Notan

When you import a photo into Calibration Flow, the app immediately suggests a histogram correction based on the darkest and lightest points found in the image. That's useful. But you don't really know what you just asked the app to do — does your image have enough contrast for the process? Is a certain result even possible? Are there weird tonal masses, poorly distributed?

Notan mode exists to answer those questions before you go further. It groups the pixels in your image into value (or color) families, and shows them to you as visible masses. No abstract histogram. A concrete tonal segmentation that lets you see the raw structure of your photo in three to fifteen distinct masses. What you're after is understanding the image's intent before you correct it.

#What you see on screen

You click the Notan button in the tonal panel. The control bar expands into three groups: Mode, Values, Grain.

Mode: Value or Color. In Value mode (the default), Calibration Flow segments your image into luminance families — everything that's mid-gray together, everything that's dark gray together, and so on. Each family displays as a gray mass representing its average value. In Color mode, the system analyzes chromatic distance in Lab space and re-segments pixels based on their hue proximity (ΔE) — you then see each family with its true average color. Color mode is four times slower (250 ms vs immediate L* segmentation) and includes a 140 ms debounce delay.

Values: 2 to 15 families. Default is 3. This slider groups your 15 potential luminance zones into N contiguous families of equal size. At 2, you have only two masses: very dark and very light. At 15, you approach a classic histogram, with one family per luminance slice. Increase the number to see more tonal gradation; lower it to see raw composition.

Grain: 0 to 5, spatial fusion. Default is 2. This slider eliminates tiny isolated spots (speckles) by merging connected components that are too small. At 0, you keep all fine detail — a tiny spot of light in a shadow zone stays visible. At 5, you clean to the maximum, the masses become smooth and refined. The calculation uses adaptive surface thresholds: each level represents an enlargement factor of the minimum area before fusion.

As you drag the sliders, you see the colored overlay recalculate in real time. Each tonal family displays in its own color (if Color mode) or gray (if Value mode). At the bottom, you also see the swatches: one swatch per detected family, showing its average value or color. The classic histogram, when you're in Value mode, also recolors — each bar takes on the tint of the tonal family it belongs to.

#Why it matters

To diagnose before you correct. Before you send your image down the correction pipeline (curve, auto-level, etc.), you now know what you really have: seven clear tonal masses? Only three? A block of shadows and a block of highlights with no gradation? This raw vision tells you whether the correction you want is even possible. If you have only two tonal families (shadows + highlights), raising Values to 15 will never give you a rich midtone image — you just learned that your image doesn't have midtones.

To verify process compatibility. Alternative processes (cyanotype, gum print, carbon) don't render fine detail — they work in discrete value masses. Before you scan your photo for calibration, launch Notan mode, set Values to 5–6, Grain to 3, and look: that's exactly what your final print will show. If it looks awful in five tonal masses, the print will too. The better you know this before you scan 50 targets, the more time you save.

To progress through the workflow. This is an upstream tool: you understand your image right at import, you choose a strategy (correct the midtones? Push contrast? Accept poor tonal definition?), and then you descend into the correction tools knowing what awaits you.

#When to skip it

You just need to correct with a curve. Notan mode is introspection. If you're after a specific curve (no prior analysis, you already know where you're going), the tool becomes a distraction. Disable it and work the curve directly.

The image is already familiar. A standard studio portrait, a photo already well-exposed that you recognize — no need to check tonal structure. Trust your photographer's instinct and move to correction.

You accept the process's limitations. On a process that renders only two or three masses (e.g., a very pronounced micro-contrast technique), knowing there are only two masses changes nothing — you move forward anyway. In that case, set Notan aside so you don't mentally block yourself.

⚠️ Technical limit: downscale. Notan calculation shrinks your image to max 340 px wide for performance. If you're hunting fine, localized detail (a texture, a watermark, grain), Notan will hide it by this reduction. For broad diagnosis of tonal architecture, it's transparent; for local precision, you need to zoom manually or check the full histogram.

⚠️ Color mode can get slow. If you pick Color mode and drag Values quickly from 2 to 15, you'll feel latency. Chromatic clustering (ΔE²) is 4× more expensive than L* segmentation. Value mode is instant. Keep Color for when you want to check chroma, not for rapid exploration.

#Worth remembering

#The test

Load a photo with strong chiaroscuro (e.g., a portrait with a dark background). Launch Notan in Value mode, Values = 3, Grain = 2. You should see three clean masses: deep shadows, face midtones, and maybe a halo of light. Drag Values up to 10. You'll gradually see the midtones subdivide and more nuance appear. Now drag Grain up to 5. Tiny isolated spots (grain speckles, fine hair) vanish, and the masses become smooth. That's Notan: you just saw exactly what the alternative process will render, depending on how you set the controls.