ImageScaleToNormPixels(imageScaleToNormPixels)

The ImageScaleToNormPixels node scales an input image by a relative multiplier. It is grouped under EasyUse/Image and returns a resized IMAGE tensor, making it useful when you want to upscale or downscale by a factor rather than targeting a specific width/height.

Inputs

• image (IMAGE, required): The input image tensor to be resized.
• upscale_method (COMBO, required): The resampling/upscaling algorithm to use. The dropdown provides the supported methods; select the one best suited for the image content and scaling direction.
• scale_by (FLOAT, required): The scale factor applied to both width and height. 1.0 leaves the image unchanged, values greater than 1.0 upscale, and values less than 1.0 downscale. Default is 1.0, with a range of 0.01 to 8.0 and a step of 0.01.

Outputs

• IMAGE: The scaled image tensor. The batch size and channel count match the input, while the spatial dimensions are multiplied by scale_by.

Usage Notes

• Because scale_by is a multiplier, 2.0 produces an image twice as large in both dimensions, while 0.5 halves both dimensions.
• This node operates directly on standard image pixel tensors rather than latent representations.
• There are no separate width or height inputs; the output resolution is determined entirely by the input resolution and the chosen scale_by value.
• Use the upscale_method dropdown to control quality/behavior when changing resolution. Some methods are better suited for photographic images, while others are better for pixel art or UI elements.