ComfyUI Node: Support empty uncond

Support empty uncond:

The Support empty uncond node is designed to enhance the flexibility and robustness of your AI model by handling cases where unconditional conditioning (uncond) might be empty or missing. This node ensures that your model can still function effectively even when the uncond is not provided or is empty, by either dividing the conditional output by the CFG (Classifier-Free Guidance) scale or cloning the conditional output to replace the uncond. This capability is particularly useful in scenarios where the absence of uncond could otherwise lead to suboptimal model performance or errors. By integrating this node, you can maintain the stability and reliability of your model across a wider range of input conditions.

Support empty uncond Input Parameters:

model

This parameter represents the AI model that you are working with. It is essential for the node to know which model to apply the patch to, ensuring that the modifications are correctly implemented. The model parameter does not have specific minimum, maximum, or default values as it is dependent on the model you are using in your workflow.

method

This parameter determines the approach the node will take when handling an empty uncond. It offers two options: from cond and divide by CFG. If from cond is selected, the node will clone the conditional output to replace the uncond. If divide by CFG is chosen, the node will divide the conditional output by the CFG scale. This parameter allows you to control how the node compensates for the absence of uncond, ensuring that the model's performance remains consistent. The default value is from cond.

Support empty uncond Output Parameters:

model

The output parameter is the modified AI model. This model has been patched to handle empty uncond scenarios according to the specified method. The importance of this output lies in its enhanced capability to manage cases where uncond is missing, thereby improving the model's robustness and reliability. The output model can be used in subsequent nodes or processes within your workflow, ensuring that the modifications are seamlessly integrated.

Support empty uncond Usage Tips:

• To ensure optimal performance, choose the divide by CFG method if you want to maintain the proportionality of the conditional output when uncond is missing.
• Use the from cond method if you prefer a straightforward approach where the conditional output is simply cloned to replace the uncond, which can be useful in simpler models or scenarios.

Support empty uncond Common Errors and Solutions:

"AttributeError: 'NoneType' object has no attribute 'clone'"

• Explanation: This error occurs when the model parameter is not correctly provided or is None.
• Solution: Ensure that you pass a valid model to the node. Verify that the model is correctly loaded and passed as an argument.

"TypeError: unsupported operand type(s) for /: 'Tensor' and 'int'"

• Explanation: This error happens when the method divide by CFG is selected, but the conditional output is not a tensor or the CFG scale is not an integer.
• Solution: Check that the conditional output is a tensor and that the CFG scale is correctly defined as an integer. Adjust the inputs accordingly to match the expected types.