoder model used to perform the decoding operation. Type is VAE.

Control Parameters (Parameters)

Parameter NameData TypeRequiredDefault ValueValue Range/OptionsDescription
per_batchINTYes16-per_batch (input parameter): Controls the sub-batch size for each decoding operation. Type is INT, default value is 16, minimum value is 1.

Output

Parameter NameData TypeDescription
IMAGEIMAGEIMAGE (output parameter): The generated image data after decoding. Type is IMAGE.

Usage Scenarios

This node is very useful in workflows for video generation or image sequence processing. For example, after using a generative model (such as Stable Diffusion) to produce a series of latent representations for images in batch, you can connect this node and specify a suitable VAE model to decode these latent representations into a complete image sequence at once for subsequent viewing, saving, or further editing.

Notes

It is important to note that the value of the per_batch parameter affects the memory usage and speed of the decoding process. Setting a value too large may lead to insufficient VRAM, while setting it too small may reduce processing efficiency. The minimum value for this parameter is 1.

The VAE Decode Batched πŸŽ₯πŸ…₯πŸ…—πŸ…’ node is from the ComfyUI-VideoHelperSuite node package.

Comments

Sign in with GitHub to join the discussion.

Loading comments…