0.1.0a21

0.1.0a21

• NEW: Refactor CAT to allow for other CAT methods: von-kries, xyz-scaling, cat02, cmccat97, cmccat2000,
  and sharp. Currently, bradford is the default and the overall preferred option.
• FIX: Fix issues gamut mapping algorithm.

0.1.0a20

0.1.0a20

• FIX: Fix lab-d65 which was not using the correct white point in all places.

0.1.0a19

0.10a19

• FIX: Ensure that subclassed Color objects are normalized when performing operations with more than one color to
  prevent issues in case one subclassed object has overridden important functions.
• FIX: Spaces like lab, lch, etc., which specify certain channels as percent only should require the color()
  format to only accept percentages for those channels and output those channels as percentages when converting to a
  string.

0.1.0a18

0.10a18

• NEW: Refactor of internals.
• NEW: interpolate and steps can now accept multiple colors and will return an interpolation function that spans all
  specified colors via the range of [0..1].
• NEW: Better control over piecewise interpolation: setting stops, adjusting options per segment, etc.
• NEW: compose can now accept multiple colors and will return a result where all colors are layered on top of each
  other.
• NEW: new method does not need to be a classmethod. Make it a normal method on the instance.
• NEW: Add Jzazbz and JzCzhz color spaces.
• NEW: Add D65 variants of CIELAB, CIELCH, and XYZ.
• NEW: Add ICtCp color space and Delta E ITP method.
• FIX: Actually make mix default to lab like interpolate and friends do.

0.1.0a17

0.1.0a17

• FIX: Ensure that both the Bradford CAT and the XYZ transformation matrix all use ASTM E308-01 white points. This fixes a number of conversion issues when going to and from D65 to D50 color spaces.

0.1.0a16

0.1.0a16

• NEW: Make mix use the same space logic as interpolate and step. Colors are mixed in CIELAB unless space is set to a different color space.
• NEW: Add support for blend modes as specified in [Compositing and Blending Level 1][compositing-level-1].
• NEW: Rename overlay to compose as all compositing (including blend modes) is done through compose now.overlay is still present and is deprecated and will be removed at some future point before a stable release. compose will assume sRGB space unless a different space is specified, but overlay will function as it always did.
• FIX: Fix some small internal issues with in_place logic.

0.1.0a15

0.1.0a15

• NEW: The adjust parameter on interpolate, steps, and mix has been dropped. Instead, a general purpose method
  has been added to the Color object to mask one or more channels at a time. This can be used to create a temporary
  color with masked channels for the purpose of interpolation.
• FIX: Ensure that when alpha is NaN that it is handled in overlay.
• FIX: When using raw data in the color() function, and there is too little data, fill data with NaN.
• FIX: Fix issue where API interpolate method does not pass out_space parameter down.
• FIX: Disabling or forcing alpha did not work properly for HSL colors via to_string.
• FIX: contrast and luminance should use XYZ with a D65 white point, not the default XYZ space which uses a D50
  white point.
• FIX: Fix bug in Delta E 2000 algorithm.

0.1.0a14

0.1.0a14

• NEW: sRGB string output parameter hex_upper has been renamed to just upper. Expose it in documentation as well.

0.1.0a13

0.1.0a13

• FIX: More efficient calculation of CIELAB, following CIE 15.3:2004. Results are still the same, but it makes the math a little simpler.
• FIX: HSV did not always set hue to NaN when saturation was 0.
• FIX: Give better conversion results by having HWB colors pass through HSV instead of sRGB.
• FIX: Fix slight issue with REC.2020 and ProPhoto color space conversion. Small issue when using <= when < was desired.

0.1.0a12

0.1.0a12

• FIX: More stable saturation calculation for HSL to ensure divide by zero doesn't occur. 2(V - L) / (1 - abs(2 * L - 1)) is likely to yield zero in the denominator when L is very small, while the equivalent (V - L) / min(L, 1 - L) is not.