Monochrome film tonal response

Visible effect

Shadows gather, midtones separate, and highlights roll toward print white

A monochrome film image rarely maps scene luminance to display brightness with a straight line. Dark values gather into a toe, middle values separate more strongly, and bright values bend into a shoulder before the material or print reaches its lightest reproducible tone.

The image can also carry a warm, cool, sepia, or process-specific cast because its darkest and lightest tones are not physically perfect digital black and white. That tint belongs to the monochrome scale; it is not the same mechanism as colour-film dye-layer response.

Physics

Exposure and development produce optical density

Incident photons create a latent image in light-sensitive silver-halide grains. Development converts affected grains into metallic silver, and fixing removes undeveloped silver halide. The resulting negative has an exposure-dependent optical density. A positive print or scan introduces another transfer stage before the reader sees a bright image.

Very low exposures do not separate cleanly above base density; high exposures approach the material and process limit. Between them lies a straighter region whose slope controls tonal separation.

Mathematics

A bounded positive-image curve with explicit toe and shoulder

The playground starts with linear-light luminance Y and applies exposure E. Toe T reduces low-value separation, shoulder S gathers high values, and C changes the midpoint slope:

The final scalar interpolates between chosen black and white endpoint colours. This construction fixes zero and one and remains monotonic, but its parameters are artistic display controls rather than measured log-exposure and density values from a particular stock and process.

Shader

GLSL linear-light tone and tint pass

The runtime compiles the same bounded function shown below. Its wrapper performs aspect-fill sampling, sRGB decoding, and output encoding.

// WHAT: Map linear luminance through a bounded monochrome film-response curve.
// HOW: Apply exposure, toe, shoulder, and contrast in sequence, use the result
// to interpolate tinted endpoints, then mix the response with the source.
// WHY: Separating curve shape from endpoint colour makes each photographic
// control legible and keeps zero mix as an exact identity.
float filmTone(
  float luminance,
  float exposureStops,
  float toe,
  float shoulder,
  float contrast
) {
  // Exposure shifts the signal before the characteristic curve is shaped.
  float exposed = clamp(luminance * exp2(exposureStops), 0.0, 1.0);
  float toeExponent = 1.0 + clamp(toe, 0.0, 1.0) * 2.0;
  float shoulderExponent = 1.0 + clamp(shoulder, 0.0, 1.0) * 2.0;
  // Toe deepens/compresses shadows; shoulder rolls off highlights.
  float toeShaped = pow(exposed, toeExponent);
  float shoulderShaped =
    1.0 - pow(1.0 - toeShaped, shoulderExponent);
  float safeContrast = max(0.25, contrast);
  // This bounded ratio pivots contrast without sending endpoints outside [0, 1].
  float low = pow(shoulderShaped, safeContrast);
  float high = pow(1.0 - shoulderShaped, safeContrast);
  return low / max(low + high, 0.000001);
}

vec3 filmTonalResponse(
  vec3 sourceLinear,
  float exposureStops,
  float toe,
  float shoulder,
  float contrast,
  vec3 shadowTint,
  vec3 highlightTint,
  float effectMix
) {
  // The response is intentionally monochrome: colour comes from endpoint tints.
  float luminance = dot(
    sourceLinear,
    vec3(0.2126, 0.7152, 0.0722)
  );
  float tone = filmTone(
    luminance,
    exposureStops,
    toe,
    shoulder,
    contrast
  );
  vec3 monochrome = mix(shadowTint, highlightTint, tone);
  // Mix last so the effect can be disabled without changing the source.
  return mix(sourceLinear, monochrome, clamp(effectMix, 0.0, 1.0));
}

Why these steps are here

1. Decode before measuring luminance. The curve acts on linear-light channel values, not gamma-encoded display samples.
2. Keep endpoints anchored. Power curves meet at zero and one, avoiding a hidden lift or clamp when controls change.
3. Separate toe, shoulder, and midpoint slope. Each control changes the region named by the article and diagnostic curve.
4. Tint after the scalar response. Black and white endpoint colours describe a monochrome print scale rather than independent RGB correction curves.
5. Mix last. Zero mix is a tested identity; full mix reveals the modeled response.

Notes

• A calibrated model would use a measured characteristic curve in log exposure versus density, then model printing or scanning separately.
• The approximation does not simulate grain, halation, chemistry stains, flare, development unevenness, paper response, or scanner tone mapping.
• Converting RGB to luminance discards spectral and filter response. Real monochrome stocks and coloured lens filters render hues with different relative brightness.
• Colour-film response remains a separate effect because dye-layer coupling and three-dimensional colour correction cannot be represented by this scalar curve.

References