mirror of
https://github.com/storytold/effectcraft.git
synced 2026-11-10 18:31:30 -06:00
FredApps
9893eb0d8d
Related to #1365. Display labels only; effect, command, parameter, panel and serialized ids and the AEP/script match-name tables are unchanged. - 69 Cycore CC effects, Key Light, Mocha shape, Lumetri Color/Scopes and the Mercury renderer labels get descriptive names; the old names stay case-insensitive lookup and search aliases. - Backend::parse accepts the new renderer labels, so reported settings round-trip. - Display names are unique per language in every effect catalog (new test). - Warp Stabilizer and Roto Brush keep their names until RFC-0029 settles. - Mapping in docs/feature-naming.md.
662 lines
27 KiB
Rust
662 lines
27 KiB
Rust
//! Utility effects: bounds growth, HDR inspection/compression, Cineon log conversion and colour
|
||
//! lookup tables (.cube).
|
||
|
||
use std::collections::HashMap;
|
||
use std::sync::{Arc, Mutex, OnceLock};
|
||
|
||
use effectcraft_color::luminance;
|
||
use effectcraft_keyframe::Value;
|
||
use effectcraft_project::ParamUi;
|
||
use rayon::prelude::*;
|
||
|
||
use crate::util::{premul, unpremul};
|
||
use crate::{Buf, EffectCtx, EffectSpec, col, num, p, popup, slider};
|
||
|
||
fn spec(id: &'static str, name: &'static str, params: Vec<crate::ParamSpec>, render: crate::RenderFn) -> EffectSpec {
|
||
EffectSpec { id, name, category: "Utility", params, render, gpu: false, float: true }
|
||
}
|
||
|
||
fn grow_bounds(ctx: &EffectCtx, mut b: Buf) -> Buf {
|
||
let px = (ctx.params.f("pixels") * b.scale).round().max(0.0) as u32;
|
||
if !ctx.adjustment {
|
||
b.pad(px);
|
||
}
|
||
b
|
||
}
|
||
|
||
/// Visualise out-of-range values: in-range shown as dim grey, > 1 in the highlight colour,
|
||
/// < 0 in its complement.
|
||
fn overbrights(ctx: &EffectCtx, mut b: Buf) -> Buf {
|
||
let ch = ctx.params.e("channel");
|
||
let hc = ctx.params.color("highlightColor");
|
||
b.img.data.par_iter_mut().for_each(|px| {
|
||
let (c, a) = unpremul(*px);
|
||
if a <= 0.0 {
|
||
return;
|
||
}
|
||
let v = match ch {
|
||
1 => c[0],
|
||
2 => c[1],
|
||
3 => c[2],
|
||
4 => a,
|
||
5 => luminance(c[0], c[1], c[2]),
|
||
_ => c[0].max(c[1]).max(c[2]),
|
||
};
|
||
let o = if v > 1.0 {
|
||
[hc[0], hc[1], hc[2]]
|
||
} else if v < 0.0 {
|
||
[1.0 - hc[0], 1.0 - hc[1], 1.0 - hc[2]]
|
||
} else {
|
||
let g = v * 0.35;
|
||
[g, g, g]
|
||
};
|
||
*px = premul(o, a);
|
||
});
|
||
b
|
||
}
|
||
|
||
/// Luma weights HDR Highlight Compression divides by (measured against After Effects 2026).
|
||
const HDR_LUMA: [f32; 3] = [0.30, 0.59, 0.11];
|
||
/// Exponent of the domain the Amount blend happens in (out^0.4 is mixed linearly).
|
||
const HDR_BLEND_GAMMA: f32 = 0.4;
|
||
|
||
/// `v^e` with the sign of `v` kept (negative input stays negative, as on the GPU).
|
||
#[inline]
|
||
fn hdr_spow(v: f32, e: f32) -> f32 {
|
||
v.signum() * v.abs().powf(e)
|
||
}
|
||
|
||
/// One straight-colour pixel through HDR Highlight Compression at blend `t` (Amount / 100, > 0).
|
||
///
|
||
/// Measured behaviour (16-bit renders of a 0–4 HDR ramp and coloured patches at Amount 100,
|
||
/// 90, 84, 75, 50 and 25): every channel is divided by `1 + Y` (`Y` = 0.30 R + 0.59 G + 0.11 B, so a grey
|
||
/// value `x` lands on `x / (1 + x)`), clipped at 1, then mixed with the input as `x^0.4` values.
