ComfyUI Node: KSampler Sequence

KSampler Sequence:

KSamplerSeq is a powerful node designed to facilitate the generation of image sequences using a k-sampling technique. This node is particularly useful for AI artists who want to create smooth transitions and animations by iteratively refining latent images through multiple steps. The primary goal of KSamplerSeq is to provide a flexible and efficient way to generate high-quality image sequences by leveraging various sampling methods, conditioning sequences, and interpolation techniques. By adjusting parameters such as denoising levels, latent interpolation, and conditioning strength, you can achieve a wide range of visual effects and styles, making it an essential tool for creative projects that require dynamic and evolving imagery.

KSampler Sequence Input Parameters:

model

This parameter specifies the model to be used for the sampling process. It is essential as it determines the underlying architecture and capabilities of the sampling process.

seed

The seed parameter is an integer that initializes the random number generator, ensuring reproducibility of the results. The default value is 0, with a minimum of 0 and a maximum of 0xffffffffffffffff.

seed_mode_seq

This parameter defines the mode of seed variation across the sequence. Options include "increment", "decrement", "random", and "fixed". Each mode affects how the seed changes from one step to the next, influencing the variability and consistency of the generated images.

alternate_values

A boolean parameter that, when set to true, alternates certain values during the sampling process. This can introduce variability and potentially more interesting results. The default value is true.

steps

This integer parameter determines the number of steps for the sampling process. More steps generally lead to higher quality images but require more computation time. The default value is 20, with a minimum of 1 and a maximum of 10000.

cfg

The cfg parameter is a float that controls the classifier-free guidance scale. It influences the strength of the guidance applied during sampling. The default value is 8.0, with a range from 0.0 to 100.0, adjustable in steps of 0.5.

sampler_name

This parameter specifies the name of the sampler to be used. It is selected from the available samplers in comfy.samplers.KSampler.SAMPLERS.

scheduler

The scheduler parameter determines the scheduling strategy for the sampling process. It is chosen from the available schedulers in comfy.samplers.KSampler.SCHEDULERS.

sequence_loop_count

An integer parameter that sets the number of loops for the sequence generation. More loops can create longer and more complex sequences. The default value is 20, with a minimum of 1 and a maximum of 1024.

positive_seq

This parameter provides the positive conditioning sequence, which guides the sampling process towards desired features and characteristics.

negative_seq

This parameter provides the negative conditioning sequence, which guides the sampling process away from undesired features and characteristics.

use_conditioning_slerp

A boolean parameter that, when enabled, uses spherical linear interpolation (slerp) for conditioning sequences. This can create smoother transitions between conditioning states. The default value is false.

cond_slerp_strength

A float parameter that controls the strength of the slerp interpolation for conditioning sequences. The default value is 0.5, with a range from 0.0 to 1.0, adjustable in steps of 0.001.

latent_image

This parameter provides the initial latent image to be used as the starting point for the sampling process.

use_latent_interpolation

A boolean parameter that, when enabled, uses interpolation techniques for latent images. This can create smoother transitions between latent states. The default value is false.

latent_interpolation_mode

This parameter specifies the mode of latent interpolation, with options including "Blend", "Slerp", and "Cosine Interp". Each mode affects how the latent images are interpolated during the sequence.

latent_interp_strength

A float parameter that controls the strength of the latent interpolation. The default value is 0.5, with a range from 0.0 to 1.0, adjustable in steps of 0.001.

denoise_start

A float parameter that sets the initial denoising level. Higher values result in more denoising. The default value is 1.0, with a range from 0.0 to 1.0, adjustable in steps of 0.01.

denoise_seq

A float parameter that sets the denoising level for the sequence. It influences the amount of noise reduction applied during each step. The default value is 0.5, with a range from 0.0 to 1.0, adjustable in steps of 0.01.

unsample_latents

A boolean parameter that, when enabled, allows for the unsampling of latent images. This can introduce additional variability and complexity to the generated sequences. The default value is false.

alternate_mode

A boolean parameter that, when enabled, alternates certain modes during the sampling process. This can create more dynamic and varied results. The default value is false.

inject_noise

A boolean parameter that, when enabled, injects noise into the sampling process. This can add texture and complexity to the generated images. The default value is true.

noise_strength

A float parameter that controls the strength of the injected noise. The default value is 0.1, with a range from 0.0 to 1.0.

denoise_sine

A boolean parameter that, when enabled, applies a sine function to the denoising process. This can create more natural and organic noise patterns. The default value is true.

denoise_max

A float parameter that sets the maximum denoising level. The default value is 0.9, with a range from 0.0 to 1.0.

seed_keying

A boolean parameter that, when enabled, uses seed keying to influence the sampling process. This can create more consistent and predictable results. The default value is true.

seed_keying_mode

This parameter specifies the mode of seed keying, with options including "sine". Each mode affects how the seed keying is applied during the sampling process.

seed_divisor

An integer parameter that sets the divisor for seed keying. This influences the frequency and pattern of seed changes. The default value is 4.

KSampler Sequence Output Parameters:

samples

The samples parameter provides the final generated image sequences. These sequences are the result of the iterative sampling process, incorporating all the specified parameters and settings. The output is a collection of images that can be used for various creative projects, animations, and visual effects.

KSampler Sequence Usage Tips:

• Experiment with different seed modes to see how they affect the variability and consistency of your generated sequences.
• Adjust the denoise_start and denoise_seq parameters to control the amount of noise reduction and achieve the desired level of detail in your images.
• Use the latent_interpolation_mode and latent_interp_strength parameters to create smooth transitions between different latent states, enhancing the fluidity of your animations.
• Enable use_conditioning_slerp and adjust cond_slerp_strength to create more natural and gradual changes in conditioning sequences.
• Try different sampler_name and scheduler combinations to find the best fit for your specific project and desired visual style.

KSampler Sequence Common Errors and Solutions:

"Invalid seed value"

• Explanation: The seed value provided is outside the acceptable range.
• Solution: Ensure that the seed value is between 0 and 0xffffffffffffffff.

"Steps value out of range"

• Explanation: The steps parameter is set to a value outside the allowed range.
• Solution: Adjust the steps parameter to be within the range of 1 to 10000.

"Invalid sampler name"

• Explanation: The specified sampler name is not recognized.
• Solution: Verify that the sampler name is one of the available options in comfy.samplers.KSampler.SAMPLERS.

"Scheduler not found"

• Explanation: The specified scheduler is not recognized.
• Solution: Ensure that the scheduler is one of the available options in comfy.samplers.KSampler.SCHEDULERS.

"Latent image not provided"

• Explanation: The latent_image parameter is missing or invalid.
• Solution: Provide a valid latent image to be used as the starting point for the sampling process.