darktable preset inspired by IlfordPan F Plus 50

What the style changes

The style converts the photo to black and white the way Pan F Plus sees color, then adds its high contrast, crisp edges and a barely visible grain. What it is based on is detailed on the Ilford Pan F Plus 50 film page. All figures below come from renders made with darktable-cli 5.6.0 at 1000 px, measured in CIELab.

• Gray from the film’s spectral sensitivity. A color calibration instance named “N&B” mixes the channels with weights computed from Harman’s published spectrogram. The curve is integrated against the CIE 1931 color matching functions, after dividing out the 2850 K tungsten light it was measured under: this is how darktable builds its own film presets, such as “monochrome | ILFORD HP5+”.
  • Compared with a plain luminance conversion, and before any curve or grain, the sky gets lighter (L 55.2 → 59.5), while skin (79.3 → 73.2) and a pink T-shirt (47.8 → 45.0) get darker.
  • These values stay within 0.4 L* of darktable’s HP5+ preset. Pan F Plus puts slightly more weight on blue (Z 0.51 against 0.49).
• High contrast. The tone curve has a long toe and its steepest slope in the light tones. Contrast rises from 14.3 to 20.0 on a landscape, more than with the Fomapan 100 style (19.0).
• Deep blacks, bright whites. The darkest 5 % of the frame drop to L 0.0 on one photo and from 29.4 to 18.5 on another. The brightest 5 % rise from 75.2 to 81.0 and from 87.9 to 89.9; white walls stay white, where the Fomapan 100 style grays them.
• Midtones that lean dark. The sea loses 5.5 L* and the pink T-shirt 6.9. Skin comes out darker than with the Fomapan 100 style (76.6 against 79.5).
• Crisp edges. A local contrast instance at darktable’s default settings stands in for the edge contrast of the datasheet. We saw no halo along a stone arch or around a face.
• Barely visible grain. At 100 % on 2560 px exports, the grain is much finer than with the Fomapan 100 and Tri-X 400 styles. At 1000 px it almost disappears.
• A truly neutral gray. Chroma falls from 8.6 to 0.0: no color cast is left.

What it leaves alone

• Your base development. The style contains no exposure, no sigmoid or filmic and no LUT. Those settings stay as you made them.

• But it switches off your main color calibration. The style contains two color calibration instances:

  • a disabled one with no instance name, which replaces your main color calibration and switches it off. It is set to “as shot in camera” with CAT16, like darktable’s default, so you can switch it back on;
  • the “N&B” instance, which is added and does the black-and-white conversion.

  Why: when an active color calibration runs after the black-and-white conversion, its chromatic adaptation tints the gray. We measured a blue cast of b* −4 to −8 in our renders before adding the disabled instance. darktable’s built-in black-and-white presets also bypass chromatic adaptation, and film does not white-balance either.

• It also replaces your tone curve, local contrast and grain. When you apply a style from the lighttable or the darkroom, each of its modules replaces the instance of the same module with the same name, if the photo has one, and is added otherwise. The style’s tone curve, local contrast and grain carry no instance name, so they replace any settings you had already made in those modules. Your earlier settings remain in the history.

• Your pipeline order. The style file carries no module order list, so it does not impose one on your image.

Exporting with a style, or rendering with darktable-cli, works differently: every module of the style is added as a separate instance, and the disabled color calibration still switches off the existing one. Our measurements were made that way, on darktable’s default rendering. The details are in Install and apply darktable styles.

Limits

• An interpretation, not a calibration. No film was scanned or measured to tune the style.
• Close to HP5 Plus in the grays. Harman’s chart has no unit; like darktable, we read it as log sensitivity. With a scale 0.8 to 1.3 times the one we read, the X weight moves only between 0.246 and 0.254. On our test photos, the Pan F Plus and HP5 Plus weights barely differ: the style’s character comes from the curve, the local contrast and the grain.
• The curve. The datasheet gives it for one developer, and the reviews say the film’s contrast depends heavily on exposure and development. The print stage is not modeled. The style follows the majority of reviews (deep blacks), not those that find the film gentle.
• The grain. Harman publishes no RMS value, so the strength was set by eye. darktable’s grain is procedural noise: it imitates the texture of film grain, not the shape of real silver grains.
• The edge effect. darktable’s local contrast imitates the look of edge contrast; it does not simulate development at the edges.
• Bright photos. Whites go all the way to 1. On an overexposed portrait, 11 % of the pixels exceeded 250/255 at 1000 px, against none in the base render. Our sample images use −0.5 EV, as for the other styles.
• No halation. None of the sources we read mentions it, so the style does not simulate it.

Tips

• Lower the exposure by about half a stop on bright scenes if the whites clip.
• Too harsh? Lower the detail setting of the local contrast module, or switch it off.
• See the spectral conversion alone by switching off the style’s tone curve, local contrast and grain instances.
• Back to color: switch the main color calibration back on and switch off the “N&B” instance.
• Installation: see Install and apply darktable styles.