Upscale Node
Learn about Rubbrband’s Upscale Node
Rubbrband Upscale Node
The Rubbrband Upscaler is an upscaler that is produced by Rubbrband that is good for maintaining all of the image’s details. Changing the parameters can seriously influence the outputs. Currently, the Rubbrband upscaler only supports up to 4096x4096 as the output. Inputs
The image that will be upscaled
Settings
Cfg is a field that is for prompt guidance. Increasing this increases the strength of the prompt in the upscale.
Increasing this will increase the granularity of the image, giving importance to small details being preserved in the upscale. It may lead to graininess if set too high.
Increasing this will increase the diversity/creativity of the upscaled image.
Increasing this will increase the inference time for the upscale and will add more iterations to the generation process to make images more refined.
Outputs
The upscaled image.
Clarity Upscale Node
The clarity upscaler is a famous upscaler in the AI community that is very good for realism and control Inputs
The prompt that will be used to upscale the image. Usually a description of the image.
The negative prompt, which represents what you do not want to see in the output.
The image that will be upscaled
Settings
Increasing this will increase the diversity/creativity of the upscaled image.
Increasing HDR increases the vivid-ness and life-likeness of an image.
Increasing this will increase the inference time for the upscale and will add more iterations to the generation process to make images more refined.
Increasing this will result in more details of the original image being preserved.
The tiling height determines the size of image “tiles” processed individually during upscaling. Increasing the tiling height reduces image fractality by using larger tiles, while decreasing it creates more subdivisions, resulting in finer initial detail.
The tiling width determines the size of image “tiles” processed individually during upscaling. Increasing the tiling width reduces image fractality by using larger tiles, while decreasing it creates more subdivisions, resulting in finer initial detail.
The upscale factor, which represents how many times higher in resolution is desired.
Outputs
The upscaled image
ESRGAN Upscale Node The ESRGAN Upscale Node does a direct upscale without any diffusion models. It is very simplistic, fast, and barren.
Inputs
The image that will be upscaled
Outputs
The upscaled image