|
||
fn hdr_compress_pixel(c: [f32; 3], t: f32) -> [f32; 3] {
|
||
let y = (HDR_LUMA[0] * c[0] + HDR_LUMA[1] * c[1] + HDR_LUMA[2] * c[2]).max(0.0);
|
||
let s = 1.0 / (1.0 + y);
|
||
let g = HDR_BLEND_GAMMA;
|
||
c.map(|x| {
|
||
let comp = (x * s).min(1.0);
|
||
let mixed = (1.0 - t) * hdr_spow(x, g) + t * hdr_spow(comp, g);
|
||
let o = hdr_spow(mixed, 1.0 / g);
|
||
if o.is_finite() { o } else { x }
|
||
})
|
||
}
|
||
|
||
/// HDR Highlight Compression: every straight-colour pixel through [`hdr_compress_pixel`] at
|
||
/// Amount / 100; Amount 0 or NaN returns the layer unchanged.
|
||
fn hdr_compression(ctx: &EffectCtx, mut b: Buf) -> Buf {
|
||
let t = (ctx.params.f("amount") / 100.0).clamp(0.0, 1.0) as f32;
|
||
if t <= 0.0 || !t.is_finite() {
|
||
return b;
|
||
}
|
||
b.img.map_straight(|c| hdr_compress_pixel(c, t));
|
||
b
|
||
}
|
||
|
||
// ---- Cineon (Kodak printing-density log encoding: 0.002 density per code value, negative gamma 0.6) ----
|
||
|
||
#[derive(Clone, Copy, Debug)]
|
||
pub(crate) struct Cineon {
|
||
pub black: f32,
|
||
pub white: f32,
|
||
pub ib: f32,
|
||
pub iw: f32,
|
||
pub gamma: f32,
|
||
pub rolloff: f32,
|
||
}
|
||
|
||
impl Cineon {
|
||
fn offset(&self) -> f32 {
|
||
10f32.powf((self.black - self.white) * 0.002 / 0.6)
|
||
}
|
||
fn knee(&self) -> f32 {
|
||
1.0 - (self.rolloff / 100.0).clamp(0.0, 0.99)
|
||
}
|
||
/// Normalized code value (code / 1023) → linear, raised to `1.7 / gamma`
|
||
/// ([`effectcraft_project::cineon_log_to_lin`], which Interpret Footage ▸ Cineon Settings
|
||
/// also uses with `1 / gamma`).
|
||
pub fn log_to_lin(&self, v: f32) -> f32 {
|
||
let lin = effectcraft_project::cineon_log_to_lin(v, self.black, self.white, 1.7, self.gamma, self.rolloff);
|
||
self.ib + (self.iw - self.ib) * lin
|
||
}
|
||
/// Linear → normalised code value.
|
||
pub fn lin_to_log(&self, v: f32) -> f32 {
|
||
let off = self.offset();
|
||
let d = self.iw - self.ib;
|
||
let mut lin = (v - self.ib) / if d.abs() < 1e-6 { 1e-6 } else { d };
|
||
let k = self.knee();
|
||
if k < 1.0 && lin > k {
|
||
let w = 1.0 - k;
|
||
lin = k - w * (1.0 - (lin - k) / w).max(1e-6).ln();
|
||
}
|
||
lin = lin.max(0.0).powf(self.gamma.max(0.01) / 1.7);
|
||
let code = self.white + (lin * (1.0 - off) + off).max(1e-9).log10() * 0.6 / 0.002;
|
||
code / 1023.0
|
||
}
|
||
}
|
||
|
||
fn cineon(ctx: &EffectCtx, mut b: Buf) -> Buf {
|
||
let c = Cineon {
|
||
black: ctx.params.f("tenBitBlackPoint") as f32,
|
||
white: ctx.params.f("tenBitWhitePoint") as f32,
|
||
ib: ctx.params.f("internalBlackPoint") as f32,
|
||
iw: ctx.params.f("internalWhitePoint") as f32,
|
||
gamma: ctx.params.f("gamma") as f32,
|
||
rolloff: ctx.params.f("highlightRolloff") as f32,
|
||
};
|
||
let std = Cineon { black: 95.0, white: 685.0, ib: 0.0, iw: 1.0, gamma: 1.7, rolloff: c.rolloff };
|
||
let kind = ctx.params.e("conversionType");
|
||
b.img.map_straight(|v| {
|
||
v.map(|x| match kind {
|
||
0 => c.lin_to_log(x),
|
||
1 => c.log_to_lin(x),
|
||
_ => std.lin_to_log(c.log_to_lin(x)),
|
||
})
|
||
});
|
||
b
|
||
}
|
||
|
||
// ---- Colour LUTs (.cube) ----
|
||
|
||
#[derive(Clone, Debug, PartialEq)]
|
||
pub struct Lut {
|
||
pub n1: usize,
|
||
pub d1: Vec<[f32; 3]>,
|
||
pub n3: usize,
|
||
pub d3: Vec<[f32; 3]>,
|
||
pub dmin: [f32; 3],
|
||
pub dmax: [f32; 3],
|
||
}
|
||
|
||
/// Parse `.cube` text (TITLE, LUT_1D_SIZE, LUT_3D_SIZE, DOMAIN_MIN/MAX, data red-fastest).
