Fills an array with colors blended between keyframes, in RGB, HSV or Oklab.
A Color_Keyframe pairs a color with a position t in [0, 1]; generate_color_samples spreads n output samples evenly over [0, 1] and blends the two keys around each. RGB and Oklab blending works on colors linearized with a plain gamma of 2.2 (GAMMA), not the exact sRGB curve, and encodes the result the same way; HSV blends hue, saturation and value directly, taking the shorter way around the hue circle. Outside the first and last key the nearest key's color holds, unless wrapping is set, which blends the last key into the first across the 1 to 0 seam.
Converts a red, green, blue color (channels in [0, 1]) to hue in degrees [0, 360), saturation and value in [0, 1]. Takes a Vector3 and returns a Vector3, or takes r, g, b and returns h, s, v. Greys (no difference between channels) give hue 0 and black gives saturation 0.
Converts hue in degrees, saturation and value (both in [0, 1]) to red, green, blue. Takes a Vector3 and returns a Vector3, or takes h, s, v and returns r, g, b. The hue wraps, so -60 and 300 are the same, and a saturation of 0 gives a grey of the given value.
Converts a linear-light sRGB color to Oklab: x is lightness (0 for black, about 1 for white), y and z the green-red and blue-yellow axes. The input must be linear, not gamma-encoded; see Srgb.srgb_to_linear. The transform runs in float64 and rounds to float at the end.
Converts an Oklab color back to linear-light sRGB. Colors outside the sRGB gamut are clamped into [0, 1] per channel (a simple clip, not a gamut mapping), so a round trip of an out-of-gamut Oklab color does not return it.
Interpolates from a to b along a smoothstep 3t^2 - 2t^3 of t, so the change starts and ends gently. Takes float or Vector3; t is expected in [0, 1] and is not clamped.
Prepares the key colors for blending in space: gamma-decoded to linear light for RGB, Oklab-converted from that for Oklab, and as hue, saturation, value for HSV. Returns one Vector3 per key (alpha is read from the keys later). The array is allocated with context.allocator and the caller frees it with array_free.
Blends one sample at position t. sorted are the keys (by t ascending), temp_colors the result of make_colors for the same keys and space, and i0 the index of the last key at or before t; use -1 for a t before the first key when wrapping. Returns the red, green, blue, alpha color, encoded again. With a single key that key's color is returned. Asserts when there are no keys or the two arrays differ in length.
Fills output_samples with a gradient through the keys: sample i of n is at t = i / (n - 1), so the first sample is at 0 and the last at 1. sorted must be sorted by t ascending. Nothing is written when there are no samples or no keys; a single sample gets the first key's color unchanged; an unsupported color space prints a message to standard error and writes nothing.
How generate_color_samples blends: the color_space the blend happens in (.RGB, .HSV or .OKLAB), the geometry that shapes the blend between two keys (.LINEAR, .QUADRIC for an ease in and out, .HERMITE for a smoothstep) and wrapping, which blends the last key into the first across the seam instead of holding the end colors. The default value is RGB, linear, not wrapping.
One key of a gradient: color_rgb is a red, green, blue, alpha color with channels in [0, 1] (encoded like an ordinary image color, not linear light) and t its position in [0, 1]. A list of keys must be sorted by t ascending. Two keys at the same position make a hard step. Alpha is always blended linearly, whatever the color space.
The gamma, 2.2, used to turn the encoded colors of the keys into linear light for blending in the RGB and Oklab spaces and back again. It is a power law, a little different from the piecewise sRGB curve of the Srgb module.