Scale

A scale specifies an attribute-specific transformation that should be carried out before mapping it to various visual channels. Note that a scale can be associated with one or more Attributes.

Uniform scale

Uniform scale rescale the attribute uniformly based on the specified factor.

s = hkw.scale.Uniform(factor = 2)

Log scale

It is quite common that we want to visualize the log of the data. The Log scale provides this capability.

s = hkw.scale.Log(base = 2)

In this example, we have created a log scale with base 2.

Clip scale

Clip scale truncates the data with the specified domain. Data values below the domain minimum will be set to the domain minimum, and vice versa for the maximum.

s = hkw.scale.Clip(domain=[0, 1])

Normalize scale

Normalize scale rescales the attribute such that the specified domain maps to the specified range. For example,

s = hkw.scale.Normalize(range_min=0, range_max=1, domain_min=-10, domain_max=10)

The scale s will rescale a attribute such that the data within [-10, 10] is now within [0, 1]. If the domain is not specified, the maximum and minimum value of the attribute will be used as the domain.

s = hkw.scale.Normalize(range_min=np.zeros(3), range_max=np.ones(3))

In this example, we are rescaling a vector attribute so that it fits within the box from [0, 0, 0] to [1, 1, 1].

Affine scale

Affine scale applies a affine transformation to a vector-valued attribute.

s = hkw.scale.Affine(matrix = np.eye(3) * 2)

Offset scale

Offset scale offsets the current attribute by another attribute. It is useful for showing deformed data where the rest position and displacement field are stored in separate attributes.

s = hkw.scale.Offset(offset=hkw.attribute("displacement"))

Custom scale

Custom scale allows one to use arbitrary scaling function. The function should take a single data value (either a scalar or a vector) as input and output the scaled data value. For example, one can reproduce the effect of uniform scaling by 2 using the following custom scale.

s = hkw.scale.Custom(function = lambda x : x * 2)

Combining multiple scales

It is often necessary to apply multiple scales on an attribute. Hakowan provides an easy way of combining scales together.

s1 = hkw.scale.Log(base = 2)
s2 = hkw.scale.Uniform(factor = 2)

# Chain s1 and s2 together. s1 will be applied before s2.
s = s1 * s2