|
||
pub fn parse_cube(text: &str) -> Option<Lut> {
|
||
let mut lut = Lut { n1: 0, d1: vec![], n3: 0, d3: vec![], dmin: [0.0; 3], dmax: [1.0; 3] };
|
||
let mut rows: Vec<[f32; 3]> = Vec::new();
|
||
let three = |it: &mut std::str::SplitWhitespace| -> Option<[f32; 3]> {
|
||
let a = it.next()?.parse().ok()?;
|
||
let b = it.next()?.parse().ok()?;
|
||
let c = it.next()?.parse().ok()?;
|
||
Some([a, b, c])
|
||
};
|
||
for line in text.lines() {
|
||
let line = line.trim();
|
||
if line.is_empty() || line.starts_with('#') {
|
||
continue;
|
||
}
|
||
let mut it = line.split_whitespace();
|
||
let key = it.clone().next()?;
|
||
match key {
|
||
"TITLE" => continue,
|
||
"LUT_1D_SIZE" => {
|
||
it.next();
|
||
lut.n1 = it.next()?.parse().ok()?;
|
||
}
|
||
"LUT_3D_SIZE" => {
|
||
it.next();
|
||
lut.n3 = it.next()?.parse().ok()?;
|
||
}
|
||
"DOMAIN_MIN" => {
|
||
it.next();
|
||
lut.dmin = three(&mut it)?;
|
||
}
|
||
"DOMAIN_MAX" => {
|
||
it.next();
|
||
lut.dmax = three(&mut it)?;
|
||
}
|
||
"LUT_1D_INPUT_RANGE" | "LUT_3D_INPUT_RANGE" => {
|
||
it.next();
|
||
let lo: f32 = it.next()?.parse().ok()?;
|
||
let hi: f32 = it.next()?.parse().ok()?;
|
||
lut.dmin = [lo; 3];
|
||
lut.dmax = [hi; 3];
|
||
}
|
||
k if k.starts_with(|c: char| c.is_ascii_alphabetic()) => continue,
|
||
_ => rows.push(three(&mut it)?),
|
||
}
|
||
}
|
||
let need1 = lut.n1;
|
||
let need3 = lut.n3 * lut.n3 * lut.n3;
|
||
if (lut.n1 == 0 && lut.n3 == 0) || lut.n1 == 1 || lut.n3 == 1 || rows.len() != need1 + need3 || lut.n3 > 256 {
|
||
return None;
|
||
}
|
||
lut.d1 = rows[..need1].to_vec();
|
||
lut.d3 = rows[need1..].to_vec();
|
||
Some(lut)
|
||
}
|
||
|
||
impl Lut {
|
||
fn norm(&self, c: [f32; 3]) -> [f32; 3] {
|
||
[0, 1, 2].map(|i| ((c[i] - self.dmin[i]) / (self.dmax[i] - self.dmin[i]).max(1e-9)).clamp(0.0, 1.0))
|
||
}
|
||
pub fn apply(&self, c: [f32; 3]) -> [f32; 3] {
|
||
let mut c = c;
|
||
if self.n1 > 1 {
|
||
let t = self.norm(c);
|
||
c = [0, 1, 2].map(|i| {
|
||
let x = t[i] * (self.n1 - 1) as f32;
|
||
let i0 = (x.floor() as usize).min(self.n1 - 2);
|
||
let f = x - i0 as f32;
|
||
self.d1[i0][i] + (self.d1[i0 + 1][i] - self.d1[i0][i]) * f
|
||
});
|
||
}
|
||
if self.n3 > 1 {
|
||
c = self.tetra(self.norm(c));
|
||
}
|
||
c
|
||
}
|
||
/// Apply with a choice of 3D interpolation: 0 nearest, 1 trilinear, otherwise tetrahedral.
|
||
pub fn apply_interp(&self, c: [f32; 3], interp: u32) -> [f32; 3] {
|
||
let mut c = c;
|
||
if self.n1 > 1 {
|
||
let t = self.norm(c);
|
||
c = [0, 1, 2].map(|i| {
|
||
let x = t[i] * (self.n1 - 1) as f32;
|
||
if interp == 0 {
|
||
return self.d1[(x.round() as usize).min(self.n1 - 1)][i];
|
||
}
|
||
let i0 = (x.floor() as usize).min(self.n1 - 2);
|
||
let f = x - i0 as f32;
|
||
self.d1[i0][i] + (self.d1[i0 + 1][i] - self.d1[i0][i]) * f
|
||
});
|
||
}
|
||
if self.n3 > 1 {
|
||
let t = self.norm(c);
|
||
c = match interp {
|
||
0 => self.nearest(t),
|
||
1 => self.trilinear(t),
|
||
_ => self.tetra(t),
|
||
};
|
||
}
|
||
c
|
||
}
|
||
fn nearest(&self, t: [f32; 3]) -> [f32; 3] {
|
||
let n = self.n3;
|
||
let s = (n - 1) as f32;
|
||
let i = |x: f32| ((x * s).round() as usize).min(n - 1);
|
||
self.d3[i(t[0]) + i(t[1]) * n + i(t[2]) * n * n]
|
||
}
|
||
/// Trilinear interpolation on the lattice (red fastest).
