Learn & Understand

When Non-Uniform Scaling Is Actually the Right Call

Uniform scaling is the right default almost every time - but "almost every time" isn't "every time," and knowing the genuine exceptions matters more than memorizing the general rule.

Exception One: Correcting Known Anisotropic Shrinkage

As covered in this category's shrinkage guide, some materials - particularly semi-crystalline plastics like nylon - don't necessarily shrink identically in every axis, with the layer-built Z axis sometimes behaving differently from the continuously-printed X/Y plane. If you've measured a finished part and found the Z-axis dimension is off by a different percentage than the X/Y dimensions, applying a single uniform shrinkage correction to all three axes will only fully fix two of them at best. In this specific, measured case, applying a separate, non-uniform scale percentage to the Z axis alone - based on actually observed Z-axis shrinkage, not guessed - is the technically correct fix, not a shortcut being taken.

Exception Two: Correcting a Model That Was Exported at the Wrong Proportions

Occasionally a downloaded or exported model simply has incorrect native proportions in one axis - a CAD export bug, a unit conversion error somewhere in the model's history (feet vs. meters vs. millimeters confusion is a common culprit), or a scan-based model with a known single-axis distortion from the scanning process. If you can identify that the error is genuinely isolated to one axis - confirmed against a known correct reference dimension in the other two axes - correcting just that one axis is appropriate. The key distinction from a mistake is that you have a specific, verified reason the model is wrong in only one direction, rather than assuming that must be the case.

Why the Default Warning Against Non-Uniform Scaling Still Matters

Outside these specific, evidence-based cases, stretching only one axis of a model without a verified reason will distort curved surfaces (a circle becomes an ellipse, a sphere becomes an egg shape) and can break tight-tolerance features like screw threads, snap-fits, or bearing seats that were designed assuming the model's original proportions. This is exactly why uniform scaling remains the correct default - the exceptions above require you to already know something specific and measured about why one axis needs different treatment, not a general hunch that "it looks a little off."

Choosing the Right Reference Dimension First

Whichever approach you use, pick the reference dimension carefully on a complex model - the first number a slicer happens to display (often just the bounding box's longest axis) isn't necessarily the dimension that actually matters for your part's fit or function. Measure the specific feature that needs to match a real-world target, not just whatever dimension is easiest to read off the slicer's summary panel.