ImageScaleFactorSimple _O

If ImageScaleFactor _O is the Swiss-army scaler, ImageScaleFactorSimple _O is the "one button" edition: instead of separate width and height factors, it takes a single Factor applied uniformly to both dimensions. Image in, image out, scaled by whatever multiplier you set.

It sits next to its sibling under O/image in the Quality of Life Suit. For the 95% of scaling jobs where you want to keep the aspect ratio - double it, halve it, bump it 1.5× for a hires pass - this is the node you actually reach for. The two-factor version only earns its keep when you genuinely want to squash or stretch an image independently, which is rare outside of very deliberate composition work.

How it works

It's the same engine as the fancier sibling: comfy.utils.common_upscale, the shared ComfyUI resize helper, driven by Factor applied to both axes. The settings that matter:

• Factor - the multiplier, default 1.25, range 0–10. Below 1 downscales, above 1 upscales. The step is 0.28125, so you can hit most reasonable values with a couple of nudges.
• upscale_method - nearest-exact, bilinear, or area. area for downscaling, bilinear for upscaling, nearest-exact if you're doing pixel art.
• MulOf46 - rounds output dimensions to a multiple of 8; on by default. Leave it on with SD-family models; an off-multiple latent size is a classic source of broken generations.
• crop - disabled or center, for when you need to crop after the resize.

Output dimensions are capped at 8192px. The single output is an IMAGE.

Installing

cd ComfyUI/custom_nodes
git clone https://github.com/omar92/ComfyUI-QualityOfLifeSuit_Omar92.git

or ComfyUI Manager → search ComfyUI-QualityOfLifeSuit_Omar92, then restart. No dependencies, no models.

The honest take

Straight up: if you're only going to install one of the two factor scalers, make it this one. Uniform scaling covers the hires-fix case (1.5–2x is the community sweet spot for a second pass) and every preprocessor-prep case. Remember it's a plain interpolation resize - no detail added, just geometry. The detail in a hires pass comes from the img2img sampling that follows, not from this node. And keep MulOf46 on; the only time you'd disable it is if you're feeding an unusual consumer that doesn't care about multiples of 8.