|
||
pub fn trilinear(&self, t: [f32; 3]) -> [f32; 3] {
|
||
let n = self.n3;
|
||
let s = (n - 1) as f32;
|
||
let idx = |x: f32| {
|
||
let v = x * s;
|
||
let i = (v.floor() as usize).min(n - 2);
|
||
(i, v - i as f32)
|
||
};
|
||
let (r, fr) = idx(t[0]);
|
||
let (g, fg) = idx(t[1]);
|
||
let (b, fb) = idx(t[2]);
|
||
let at = |dr: usize, dg: usize, db: usize| self.d3[(r + dr) + (g + dg) * n + (b + db) * n * n];
|
||
let l = |a: [f32; 3], b: [f32; 3], f: f32| [a[0] + (b[0] - a[0]) * f, a[1] + (b[1] - a[1]) * f, a[2] + (b[2] - a[2]) * f];
|
||
let c00 = l(at(0, 0, 0), at(1, 0, 0), fr);
|
||
let c10 = l(at(0, 1, 0), at(1, 1, 0), fr);
|
||
let c01 = l(at(0, 0, 1), at(1, 0, 1), fr);
|
||
let c11 = l(at(0, 1, 1), at(1, 1, 1), fr);
|
||
l(l(c00, c10, fg), l(c01, c11, fg), fb)
|
||
}
|
||
fn tetra(&self, t: [f32; 3]) -> [f32; 3] {
|
||
let n = self.n3;
|
||
let s = (n - 1) as f32;
|
||
let idx = |x: f32| {
|
||
let v = x * s;
|
||
let i = (v.floor() as usize).min(n - 2);
|
||
(i, v - i as f32)
|
||
};
|
||
let (r, fr) = idx(t[0]);
|
||
let (g, fg) = idx(t[1]);
|
||
let (b, fb) = idx(t[2]);
|
||
let at = |dr: usize, dg: usize, db: usize| self.d3[(r + dr) + (g + dg) * n + (b + db) * n * n];
|
||
let c000 = at(0, 0, 0);
|
||
let c111 = at(1, 1, 1);
|
||
let (w1, w2, w3, p1, p2) = if fr > fg {
|
||
if fg > fb {
|
||
(fr, fg, fb, at(1, 0, 0), at(1, 1, 0))
|
||
} else if fr > fb {
|
||
(fr, fb, fg, at(1, 0, 0), at(1, 0, 1))
|
||
} else {
|
||
(fb, fr, fg, at(0, 0, 1), at(1, 0, 1))
|
||
}
|
||
} else if fb > fg {
|
||
(fb, fg, fr, at(0, 0, 1), at(0, 1, 1))
|
||
} else if fb > fr {
|
||
(fg, fb, fr, at(0, 1, 0), at(0, 1, 1))
|
||
} else {
|
||
(fg, fr, fb, at(0, 1, 0), at(1, 1, 0))
|
||
};
|
||
// c = c000 + w1 (p1 - c000) + w2 (p2 - p1) + w3 (c111 - p2), w1 ≥ w2 ≥ w3.
|
||
[0, 1, 2].map(|i| c000[i] + w1 * (p1[i] - c000[i]) + w2 * (p2[i] - p1[i]) + w3 * (c111[i] - p2[i]))
|
||
}
|
||
}
|
||
|
||
type LutCache = Mutex<HashMap<String, Option<Arc<Lut>>>>;
|
||
|
||
/// Load (and cache) a LUT from inline `.cube` text or a file path.
