EasyControlnet (Advanced)(controlnetAdvanced)

Description

EasyControlnet (Advanced) applies a ControlNet to the conditioning contained in an EasyUse PIPE_LINE. It is an advanced variant of EasyControlnet that also exposes scale_soft_weights, which adjusts the soft-weight scaling of the loaded ControlNet. The image input provides the ControlNet hint image, such as a Canny edge map, depth map, or pose skeleton.

Inputs

pipe

A PIPE_LINE containing the model, clip, VAE, positive conditioning, negative conditioning, and latent data. The node reads the positive and negative conditioning from this pipe and writes the ControlNet-applied conditionings back into the returned pipe.

image

A required IMAGE tensor used as the ControlNet hint. This should be the preprocessed control image matching the type of the selected ControlNet model.

control_net_name

A required combo input selecting the ControlNet model from the available files in ComfyUI's controlnet folder.

control_net

An optional CONTROL_NET input. When supplied, this ControlNet object is used directly instead of loading one from control_net_name. This is useful for passing in a ControlNet that was already loaded or modified elsewhere.

strength

A float, default 1.0, range 0.0 to 10.0, step 0.01. Controls the overall strength of the ControlNet application. A value of 0 effectively disables the ControlNet and leaves the conditioning unchanged.

start_percent

A float, default 0.0, range 0.0 to 1.0, step 0.001. Defines the denoising fraction at which the ControlNet starts influencing the generation. 0.0 means the very beginning of sampling.

end_percent

A float, default 1.0, range 0.0 to 1.0, step 0.001. Defines the denoising fraction at which the ControlNet stops influencing the generation. 1.0 means the end of sampling.

scale_soft_weights

A float, default 1.0, range 0.0 to 1.0, step 0.001. Scales the soft weighting applied to the ControlNet when it is loaded from control_net_name. A value of 1.0 uses the standard soft-weight behavior; lower values attenuate it. If you are not intentionally tuning this, leave it at 1.0.

Outputs

• PIPE_LINE: the input pipeline with its conditioning updated to include the applied ControlNet.
• CONDITIONING: the positive conditioning after ControlNet application.
• CONDITIONING: the negative conditioning after ControlNet application.

The two CONDITIONING outputs match the positive and negative conditioning values stored in the returned pipeline, so either the pipe or the individual conditionings can be wired onward.

Usage Notes

• start_percent and end_percent are fractions of the full denoising schedule, not absolute step counts. For example, start_percent=0.2 and end_percent=0.8 applies the ControlNet only during the middle 60% of sampling.
• strength can go above 1.0, making it possible to push a weak ControlNet further than the normal 0.0–1.0 range.
• Set strength to 0 if you want the pipe's conditioning to pass through without ControlNet influence.
• The image should be the preprocessed hint expected by the chosen ControlNet, such as a Canny edge map, depth map, or OpenPose skeleton.
• If you connect a ControlNet to the optional control_net input, the node uses that object directly and does not need to load from control_net_name.