ComfyUI Node: Settings (smZ)

Settings (smZ):

The smZ_Settings node is designed to enhance the flexibility and control of your AI art generation process by providing a range of customizable settings. This node allows you to fine-tune various parameters related to noise, sampling, and optimization, ensuring that you can achieve the desired artistic effects with greater precision. By integrating these settings, you can reproduce specific image results, optimize performance, and manage memory usage more effectively. The primary goal of the smZ_Settings node is to offer you a comprehensive set of tools to refine your AI-generated art, making it a valuable addition to your creative workflow.

Settings (smZ) Input Parameters:

info_eta_noise_seed_delta

This parameter provides information about the eta noise seed delta, which does not improve the image quality but produces different results for ancestral samplers. It is only useful for reproducing specific images. The description is multiline and serves as a placeholder to explain its purpose.

ENSD

This integer parameter sets the eta noise seed delta. It allows you to control the variability in the results produced by ancestral samplers. The default value is set to the current eta noise seed delta, with a minimum value of 0 and a maximum value of 0xffffffffffffffff, and it increments in steps of 1.

info_sgm_noise_multiplier

This parameter provides information about the SGM noise multiplier, which matches the initial noise to the official SDXL implementation. It is only useful for reproducing specific images. The description is multiline and serves as a placeholder to explain its purpose.

sgm_noise_multiplier

This boolean parameter enables or disables the SGM noise multiplier. When enabled, it matches the initial noise to the official SDXL implementation, which can be useful for reproducing specific images. The default value is set to the current SGM noise multiplier setting.

info_upcast_sampling

This parameter provides information about upcast sampling, which has no effect when --force-fp32 is used. It usually produces similar results to --force-fp32 with better performance while using less memory. The description is multiline and serves as a placeholder to explain its purpose.

upcast_sampling

This boolean parameter enables or disables upcast sampling. When enabled, it usually produces similar results to --force-fp32 with better performance and less memory usage. The default value is set to the current upcast sampling setting.

info_NGMS

This parameter provides information about the Negative Guidance minimum sigma (NGMS), which skips the negative prompt for some steps when the image is almost ready. A value of 0 disables this feature, while higher values result in faster processing. It is only applicable for CFGDenoiser. The description is multiline and serves as a placeholder to explain its purpose.

NGMS

This float parameter sets the Negative Guidance minimum sigma. It allows you to control the skipping of the negative prompt for some steps when the image is almost ready. The default value is set to the current minimum sigma, with a minimum value of 0.0, a maximum value of 15.0, and it increments in steps of 0.01.

Settings (smZ) Output Parameters:

smZ_opts

This output parameter provides a dictionary of the settings configured in the smZ_Settings node. It includes all the customized parameters such as eta noise seed delta, SGM noise multiplier, upcast sampling, and Negative Guidance minimum sigma. This dictionary is used to apply the configured settings to the model and sampler functions, ensuring that the desired effects are achieved in the generated images.

Settings (smZ) Usage Tips:

• To reproduce specific images, adjust the ENSD parameter to match the eta noise seed delta used in the original generation process.
• Enable the sgm_noise_multiplier to match the initial noise to the official SDXL implementation, which can be useful for reproducing specific images.
• Use the upcast_sampling parameter to achieve similar results to --force-fp32 with better performance and less memory usage.
• Adjust the NGMS parameter to control the skipping of the negative prompt for some steps when the image is almost ready, which can speed up the processing time.

Settings (smZ) Common Errors and Solutions:

"Invalid value for ENSD"

• Explanation: The value provided for the ENSD parameter is outside the acceptable range.
• Solution: Ensure that the value for ENSD is between 0 and 0xffffffffffffffff and increments in steps of 1.

"Invalid value for NGMS"

• Explanation: The value provided for the NGMS parameter is outside the acceptable range.
• Solution: Ensure that the value for NGMS is between 0.0 and 15.0 and increments in steps of 0.01.

"Missing smZ_opts in model_options"

• Explanation: The smZ_opts dictionary is not found in the model_options.
• Solution: Ensure that the smZ_Settings node is properly configured and that the smZ_opts dictionary is included in the model_options.

"Sampler function not found"

• Explanation: The sampler function is not found or not properly configured.
• Solution: Ensure that the sampler function is correctly set up and that the smZ_Settings node is properly integrated into the workflow.