|
||
pub fn load_lut(src: &str) -> Option<Arc<Lut>> {
|
||
if src.trim().is_empty() {
|
||
return None;
|
||
}
|
||
static CACHE: OnceLock<LutCache> = OnceLock::new();
|
||
let cache = CACHE.get_or_init(|| Mutex::new(HashMap::new()));
|
||
if let Ok(m) = cache.lock()
|
||
&& let Some(v) = m.get(src)
|
||
{
|
||
return v.clone();
|
||
}
|
||
let inline = src.contains('\n') || src.contains("LUT_");
|
||
let parsed = if inline { parse_cube(src) } else { std::fs::read_to_string(src.trim()).ok().and_then(|t| parse_cube(&t)) }.map(Arc::new);
|
||
if let Ok(mut m) = cache.lock() {
|
||
m.insert(src.to_string(), parsed.clone());
|
||
}
|
||
parsed
|
||
}
|
||
|
||
fn apply_lut(ctx: &EffectCtx, mut b: Buf) -> Buf {
|
||
let Some(lut) = load_lut(ctx.params.s("lut")) else {
|
||
return b;
|
||
};
|
||
b.img.map_straight(|c| lut.apply(c));
|
||
b
|
||
}
|
||
|
||
pub fn specs() -> Vec<EffectSpec> {
|
||
vec![
|
||
spec("ec.utility.growbounds", "Grow Bounds", vec![p("pixels", "Pixels", num(0.0), slider(0.0, 4000.0, 0.0, 500.0, 0))], grow_bounds),
|
||
spec(
|
||
"ec.utility.ccoverbrights",
|
||
"Out-of-Range Display",
|
||
vec![
|
||
p("channel", "Channel", Value::Enum(0), popup(&["Max RGB", "Red", "Green", "Blue", "Alpha", "Luminance"])),
|
||
p("highlightColor", "Highlight Color", col(1.0, 0.0, 0.0), ParamUi::Color),
|
||
],
|
||
overbrights,
|
||
),
|
||
spec(
|
||
"ec.utility.hdrcompression",
|
||
"HDR Highlight Compression",
|
||
vec![p("amount", "Amount", num(100.0), slider(0.0, 100.0, 0.0, 100.0, 1))],
|
||
hdr_compression,
|
||
),
|
||
spec(
|
||
"ec.utility.cineon",
|
||
"Cineon Converter",
|
||
vec![
|
||
p("conversionType", "Conversion Type", Value::Enum(1), popup(&["Linear to Log", "Log to Linear", "Log to Log"])),
|
||
p("tenBitBlackPoint", "10 Bit Black Point", num(95.0), slider(0.0, 1023.0, 0.0, 1023.0, 0)),
|
||
p("internalBlackPoint", "Internal Black Point", num(0.0), slider(-1.0, 2.0, 0.0, 1.0, 3)),
|
||
p("tenBitWhitePoint", "10 Bit White Point", num(685.0), slider(0.0, 1023.0, 0.0, 1023.0, 0)),
|
||
p("internalWhitePoint", "Internal White Point", num(1.0), slider(-1.0, 2.0, 0.0, 1.0, 3)),
|
||
p("gamma", "Gamma", num(1.7), slider(0.1, 5.0, 0.1, 5.0, 2)),
|
||
p("highlightRolloff", "Highlight Rolloff", num(20.0), slider(0.0, 100.0, 0.0, 100.0, 0)),
|
||
],
|
||
cineon,
|
||
),
|
||
// `lut`: a path to a `.cube` file (Effect Controls' Choose…, asked for when the effect is
|
||
// applied, as in After Effects), or the `.cube` text itself (commands/automation).
|
||
spec("ec.utility.applylut", "Apply Color LUT", vec![p("lut", "LUT", Value::Str(String::new()), crate::file_ui(&["cube"]))], apply_lut),
|
||
]
|
||
}
|
||
|
||
#[cfg(test)]
|
||
mod tests {
|
||
use super::*;
|
||
use crate::{Image, Params};
|
||
|
||
fn identity_cube(n: usize) -> String {
|
||
let mut s = format!("TITLE \"id\"\n# comment\nLUT_3D_SIZE {n}\n");
|
||
for b in 0..n {
|
||
for g in 0..n {
|
||
for r in 0..n {
|
||
let f = |v: usize| v as f32 / (n - 1) as f32;
|
||
s += &format!("{} {} {}\n", f(r), f(g), f(b));
|
||
}
|
||
}
|
||
}
|
||
s
|
||
}
|
||
|
||
fn run(id: &str, vals: &[(&str, Value)], img: Image) -> Buf {
|
||
let s = crate::find(id).unwrap();
|
||
let mut params = Params { values: s.params.iter().map(|p| (p.id.to_string(), p.default.clone())).collect() };
|
||
for (k, v) in vals {
|
||
params.values.insert(k.to_string(), v.clone());
|
||
}
|
||
let ctx =
|
||
EffectCtx { params: ¶ms, time: 0.0, layer_size: [img.width as f64, img.height as f64], seed: 1, adjustment: false, env: Default::default() };
|
||
crate::apply(s, &ctx, Buf { img, offset: [0.0, 0.0], scale: 1.0 })
|
||
}
|
||
|
||
fn test_img() -> Image {
|
||
let mut img = Image::new(8, 8);
|
||
for y in 0..8 {
|
||
for x in 0..8 {
|
||
let a = 0.5 + (x as f32) / 16.0;
|
||
img.set(x, y, premul([x as f32 / 7.0, y as f32 / 7.0, ((x + y) % 5) as f32 / 4.0], a));
|
||
}
|
||
}
|
||
img
|
||
}
|
||
|
||
#[test]
|
||
fn identity_lut_is_identity() {
|
||
