ComfyUI Node: Image Injection 🎭🅐🅓

Image Injection 🎭🅐🅓:

The ADE_NoisedImageInjection node is designed to facilitate the injection of noised images into a generative model's sampling process. This node is particularly useful for AI artists who want to blend specific images into the generative process, allowing for more control over the final output. By injecting images at various stages of the sampling process, you can achieve unique artistic effects and ensure certain visual elements are preserved or emphasized. The node supports various customization options, such as masking, resizing, and strength modulation, making it a versatile tool for creative experimentation.

Image Injection 🎭🅐🅓 Input Parameters:

image

This parameter represents the image tensor that you want to inject into the generative process. The image will be blended with the generated content based on the specified options.

vae

The Variational Autoencoder (VAE) model used for encoding and decoding the image. This is essential for transforming the image into a latent space that the generative model can work with.

mask_opt

An optional mask tensor that defines which parts of the image should be injected. Areas covered by the mask will be affected by the injection, allowing for selective blending.

invert_mask

A boolean parameter that, when set to true, inverts the mask. This means that the areas not covered by the original mask will be affected by the injection. Default is false.

resize_image

A boolean parameter that determines whether the image should be resized to match the dimensions of the latent space. Default is true.

start_percent

A float value that specifies the starting point of the injection as a percentage of the total sampling process. This allows you to control when the image injection begins. The value ranges from 0.0 to 1.0, with a default of 0.0.

guarantee_steps

An integer that ensures the image injection occurs for a minimum number of steps during the sampling process. This helps in maintaining the influence of the injected image. The value ranges from 1 to a large maximum value, with a default of 1.

img_inject_opts

An optional parameter that provides additional options for image injection, such as positioning and other custom settings. This allows for fine-tuning the injection process.

strength_multival

A parameter that controls the strength of the image injection. It can be a float or a tensor, allowing for dynamic strength modulation. If not specified, the default value is 1.0.

prev_image_inject

An optional parameter that allows you to chain multiple image injections. This can be useful for complex scenarios where multiple images need to be injected at different stages.

Image Injection 🎭🅐🅓 Output Parameters:

IMAGE_INJECT

This output parameter represents the modified image injection group that includes the newly injected image. It can be used in subsequent nodes to continue the generative process with the injected image.

Image Injection 🎭🅐🅓 Usage Tips:

• Experiment with the start_percent parameter to find the optimal point in the sampling process for your image injection. Starting too early or too late can significantly affect the final output.
• Use the mask_opt parameter to selectively inject parts of the image. This can help in blending specific elements without overwhelming the entire generated content.
• Adjust the strength_multival to control the influence of the injected image. A higher value will make the injected image more prominent, while a lower value will make it more subtle.

Image Injection 🎭🅐🅓 Common Errors and Solutions:

"Invalid image tensor"

• Explanation: The provided image tensor is not in the correct format or is corrupted.
• Solution: Ensure that the image tensor is correctly formatted and not corrupted. Verify the dimensions and data type of the tensor.

"VAE model not provided"

• Explanation: The VAE model is missing or not correctly specified.
• Solution: Provide a valid VAE model to the vae parameter. Ensure that the model is compatible with the image tensor.

"Mask dimensions do not match image dimensions"

• Explanation: The dimensions of the mask tensor do not match those of the image tensor.
• Solution: Ensure that the mask tensor has the same dimensions as the image tensor. Resize the mask if necessary.

"Invalid start_percent value"

• Explanation: The start_percent value is out of the acceptable range.
• Solution: Set the start_percent value within the range of 0.0 to 1.0. Verify that the value is a float.

"Invalid guarantee_steps value"

• Explanation: The guarantee_steps value is out of the acceptable range.
• Solution: Set the guarantee_steps value to an integer within the acceptable range. Verify that the value is not less than 1.