Appearance
Colour management means building an unbroken ICC profile chain: profile your capture device against a measured target, work on a hardware-calibrated monitor, embed the profile in every master file, and convert (never just assign) when you make derivatives. Done right, a grey card reads neutral, your ΔE against a reference chart stays low, and the same file looks consistent on any calibrated screen years from now. The three non-negotiable pieces are a capture profile, a calibrated display, and embedded profiles in every file.
Why does colour management matter for archives?
A digitised object is a surrogate that others will trust for research, exhibition and rights decisions. If the cyan of a watercolour or the foxing on a page shifts unpredictably between capture, screen and print, the surrogate lies. Colour management makes the surrogate measurable and defensible: you can state, with a number, how close your capture is to the original under reference conditions.
What gear and software do I need?
You need three categories of kit:
- A measured target in the capture frame — an X-Rite ColorChecker, a Golden Thread object target, or an IT8.7 chart for transmissive film.
- A hardware colorimeter or spectrophotometer for the display (Calibrite Display, X-Rite i1, Datacolor Spyder).
- Profiling software: basICColor, X-Rite i1Profiler, Argyll CMS (open source), or the camera-target tooling in Capture One / Lightroom.
How do I calibrate the monitor first?
The display is where you judge everything, so it comes first. Set a sensible target and let the colorimeter do the measuring.
text
Monitor calibration targets (typical archival desk):
White point: D65 (6500 K) # or D50 to match a print viewing booth
Luminance: 100–120 cd/m^2 # lower for dim rooms
Gamma: 2.2 (or L* for some workflows)
Ambient: neutral grey walls, ~32–64 lux, no coloured lightCalibration sets these states; profiling then writes an ICC profile describing the calibrated monitor. Re-run every 2–4 weeks.
How do I build and apply a capture profile?
Shoot or scan your target under your production lighting, then generate a device profile from it. With the open-source Argyll CMS the flow is:
bash
# 1. Extract patch values from a shot of the chart
scanin chart_capture.tiff ColorChecker.cht ColorChecker.cie
# 2. Build an ICC input profile from those measurements
colprof -v -D "Studio_2025_StrobeA" -qm chart_capture
# 3. Assign the profile to raw captures (assign, don't convert)
# then convert to your working space for the masterCrucially: assign the capture profile to interpret the raw numbers, then convert to your archival working space. Assigning re-labels the data; converting re-maps it. Mixing these up is the classic colour-shift bug.
Which colour space belongs in the master vs the derivative?
| Stage | Colour space | Why |
|---|---|---|
| Capture working space | Adobe RGB 1998 / eciRGB v2 | Wide gamut keeps saturated pigments and dye-based colour |
| Archival master (TIFF) | Same wide space, profile embedded | Future-proof, device-independent |
| Web/access derivative | sRGB IEC61966-2.1 | Matches most uncalibrated displays |
| Print/exhibition proof | Soft-proof via the printer/paper profile | Predicts the press or inkjet result |
Always embed the profile. An untagged file is ambiguous; viewers will guess, usually wrongly, that it is sRGB.
How do I verify colour accuracy with a number?
Measure colour error as ΔE (CIEDE2000) between your captured chart patches and the chart's reference values. FADGI and Metamorfoze set tolerances; a mean ΔE2000 under about 3, with no single patch wildly out, is a respectable target for cultural-heritage capture. Tools like the imaging plugin DICE/DCEval, Imatest, or the open delta-e checks in a custom script give you a pass/fail per session.
bash
# Conceptual: compare measured vs reference patches
python compute_delta_e.py --captured patches.csv --reference cc24.cie --formula ciede2000
# -> mean dE2000: 1.8, max patch dE2000: 4.1 (patch 13) -> PASS within FADGI 2-starWhat should I record in paradata?
Log the capture profile name, lighting setup, white point, the chart used, the date, and the measured ΔE for the session. This paradata lets a future curator reproduce or correct your colour and is part of a credible preservation record.
Key Takeaways
- Build an unbroken chain: capture profile → calibrated display → embedded profiles → controlled conversions.
- Calibrate then profile — they are different steps; calibration sets the state, profiling describes it.
- Use a wide working space (Adobe RGB / eciRGB) for masters, sRGB only for web derivatives.
- Assign a capture profile to interpret data; convert to change colour spaces. Never confuse them.
- Verify with ΔE2000 against a reference chart and keep it under roughly 3 for heritage work.
- A hardware colorimeter is mandatory; eyeballing the monitor is not colour management.
- Record lighting, profiles and measured error as paradata every session.
Frequently Asked Questions
What is colour management in a digitisation workflow?
Colour management is the chain of profiling, embedding and converting ICC profiles so a colour captured under your lights is reproduced faithfully on any calibrated display or print. It rests on a known capture profile, a calibrated monitor and embedded profiles in every file.
Which colour space should I use for archival masters?
Use a wide working space such as Adobe RGB 1998 or eciRGB v2 for masters, and embed the ICC profile in the file. Reserve sRGB for web access derivatives, where the smaller gamut matches typical displays.
Do I need a colorimeter to manage colour?
Yes, for the display side. A hardware colorimeter (X-Rite i1Display, Calibrite, or Datacolor Spyder) is the only reliable way to calibrate and profile your monitor; eyeballing it is not colour management.
How often should I re-profile my equipment?
Re-calibrate the monitor every 2–4 weeks, and re-shoot a camera/scanner target whenever you change lighting, lens, aperture or after about a month of stable use. Log every profiling event in your paradata.
What is the difference between calibration and profiling?
Calibration sets a device to a known, repeatable state (white point, luminance, tone response). Profiling then measures and describes that state in an ICC profile. You calibrate first, then profile the calibrated state.