Resize

Overview

Resize node scales the input image to a target resolution.

Use this node to standardize frame dimensions before feeding frames into models with fixed input sizes, or to reduce resolution to improve downstream processing speed.

Input

Input Image

image required

The image frame to resize. Connect this to a camera or upstream processing output.

Size

array required

Target output dimensions as [width, height] in pixels.

Default: [640, 480]

Keep Aspect Ratio

boolean required

When enabled, the image is scaled uniformly so that neither dimension exceeds the target size. The remaining area is padded. When disabled, the image is stretched to fill the exact target size.

Default: true

Interpolation

string required

Resampling algorithm used when scaling pixels.

Values:

• NEAREST — fastest; blocky result; suitable for binary masks.
• LINEAR — bilinear interpolation; good general-purpose choice.
• CUBIC — bicubic interpolation; smoother result than LINEAR but slower.
• AREA — pixel area averaging; best for downscaling to avoid moiré.
• LANCZOS4 — high-quality Lanczos resampling; slowest; best for upscaling.
• AUTO_BEST — automatically selects AREA for downscaling and CUBIC for upscaling.

Output

Frame Out

image

Resized image.