ComfyUI Node: KSampler FABRIC (Simple)

KSampler FABRIC (Simple):

KSamplerFABRIC is a specialized node designed to enhance the sampling process in AI art generation by incorporating FABRIC (Feedback and Adaptive Bias in Recurrent Image Creation) techniques. This node extends the capabilities of the regular KSampler by allowing for more nuanced control over the sampling process through additional parameters. It is particularly useful for artists looking to fine-tune their generated images by adjusting weights and feedback mechanisms, thereby achieving more precise and desired outcomes. The main goal of KSamplerFABRIC is to provide a more flexible and adaptive sampling method that can handle complex conditioning scenarios, making it a valuable tool for advanced AI art creation.

KSampler FABRIC (Simple) Input Parameters:

model

This parameter specifies the model to be used for sampling. It is a required input and determines the underlying architecture that will generate the images.

seed

This integer parameter sets the random seed for the sampling process. It ensures reproducibility of results. The default value is 0, with a minimum of 0 and a maximum of 0xffffffffffffffff.

steps

This integer parameter defines the number of steps to be used in the sampling process. More steps generally lead to higher quality images but take longer to compute. The default value is 20, with a minimum of 1 and a maximum of 10000.

cfg

This float parameter stands for "Classifier-Free Guidance" and controls the strength of the guidance during sampling. Higher values lead to images that more closely follow the conditioning. The default value is 8.0, with a minimum of 0.0 and a maximum of 100.0, adjustable in steps of 0.1.

sampler_name

This parameter allows you to choose the sampling algorithm to be used. It is a required input and offers various options provided by the comfy.samplers.KSampler.SAMPLERS.

scheduler

This parameter specifies the scheduler to be used during the sampling process. It is a required input and offers various options provided by the comfy.samplers.KSampler.SCHEDULERS.

positive

This conditioning parameter provides the positive conditioning for the sampling process. It is a required input and significantly influences the generated image.

negative

This conditioning parameter provides the negative conditioning for the sampling process. It is a required input and helps in steering the generated image away from undesired features.

latent_image

This parameter provides the latent image to be used as a starting point for the sampling process. It is a required input and serves as the initial state for the generation.

denoise

This float parameter controls the amount of denoising applied during the sampling process. The default value is 1.0, with a minimum of 0.0 and a maximum of 1.0, adjustable in steps of 0.01.

clip

This parameter provides the CLIP model to be used for encoding text into conditioning vectors. It is a required input and is essential for generating the null conditioning vectors.

pos_weight

This float parameter adjusts the weight of the positive conditioning. The default value is 1.0, with a minimum of 0.0 and a maximum of 1.0, adjustable in steps of 0.01.

neg_weight

This float parameter adjusts the weight of the negative conditioning. The default value is 1.0, with a minimum of 0.0 and a maximum of 1.0, adjustable in steps of 0.01.

feedback_percent

This float parameter specifies the percentage of steps at which feedback is applied during the sampling process. The default value is 0.8, with a minimum of 0.0 and a maximum of 1.0, adjustable in steps of 0.01.

pos_latents

This optional parameter provides the positive latents to be used during the sampling process. If not provided, an empty tensor is used.

neg_latents

This optional parameter provides the negative latents to be used during the sampling process. If not provided, an empty tensor is used.

KSampler FABRIC (Simple) Output Parameters:

LATENT

The output of this node is a latent tensor that represents the generated image in its latent space. This tensor can be further processed or decoded to obtain the final image. The latent output is crucial for understanding the intermediate state of the generated image and can be used for further refinement or analysis.

KSampler FABRIC (Simple) Usage Tips:

• Experiment with different values of cfg to find the optimal balance between adherence to conditioning and creative freedom.
• Use the feedback_percent parameter to control how much feedback is applied during the sampling process, which can help in fine-tuning the generated images.
• Adjust the pos_weight and neg_weight parameters to emphasize or de-emphasize certain features in the generated image based on your artistic goals.

KSampler FABRIC (Simple) Common Errors and Solutions:

"No reference latents found. Defaulting to regular KSampler."

• Explanation: This error occurs when both pos_latents and neg_latents are not provided or are empty.
• Solution: Ensure that you provide valid pos_latents and neg_latents if you want to use the advanced FABRIC features. Otherwise, the node will default to regular KSampler behavior.

"Invalid seed value."

• Explanation: This error occurs when the seed value is outside the acceptable range.
• Solution: Ensure that the seed value is within the range of 0 to 0xffffffffffffffff.

"Invalid steps value."

• Explanation: This error occurs when the steps value is outside the acceptable range.
• Solution: Ensure that the steps value is within the range of 1 to 10000.

"Invalid cfg value."

• Explanation: This error occurs when the cfg value is outside the acceptable range.
• Solution: Ensure that the cfg value is within the range of 0.0 to 100.0.

"Invalid denoise value."

• Explanation: This error occurs when the denoise value is outside the acceptable range.
• Solution: Ensure that the denoise value is within the range of 0.0 to 1.0.