for n in [2, 17] {
|
||
let lut = parse_cube(&identity_cube(n)).unwrap();
|
||
for c in [[0.1, 0.5, 0.9], [0.33, 0.77, 0.0], [1.0, 1.0, 1.0], [0.0, 0.0, 0.0], [0.42, 0.42, 0.9]] {
|
||
let o = lut.apply(c);
|
||
for i in 0..3 {
|
||
assert!((o[i] - c[i]).abs() < 1e-5, "n={n} {c:?} -> {o:?}");
|
||
}
|
||
}
|
||
let img = test_img();
|
||
let out = run("ec.utility.applylut", &[("lut", Value::Str(identity_cube(n)))], img.clone());
|
||
for (a, b) in out.img.data.iter().zip(img.data.iter()) {
|
||
for c in 0..4 {
|
||
assert!((a[c] - b[c]).abs() < 1e-5);
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn swap_lut_swaps_channels() {
|
||
let mut s = String::from("LUT_3D_SIZE 2\n");
|
||
for b in 0..2 {
|
||
for g in 0..2 {
|
||
for r in 0..2 {
|
||
s += &format!("{b} {g} {r}\n");
|
||
}
|
||
}
|
||
}
|
||
let lut = parse_cube(&s).unwrap();
|
||
let o = lut.apply([0.2, 0.5, 0.8]);
|
||
assert!((o[0] - 0.8).abs() < 1e-5 && (o[1] - 0.5).abs() < 1e-5 && (o[2] - 0.2).abs() < 1e-5, "{o:?}");
|
||
}
|
||
|
||
#[test]
|
||
fn bad_cube_is_passthrough() {
|
||
assert!(parse_cube("LUT_3D_SIZE 2\n0 0 0\n1 1 1\n").is_none());
|
||
assert!(parse_cube("LUT_3D_SIZE 2\n0 0 zz\n").is_none());
|
||
let img = test_img();
|
||
let out = run("ec.utility.applylut", &[("lut", Value::Str("LUT_3D_SIZE 3\n0 0 0\n".into()))], img.clone());
|
||
assert_eq!(out.img, img);
|
||
let out = run("ec.utility.applylut", &[("lut", Value::Str("/nonexistent/path.cube".into()))], img.clone());
|
||
assert_eq!(out.img, img);
|
||
}
|
||
|
||
#[test]
|
||
fn one_d_lut_with_domain() {
|
||
let lut = parse_cube("LUT_1D_SIZE 2\nDOMAIN_MIN 0 0 0\nDOMAIN_MAX 2 2 2\n1 1 1\n0 0 0\n").unwrap();
|
||
let o = lut.apply([0.5, 1.0, 2.0]);
|
||
assert!((o[0] - 0.75).abs() < 1e-5 && (o[1] - 0.5).abs() < 1e-5 && o[2].abs() < 1e-5, "{o:?}");
|
||
}
|
||
|
||
#[test]
|
||
fn cineon_round_trip() {
|
||
let c = Cineon { black: 95.0, white: 685.0, ib: 0.0, iw: 1.0, gamma: 1.7, rolloff: 20.0 };
|
||
for code in [100.0f32, 200.0, 350.0, 500.0, 600.0] {
|
||
let v = code / 1023.0;
|
||
let back = c.lin_to_log(c.log_to_lin(v));
|
||
assert!((back - v).abs() < 1e-3, "{code}: {back}");
|
||
}
|
||
assert!(c.log_to_lin(95.0 / 1023.0).abs() < 1e-4);
|
||
}
|
||
|
||
/// Interpret Footage ▸ Cineon Settings and the Cineon Converter's Log to Linear give the same
|
||
/// values for the same points and Highlight Rolloff when the converter's Gamma is 1.7 times
|
||
/// Cineon Settings' Gamma: the converter's default 1.7 and Cineon Settings' Gamma 1 both give
|
||
/// linear light (code 400 is 0.10251).
|
||
#[test]
|
||
fn cineon_converter_matches_interpret_footage() {
|
||
let codes = [0.0f32, 95.0, 250.0, 400.0, 560.0, 685.0, 800.0, 1023.0];
|
||
for (gamma, rolloff) in [(1.0, 20.0), (1.0, 0.0), (1.7, 50.0), (2.4, 50.0)] {
|
||
let mut img = Image::new(codes.len() as u32, 1);
|
||
for (x, code) in codes.iter().enumerate() {
|
||
let v = code / 1023.0;
|
||
img.set(x as u32, 0, [v, v, v, 1.0]);
|
||
}
|
||
let vals = [("conversionType", Value::Enum(1)), ("gamma", num(gamma * 1.7)), ("highlightRolloff", num(rolloff))];
|
||
let out = run("ec.utility.cineon", &vals, img);
|
||
let settings = effectcraft_project::CineonSettings { log_conversion: true, gamma: gamma as f32, rolloff: rolloff as f32, ..Default::default() };
|
||
for (x, code) in codes.iter().enumerate() {
|
||
let footage = settings.convert(code / 1023.0);
|
||
let effect = out.img.data[x][0];
|
||
assert!((footage - effect).abs() < 1e-6, "gamma {gamma}, rolloff {rolloff}, code {code}: {footage} vs {effect}");
|
||
}
|
||
}
|
||
let linear = effectcraft_project::CineonSettings { gamma: 1.0, rolloff: 0.0, ..Default::default() };
|
||
assert!((linear.convert(400.0 / 1023.0) - 0.10251).abs() < 1e-4);
|
||
}
|
||
|
||
/// The Cineon Converter gives the same values as before Cineon Settings shared its
|
||
/// log-to-linear function, at the default Gamma (1.7) and at Gamma 2.4, for all three
|
||
/// conversion types.
