Nodes/Post Processing PRO/Auto Resize Match PRO
ComfyUI Node

Auto Resize Match PRO

The boring node that quietly saves your compositing workflow

By Guillaume-127·Created 5 months ago·Updated 5 months ago· 2
Auto Resize Match PRO
  • image
  • reference_image
  • IMAGE
interpolation

Every other node in the Post Processing PRO pack is a fancy optical or film emulation. This one is just: "make image A the same size as image B." That's it. And honestly, it might be the node you actually use most, because size mismatches are the silent killer of ComfyUI workflows - you paste an inpainted patch, you composite a detailer pass, you overlay a mask - and nothing lines up because one image is 1344×768 and the other is 1024×1024, and the whole graph quietly breaks or looks wrong.

Auto Resize Match PRO takes an input image, reads the exact width and height of a reference image, and resizes the input to match - pixel-perfect, with your choice of interpolation quality. It's the alignment tool you didn't know you needed until a two-minute job turned into a mismatch hunt.

How it works

The mechanism is refreshingly unmagical: it takes the resolution from reference_image, resizes image to those exact dimensions with OpenCV, and returns it. A few details worth knowing:

  • If the input and reference are already the same size, it returns the original untouched - zero cost, no pointless resample.
  • interpolation is a dropdown, not a slider: lanczos, bicubic, bilinear, nearest. Lanczos is the default and the right choice for almost everything - sharpest high-quality downscale and upscale. Nearest is there for pixel-art or mask alignment where you don't want any smoothing at all.
  • It handles batches - if your input is a batch of frames or images, every one of them gets resized to the reference dimensions.

The inputs that matter

  • image - the thing being resized.
  • reference_image - the thing whose dimensions you want to match. This is the anchor; wire your main image here and the layer/mask into image. Getting these backwards is the #1 mistake, and the result is usually a silently squashed image.
  • interpolation - lanczos (default) for photographic work, nearest for masks and pixel art.

Output is an IMAGE at the reference resolution. Wire it into a composite, a mask operation, or anything downstream that expects matching dimensions.

Installing it

Same pack as the rest - install once, get all 14 nodes. ComfyUI Manager → search "Post Processing PRO", or:

cd ComfyUI/custom_nodes
git clone https://github.com/Guillaume-127/ConfyUI_Post_Processing_PRO

Restart, done. No requirements, no models, no GPU needed - this one is pure OpenCV resize math and runs instantly.

Where people get burned

The honest limitation: it matches size only, not aspect - it stretches to fill the reference exactly. If your input is 16:9 and the reference is 1:1, you get a squashed image, because there's no crop-and-fill option here. For compositing masks that's usually fine (masks stretch happily); for photographic layers, resize your layer to the right aspect first with a normal image resize node, then use this for the final pixel-exact match. It's a utility node, so there's no artistic subtlety to dial in - it either lines up or it doesn't, which is exactly what you want from a tool in the Utilities category.

CategoryPost_Processing_PRO/Utilities

Inputs (3)

NameTypeDefaultDescription
imageIMAGE
reference_imageIMAGE
interpolationCOMBO4 options: lanczos, bicubic, bilinear, nearest

Outputs (1)

NameTypeDescription
IMAGEIMAGE