ComfyUI Node: Manual Sampler

Manual Sampler:

The ManualSampler node is designed to provide a flexible and user-friendly interface for customizing k_diffusion samplers within the ComfyUI framework. Its primary purpose is to allow you to adjust key parameters such as eta and s_noise, which influence the behavior of the sampling process, without the need to delve into complex integrator code. This node is particularly beneficial for those who wish to fine-tune the sampling process to achieve specific artistic effects while ensuring that the output remains stable and sensible. By wrapping existing samplers and offering a preset system, ManualSampler ensures that you can save and reuse configurations easily, enhancing workflow efficiency. The node's design philosophy emphasizes safety and reliability, preventing issues like NaN (Not a Number) errors or divergence that can occur with improperly configured samplers.

Manual Sampler Input Parameters:

base_sampler

The base_sampler parameter specifies the foundational sampler that the ManualSampler will wrap and customize. It is crucial as it determines the core algorithm used for sampling. The choice of base sampler can significantly impact the style and quality of the generated output, as different samplers have unique characteristics and strengths.

stages

The stages parameter allows you to define the number of stages or steps in the sampling process. This can affect the granularity and detail of the output, with more stages potentially leading to finer results. The parameter accepts integer values, and its configuration should align with the desired level of detail in the final output.

eta_override

The eta_override parameter provides a way to adjust the eta value, which influences the randomness and diversity of the sampling process. A higher eta can lead to more varied outputs, while a lower value may produce more consistent results. The parameter accepts a float value, with a default setting that can be overridden as needed.

s_noise

The s_noise parameter controls the noise level in the sampling process, affecting the texture and smoothness of the output. A higher noise level can introduce more texture, while a lower level may result in smoother images. This parameter accepts a float value, allowing for precise adjustments to achieve the desired artistic effect.

preset_data

The preset_data parameter is used to load or save preset configurations for the sampler. This feature is particularly useful for maintaining consistency across projects or sharing configurations with others. The parameter accepts a string that represents the preset data, enabling easy management of sampler settings.

Manual Sampler Output Parameters:

SAMPLER

The SAMPLER output parameter represents the customized sampler instance created by the ManualSampler node. This output is crucial as it encapsulates all the adjustments made to the base sampler, including any overrides and preset configurations. The resulting sampler can be used in subsequent nodes or processes within ComfyUI to generate images or other outputs, reflecting the specific artistic choices made during configuration.

Manual Sampler Usage Tips:

• Experiment with different base_sampler options to find the one that best suits your artistic style or project requirements. Each sampler has unique characteristics that can influence the final output.
• Utilize the preset_data feature to save your favorite configurations. This can save time and ensure consistency across different projects or when collaborating with others.

Manual Sampler Common Errors and Solutions:

" Manual Sampler: base sampler '<base_name>' not found in comfy.samplers.k_diffusion_sampling"

• Explanation: This error occurs when the specified base_sampler is not recognized or available in the system.
• Solution: Ensure that the base_sampler name is correctly spelled and that it is a valid sampler within the ComfyUI framework. Check for any updates or changes in the available samplers.

"Invalid parameter value for 'eta_override' or 's_noise'"

• Explanation: This error indicates that the provided value for eta_override or s_noise is outside the acceptable range or format.
• Solution: Verify that the values for eta_override and s_noise are within the specified range and are of the correct data type (float). Adjust the values accordingly to meet the node's requirements.