|
||
#[test]
|
||
fn cineon_converter_output_is_unchanged() {
|
||
fn log_to_lin(v: f32, gamma: f32, rolloff: f32, black: f32, white: f32) -> f32 {
|
||
let off = 10f32.powf((black - white) * 0.002 / 0.6);
|
||
let mut lin = (10f32.powf((v * 1023.0 - white) * 0.002 / 0.6) - off) / (1.0 - off).max(1e-6);
|
||
lin = lin.max(0.0).powf(1.7 / gamma.max(0.01));
|
||
let k = 1.0 - (rolloff / 100.0).clamp(0.0, 0.99);
|
||
if k < 1.0 && lin > k {
|
||
let w = 1.0 - k;
|
||
lin = k + w * (1.0 - (-(lin - k) / w).exp());
|
||
}
|
||
lin
|
||
}
|
||
fn lin_to_log(v: f32, gamma: f32, rolloff: f32, black: f32, white: f32) -> f32 {
|
||
let off = 10f32.powf((black - white) * 0.002 / 0.6);
|
||
let mut lin = v;
|
||
let k = 1.0 - (rolloff / 100.0).clamp(0.0, 0.99);
|
||
if k < 1.0 && lin > k {
|
||
let w = 1.0 - k;
|
||
lin = k - w * (1.0 - (lin - k) / w).max(1e-6).ln();
|
||
}
|
||
lin = lin.max(0.0).powf(gamma.max(0.01) / 1.7);
|
||
(white + (lin * (1.0 - off) + off).max(1e-9).log10() * 0.6 / 0.002) / 1023.0
|
||
}
|
||
let codes = [0.0f32, 95.0, 250.0, 400.0, 560.0, 685.0, 800.0, 1023.0];
|
||
for gamma in [None, Some(2.4)] {
|
||
for kind in 0..3 {
|
||
let mut img = Image::new(codes.len() as u32, 1);
|
||
for (x, code) in codes.iter().enumerate() {
|
||
let v = code / 1023.0;
|
||
img.set(x as u32, 0, [v, v, v, 1.0]);
|
||
}
|
||
let mut vals = vec![("conversionType", Value::Enum(kind))];
|
||
if let Some(g) = gamma {
|
||
vals.push(("gamma", num(g)));
|
||
}
|
||
let out = run("ec.utility.cineon", &vals, img);
|
||
let g = gamma.unwrap_or(1.7) as f32;
|
||
for (x, code) in codes.iter().enumerate() {
|
||
let v = code / 1023.0;
|
||
let want = match kind {
|
||
0 => lin_to_log(v, g, 20.0, 95.0, 685.0),
|
||
1 => log_to_lin(v, g, 20.0, 95.0, 685.0),
|
||
_ => lin_to_log(log_to_lin(v, g, 20.0, 95.0, 685.0), 1.7, 20.0, 95.0, 685.0),
|
||
};
|
||
let got = out.img.data[x][0];
|
||
assert!((want - got).abs() < 1e-6, "gamma {g}, type {kind}, code {code}: {want} vs {got}");
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn hdr_compression_behaviour() {
|
||
let mut img = Image::new(2, 1);
|
||
img.set(0, 0, [0.2, 0.3, 0.4, 1.0]);
|
||
img.set(1, 0, [4.0, 2.0, 1.0, 1.0]);
|
||
let out = run("ec.utility.hdrcompression", &[("amount", num(0.0))], img.clone());
|
||
assert_eq!(out.img.data, img.data, "Amount 0 leaves the image alone");
|
||
let out = run("ec.utility.hdrcompression", &[], img);
|
||
for o in &out.img.data {
|
||
assert!(o.iter().all(|v| v.is_finite() && *v <= 1.0), "{o:?}");
|
||
}
|
||
// Negative and huge input stays finite.
