Purpose

StyleGAN runs an ONNX image generator. The model determines what kinds of pictures it can produce. Its runner configuration supplies the input controls, latent handling and image output.

The included Voro Generator Starter makes procedural 64 by 64 images. To generate a trained subject such as faces, select a compatible trained generator and its matching configuration.

Use

Place StyleGAN under Generate in the Image view. Set Your ONNX Export and Runner Config to a matching pair.

With the starter configuration, choose Seed and Seed B, then move Blend between 0 and 1. Use that small setup to check the image output before switching models.

Inputs and outputs

• The starter configuration generates its input from parameters and has no input wire.
• out is the generated image.
• A configuration that accepts an external latent can take a compatible Latent Seed or Latent Compose output.

The loaded configuration determines which ports are present. The base table below shows an in port that the starter does not use.

Controls

• Your ONNX Export selects the generator.
• Runner Config defines how to prepare its inputs and publish its output. Reload Config applies edits.
• The starter adds Seed, Seed B and Blend.
• Truncation applies when the selected export exposes it.

Notes

An ONNX file needs a configuration that matches its inputs and outputs. Check that pairing before treating every StyleGAN2 export as interchangeable. The starter supplies a procedural example without trained image weights.

Ports

Inputs

• in image
  1 x 512 x 4

Outputs

• out image
  RGBA

Parameters

StyleGAN

Your ONNX Export (Modelpath) Str op('stylegan').par.Modelpath

Default:

"" (Empty String)

Runner Config (Configpath) Str op('stylegan').par.Configpath

Default:

"" (Empty String)

Reload Config (Reload) Pulse op('stylegan').par.Reload

Default:

None