ComfyUI Node: Scheduled PerpNeg CFGGuider (Inspire)

Scheduled PerpNeg CFGGuider (Inspire):

The ScheduledPerpNegCFGGuider __Inspire node is designed to enhance the sampling process in AI art generation by providing a guided approach that incorporates both positive and negative conditioning. This node allows for a scheduled transition of the classifier-free guidance (CFG) scale from an initial value to a final value, following a specified schedule type (linear, logarithmic, or exponential). By doing so, it helps in fine-tuning the influence of the conditioning inputs over the sampling iterations, leading to more controlled and refined outputs. The node is particularly useful for artists looking to achieve a balance between adhering to the desired conditioning and exploring creative variations.

Scheduled PerpNeg CFGGuider (Inspire) Input Parameters:

model

This parameter specifies the model to be used for the sampling process. It is a required input and ensures that the node has the necessary model architecture and weights to perform the guided sampling.

positive

This parameter represents the positive conditioning input, which guides the model towards desired features or characteristics in the generated output. It is a required input and plays a crucial role in shaping the final result according to the artist's intent.

negative

This parameter represents the negative conditioning input, which helps the model avoid undesired features or characteristics in the generated output. It is a required input and is essential for refining the output by suppressing unwanted elements.

empty_conditioning

This parameter provides an empty conditioning input, which can be used as a baseline or neutral reference during the sampling process. It is a required input and helps in balancing the influence of positive and negative conditioning.

neg_scale

This parameter controls the scaling factor for the negative conditioning input. It has a default value of 1.0, with a minimum value of 0.0 and a maximum value of 100.0, adjustable in steps of 0.01. Adjusting this parameter allows for fine-tuning the strength of the negative conditioning, impacting the suppression of undesired features.

sigmas

This parameter represents the noise levels (sigmas) used during the sampling process. It is a required input and is crucial for determining the amount of noise added at each step, influencing the diversity and quality of the generated output.

from_cfg

This parameter sets the initial value of the classifier-free guidance (CFG) scale. It has a default value of 6.5, with a minimum value of 0.0 and a maximum value of 100.0, adjustable in steps of 0.1. This parameter determines the starting influence of the conditioning inputs on the sampling process.

to_cfg

This parameter sets the final value of the classifier-free guidance (CFG) scale. It has a default value of 1.0, with a minimum value of 0.0 and a maximum value of 100.0, adjustable in steps of 0.1. This parameter determines the ending influence of the conditioning inputs on the sampling process.

schedule

This parameter specifies the type of schedule to be used for transitioning the CFG scale from the initial value to the final value. The available options are "linear", "log", and "exp", with "log" being the default. The choice of schedule affects how the influence of the conditioning inputs changes over the sampling iterations.

Scheduled PerpNeg CFGGuider (Inspire) Output Parameters:

GUIDER

This output parameter represents the configured guider object that incorporates the specified model, conditioning inputs, and scheduling parameters. It is used to guide the sampling process according to the provided settings.

SIGMAS

This output parameter returns the noise levels (sigmas) used during the sampling process. It is essential for understanding the noise profile applied at each step and its impact on the generated output.

Scheduled PerpNeg CFGGuider (Inspire) Usage Tips:

• Experiment with different schedule types (linear, log, exp) to see how they affect the transition of the CFG scale and the resulting output.
• Adjust the neg_scale parameter to fine-tune the suppression of undesired features, especially when dealing with complex conditioning inputs.
• Use the from_cfg and to_cfg parameters to control the starting and ending influence of the conditioning inputs, which can help in achieving a balance between adherence to the conditioning and creative exploration.

Scheduled PerpNeg CFGGuider (Inspire) Common Errors and Solutions:

"Model not specified"

• Explanation: The model parameter is missing or not provided.
• Solution: Ensure that a valid model is specified in the model parameter before executing the node.

"Invalid schedule type"

• Explanation: The schedule parameter has an invalid value.
• Solution: Check that the schedule parameter is set to one of the valid options: "linear", "log", or "exp".

"Negative scale out of range"

• Explanation: The neg_scale parameter is set to a value outside the allowed range.
• Solution: Adjust the neg_scale parameter to be within the range of 0.0 to 100.0.

"CFG scale out of range"

• Explanation: The from_cfg or to_cfg parameters are set to values outside the allowed range.
• Solution: Ensure that both from_cfg and to_cfg parameters are within the range of 0.0 to 100.0.