⬡ Pack Pipeline

A standalone node that creates a generation pipeline from an existing image, bypassing the initial text-to-image generation phase.

The Pack Pipeline is used to enable direct image upscaling or post-processing pipelines. It creates a UME_PIPELINE from raw components without executing a sampling block, making it ideal for pure image-to-image workflows like UltimateSD or SeedVR2 Upscale.

Inputs

Name	Type	Required	Default	Description
image	IMAGE	✅	—	The source image to pack into the pipeline
bundle	UME_BUNDLE	❌	—	Optional model bundle (Model, VAE, CLIP)
settings	UME_SETTINGS	❌	—	Optional generation settings
positive	POSITIVE	❌	—	Optional positive conditioning
negative	NEGATIVE	❌	—	Optional negative conditioning

Outputs

Name	Type	Description
gen_pipe	UME_PIPELINE	A valid pipeline context ready for post-processing nodes

How it Works

Instead of relying on a KSampler to generate an initial image, the Pack Pipeline takes an already existing image and wraps it in a GenerationContext. Any optional inputs provided (models, prompts, settings) are attached to the context.

This allows downstream nodes like ⬡ UltimateSD Upscale or ⬡ SeedVR2 Upscale to function directly on user-provided images without throwing errors about missing pipeline dependencies.