When you calibrate a color process, the tonal curve fixes one problem — density — but not the other. Your printed Cyan isn't exactly the reference Cyan. Your Magenta leans toward orange, your Yellow toward green. Each pigment has its own hue drift, and no black-and-white curve corrects it: a curve acts on luminance, not on color.
iScan goes after that drift. You show it a photo of two targets side by side — the reference target and your print of the same target — and it measures, channel by channel, how far your color strays from the target. It produces a tint profile that you then apply to your images, before generating your negatives.
#What you see
iScan is laid out in four numbered steps, in a dedicated tool.
1. Load the photo. You take a single photo containing both targets side by side: the printed reference target and your own print of that target. Natural light, flat, one shot of the two. You load it into iScan.
2. Align the targets. A grid of patches in a Venn diagram overlays the photo. You move and rotate the image under the grid — by finger (pinch, rotate) or mouse — so each patch lands on the matching area of the target. You do this for the reference, then for your print.
3. Profile. You click "Compute the profile." iScan compares the colors read on the two targets and shows a confidence score as a percentage, plus a per-channel table: the theoretical reference, the analyzed reference, your analyzed print, and the deviations — ΔH (hue deviation) and ΔE (overall color deviation). A high confidence means the measurement is reliable; a low one tells you to retake the photo.
4. Correct an image. You apply the tint profile to an image. iScan corrects the colors based on what it measured, and the corrected image feeds the color generation mode.
<!-- src: assets/iscan/app.js:20 (STEP upload/align/result/apply) ; app.js:226 buildProfileFromScans ; app.js:239 confidence % ; app.js:250 ΔH/ΔE table ; app.js:288 applyProfileToImageData -->#Why it's valuable
Because it separates two corrections that everything conflates. Tint and density are two distinct problems. Your black-and-white calibration curve handles contrast and density; iScan handles color. By treating them separately, you keep a clean tonal curve and add tint correction on top, without either polluting the other.
Because a single photo is enough. No spectrophotometer, no patch-by-patch measurement. One shot of the two targets placed side by side, and iScan derives the full profile. The hardware is your camera or your phone.
Because the profile is reusable. Once computed, you can save it and reapply it. As long as your chemistry and pigments don't change, the hue drift stays the same — you measure once, you correct as many images as you want.
#When you don't need it
On a monochrome process. Cyanotype, platinotype, Carbon Museum Black, Aquaprint Sanguine, bromoil, gumoil, resinotype — a single pigment, no tint to correct. iScan has nothing to measure. The Color Venn and the classic gray target cover these processes.
When your photo of the targets is poor. Uneven light, a crooked target, reflections: the profile's confidence drops and the correction becomes a gamble. iScan tells you so with its confidence score, but it can't invent a clean measurement from a dirty shot. Retake the photo flat, in diffuse light, before trusting the profile.
When your pigments already land true. If the deviation table shows low ΔE across all channels, your color is already faithful to the reference. Applying a near-zero correction brings nothing — keep the profile on hand for the day the chemistry drifts.
#Worth remembering
| Item | Value |
|---|---|
| Steps | 4 — load, align, profile, correct |
| Input | A single photo of the two targets side by side |
| Target | CMY Venn diagram (3 circles) |
| Output | Tint profile + confidence score (%) |
| Deviations measured | ΔH (hue) and ΔE (overall), per channel |
| Corrects | Color (pigment hue) |
| Doesn't correct | Contrast and density (→ black-and-white calibration curve) |
| Target processes | Aquaprint four-color, Carbon color three-color |
#The test
Print the reference target and your own print of that target in four-color. Place the two side by side under diffuse light and take a photo flat-on. Load it into iScan, align the patch grid on each target, then compute the profile. Look at the confidence score: above a good threshold, the profile is usable. Read the table — if your Magenta shows a high ΔH, it's leaning toward another hue, and that's exactly what the profile will pull back when you apply it to your images.