|
||
let mut img = Image::new(2, 1);
|
||
img.set(0, 0, [-0.5, 0.2, 3.0, 1.0]);
|
||
img.set(1, 0, [1e30, -1e30, 0.0, 1.0]);
|
||
let out = run("ec.utility.hdrcompression", &[("amount", num(50.0))], img);
|
||
assert!(out.img.data.iter().flatten().all(|v| v.is_finite()), "{:?}", out.img.data);
|
||
}
|
||
|
||
/// A linear value up to 1 as a code of After Effects' 16-bit PNG renders: the sRGB curve × 6553.5.
|
||
fn png_code(x: f32) -> f32 {
|
||
(if x <= 0.003_130_8 { 12.92 * x } else { 1.055 * x.powf(1.0 / 2.4) - 0.055 }) * 6553.5
|
||
}
|
||
|
||
/// Our own measurements of After Effects 2026 (32 bpc, no working space; 16-bit PNG renders
|
||
/// decoded back to linear): input pixel, then the output at Amount 100, 90, 84, 75, 50 and 25.
|
||
#[test]
|
||
fn hdr_compression_matches_measured_reference() {
|
||
const AMOUNTS: [f64; 6] = [100.0, 90.0, 84.0, 75.0, 50.0, 25.0];
|
||
#[rustfmt::skip]
|
||
let table: [([f32; 3], [[f32; 3]; 6]); 9] = [
|
||
([0.18005371; 3], [[0.1524; 3], [0.1550; 3], [0.1566; 3], [0.1590; 3], [0.1660; 3], [0.1729; 3]]),
|
||
([0.49365234; 3], [[0.3302; 3], [0.3450; 3], [0.3540; 3], [0.3674; 3], [0.4073; 3], [0.4492; 3]]),
|
||
([1.0058594; 3], [[0.5014; 3], [0.5427; 3], [0.5683; 3], [0.6085; 3], [0.7280; 3], [0.8602; 3]]),
|
||
([2.0058594; 3], [[0.6673; 3], [0.7637; 3], [0.8256; 3], [0.9232; 3], [1.2292; 3], [1.5892; 3]]),
|
||
([2.9941406; 3], [[0.7496; 3], [0.8962; 3], [0.9922; 3], [1.1465; 3], [1.6475; 3], [2.2604; 3]]),
|
||
([3.9941406; 3], [[0.7998; 3], [0.9929; 3], [1.1213; 3], [1.3303; 3], [2.0312; 3], [2.9141; 3]]),
|
||
([0.7998047, 0.099975586, 0.099975586], [
|
||
[0.6101, 0.0762, 0.0762], [0.6279, 0.0784, 0.0784], [0.6386, 0.0798, 0.0798],
|
||
[0.6547, 0.0817, 0.0817], [0.7012, 0.0877, 0.0877], [0.7496, 0.0937, 0.0937],
|
||
]),
|
||
([0.099975586, 0.7001953, 0.19995117], [
|
||
[0.0681, 0.4776, 0.1363], [0.0711, 0.4976, 0.1420], [0.0727, 0.5097, 0.1456],
|
||
[0.0754, 0.5287, 0.1508], [0.0832, 0.5829, 0.1662], [0.0912, 0.6397, 0.1825],
|
||
]),
|
||
([0.30004883, 0.5, 2.5], [
|
||
[0.1807, 0.3012, 0.9998], [0.1911, 0.3184, 1.1147], [0.1972, 0.3288, 1.1860],
|
||
[0.2071, 0.3453, 1.2995], [0.2359, 0.3931, 1.6484], [0.2669, 0.4444, 2.0482],
|
||
]),
|
||
];
|
||
let mut img = Image::new(table.len() as u32, 1);
|
||
for (x, (c, _)) in table.iter().enumerate() {
|
||
img.set(x as u32, 0, [c[0], c[1], c[2], 1.0]);
|
||
}
|
||
for (i, amount) in AMOUNTS.into_iter().enumerate() {
|
||
let out = run("ec.utility.hdrcompression", &[("amount", num(amount))], img.clone());
|
||
for (x, (input, want)) in table.iter().enumerate() {
|
||
let got = out.img.data[x];
|
||
for c in 0..3 {
|
||
let w = want[i][c];
|
||
// Up to 1: within 20 codes of the 16-bit PNG (the sRGB curve × 6553.5) the
|
||
// renders were read from. Above 1 the PNG's lookup table is not known: 0.1 %.
|
||
let ok = if w <= 1.0 { (png_code(got[c]) - png_code(w)).abs() <= 20.0 } else { (got[c] - w).abs() <= w * 1e-3 };
|
||
assert!(ok, "amount {amount}, input {input:?}, channel {c}: {} vs AE {w}", got[c]);
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
#[test]
|
||
fn grow_bounds_pads() {
|
||
let out = run("ec.utility.growbounds", &[("pixels", num(5.0))], test_img());
|
||
assert_eq!((out.img.width, out.img.height), (18, 18));
|
||
assert_eq!(out.offset, [5.0, 5.0]);
|
||
}
|
||
}
|