Difference in intended use between the Hypotenuse node and the Vector2dLength node?

Development Programming & Scripting Blueprint
, ,
1

As far as I can tell they should be the same mathematically however the 2D vector length node works for all possible input vectors while the hypotenuse node does not for the edge case of both inputs being zero. (UE-69468)
Are there different intended use cases for these two nodes and is there a reason for using one over the other (aside from the hypotenuse node not working for some values)?

2

If you look through the source code, Vector2dLength (or Vector2D size) represents the Pythagorean Theorem whereas Hypotenuse is modified to be safer way to avoid overflow/underflow caused by squaring the width and height (as stated in the implementation comment):

Vector2dLength implementation

FORCEINLINE float FVector2D::Size() const
{
	return FMath::Sqrt(X*X + Y*Y);
}

Hypotenuse implementation

float UKismetMathLibrary::Hypotenuse(float Width, float Height)
{
	// This implementation avoids overflow/underflow caused by squaring width and height:
	Width = FMath::Abs(Width);
	Height = FMath::Abs(Height);

	float Min = FGenericPlatformMath::Min(Width, Height);
	float Max = FGenericPlatformMath::Max(Width, Height);
	float Ratio = Min / Max;
	return Max * FMath::Sqrt(1.f + Ratio * Ratio);
}

As you can see above, Hypothenuse has more going on to arrive on the same result with Vector2dLength.

As for what to use where, it depends on your needs. If you prefer to be safe, then go with Hypotenuse by all means. If you prefer every bit of performance (no matter how unnoticeable it is), then go with Vector2dLength.

3

Thanks, this answers my question quite well, it also seems to explain why the hypotenuse node fails with width and height of zero.