ComfyUI Node: Lying Sigma Sampler

Lying Sigma Sampler:

The LyingSigmaSampler is a specialized node designed to manipulate the sigma values passed to a model during the sampling process. Its primary purpose is to subtly adjust these sigma values, which are crucial in controlling the noise level during the generation of images or other outputs. By applying a "dishonesty factor," this node can either increase or decrease the sigma values within a specified range, allowing for more nuanced control over the sampling process. This can be particularly beneficial in fine-tuning the output quality, as it enables you to experiment with different levels of noise reduction or enhancement, potentially leading to more detailed or creatively altered results. The LyingSigmaSampler is an essential tool for AI artists looking to explore the boundaries of their models' capabilities by tweaking the underlying sampling mechanics.

Lying Sigma Sampler Input Parameters:

sampler

The sampler parameter is a required input that specifies the base sampling method to be used. It acts as the foundation upon which the LyingSigmaSampler applies its adjustments. This parameter is crucial as it determines the initial behavior and characteristics of the sampling process before any sigma modifications are applied.

dishonesty_factor

The dishonesty_factor is a required float parameter that dictates the degree to which the sigma values are adjusted. With a default value of -0.05, it reduces the sigma by 5%, effectively decreasing the noise level. The range for this parameter is from -0.999 to a positive value, allowing for both reduction and amplification of sigma values. This parameter is key to controlling the extent of the modification applied to the sigma values, thereby influencing the final output's detail and quality.

start_percent

The start_percent is an optional float parameter that defines the starting point of the sigma adjustment as a percentage of the total sampling process. It has a default value of 0.1, with a range from 0.0 to 1.0. This parameter allows you to specify when the sigma adjustments should begin, providing control over the timing of the effect within the sampling sequence.

end_percent

The end_percent is an optional float parameter that sets the endpoint of the sigma adjustment as a percentage of the total sampling process. With a default value of 0.9, it ranges from 0.0 to 1.0. This parameter determines when the sigma adjustments should cease, allowing for precise control over the duration of the effect during the sampling process.

Lying Sigma Sampler Output Parameters:

SAMPLER

The output of the LyingSigmaSampler is a modified SAMPLER object. This output retains the original sampling method's characteristics but with adjusted sigma values as specified by the input parameters. The modified sampler is crucial for generating outputs with the desired level of detail and noise, as it incorporates the nuanced sigma adjustments made by the LyingSigmaSampler.

Lying Sigma Sampler Usage Tips:

• Experiment with different dishonesty_factor values to find the optimal balance between noise reduction and detail enhancement for your specific project.
• Use the start_percent and end_percent parameters to control the timing of sigma adjustments, which can be particularly useful for achieving gradual transitions in noise levels throughout the sampling process.

Lying Sigma Sampler Common Errors and Solutions:

"Invalid dishonesty_factor value"

• Explanation: The dishonesty_factor value provided is outside the acceptable range of -0.999 to a positive value.
• Solution: Ensure that the dishonesty_factor is set within the specified range to avoid this error.

"Start percent greater than end percent"

• Explanation: The start_percent value is set higher than the end_percent, causing a logical error in the timing of sigma adjustments.
• Solution: Adjust the start_percent and end_percent values to ensure that the start is less than or equal to the end, maintaining a logical sequence for sigma adjustments.