ComfyUI Node: Image Upscale With Model Batched

Image Upscale With Model Batched:

The ImageUpscaleWithModelBatched node is designed to upscale images using a specified model while managing memory usage efficiently. This node is particularly useful for AI artists who work with high-resolution images and need to upscale them without running into memory limitations. By processing images in smaller sub-batches, it reduces the VRAM (Video RAM) usage, making it possible to upscale large sets of images even on hardware with limited memory. This node leverages the power of deep learning models to enhance image resolution, providing high-quality upscaled images suitable for various artistic and professional applications.

Image Upscale With Model Batched Input Parameters:

upscale_model

This parameter specifies the model used for upscaling the images. The model should be compatible with the node and capable of performing the upscaling task. The choice of model can significantly impact the quality and characteristics of the upscaled images.

images

This parameter represents the set of images that you want to upscale. The images should be provided in a format that the node can process, typically as a tensor or array. The quality and resolution of the input images will affect the final upscaled output.

per_batch

This parameter determines the number of images processed in each sub-batch. It helps manage VRAM usage by breaking down the upscaling task into smaller, more manageable chunks. The default value is 16, with a minimum of 1 and a maximum of 4096. Adjusting this value can help balance between processing speed and memory usage.

Image Upscale With Model Batched Output Parameters:

IMAGE

The output is a set of upscaled images. These images have been processed by the specified model and upscaled according to the parameters provided. The output images will have higher resolution and improved quality compared to the input images, making them suitable for high-detail artistic work or professional use.

Image Upscale With Model Batched Usage Tips:

• Adjust the per_batch parameter to find the optimal balance between processing speed and memory usage. Lower values reduce VRAM usage but may increase processing time.
• Ensure that the upscale_model is compatible and optimized for the type of images you are working with to achieve the best results.
• Use high-quality input images to maximize the benefits of the upscaling process, as the quality of the input directly affects the output.

Image Upscale With Model Batched Common Errors and Solutions:

CUDA out of memory

• Explanation: This error occurs when the VRAM is insufficient to process the current batch of images.
• Solution: Reduce the per_batch parameter to process fewer images at a time, thereby lowering the VRAM usage.

Model not found

• Explanation: The specified upscale_model is not available or not loaded correctly.
• Solution: Ensure that the model is correctly installed and specified. Verify the model path and compatibility.

Invalid image format

• Explanation: The input images are not in a format that the node can process.
• Solution: Convert the images to a compatible format, such as a tensor or array, before passing them to the node.