TL;DR: Getting an ICC profile installed correctly matters as much as the profile quality itself — a misassigned rendering intent or wrong connection space will shift neutrals by 3–5 ΔE even on a well-characterised press.
TL;DR: Our commissioning checklist requires 24-hour press stabilisation and a minimum of 3 consecutive characterisation print runs before any profile is approved for production use.
What “Installation” Actually Means in a Pressroom Workflow #
Most colour management guides treat ICC profile installation as a 30-second step: copy the file to the correct system folder, restart the RIP, done. On a standalone design workstation, that’s roughly accurate. In a pressroom workflow feeding multiple substrates, print processes, and downstream soft-proof approvals, installation is a commissioning sequence — not a file-copy operation.
The distinction matters because an ICC profile is only valid under the exact conditions it was built for: a specific substrate, ink set, screen ruling, press speed, and drying or curing configuration. When any of those variables drift, the profile becomes inaccurate without triggering any error message. The RIP keeps rendering, the press keeps printing, and the delta accumulates silently.
We manage this through what we call our PC-04 Profile Commissioning Record, which ties each .icc file to a frozen press configuration entry. No profile goes live on a production queue until that record is signed off by both the pre-press lead and the press operator responsible for the substrate class.
Pre-Installation Checklist — Press and Environment Conditions #
Before any characterisation data is used to build a production profile, the press environment must be stable. These are our minimum hold conditions before we begin the characterisation print run:
- Ink viscosity: within ±0.5 seconds of target (measured by Zahn cup #2 for flexo; equivalent for offset by inkometer)
- Substrate moisture content: within ±2% RH of the conditioning target — we condition all paperboard at 50 ±5% RH per ISO 187 before characterisation runs
- Press speed: locked at production speed (we do not profile at reduced speed — profiles built at 60% of rated speed are not valid at 100%)
- Ambient temperature: 20–24°C during measurement; spectrophotometer warm-up time minimum 15 minutes
- Ink trapping: verified by first-down / last-down trap test before characterisation patch set is printed
We also require a 24-hour press stabilisation window after any ink lot change before beginning new characterisation work. This is not conservative — an ink lot change can shift solid density by 0.05–0.12 D (densitometric) depending on pigment concentration tolerance at the mill.
Integration with RIP Software — Colour Space Assignment and Rendering Intent #
Installing a profile file is straightforward. Assigning it correctly inside the RIP is where most integration errors occur.
Comparison: Common ICC Integration Scenarios and Where Each Fails
| Integration Point | Correct Assignment | Common Error | Consequence |
|---|---|---|---|
| CMYK input profile (design file) | Match source profile used at file creation (e.g. ISOcoated_v2_eci) | Assigning press output profile as input | Double-application of gamut compression; neutrals shift warm |
| Output / device profile | Substrate-specific press profile (.icc built on that substrate) | Using generic profile for all substrates | SID / TVI errors up to 8% on uncoated vs coated stock |
| Rendering intent | Perceptual or Relative Colorimetric depending on image type | Applying Absolute Colorimetric to CMYK images | Forced paper white compensation distorts midtones |
| Proofing profile (soft proof) | Same output profile sent to brand/client | Outdated profile version from previous characterisation | Client approves colour that no longer matches press state |
| Connection space (PCS) | CIE Lab* via PCS inside ICC spec | Direct CMYK-to-CMYK transform bypassing PCS | Gamut mapping is skipped; out-of-gamut colours clip without warning |
For packaging specifically, we default to Relative Colorimetric with Black Point Compensation for all spot-colour-adjacent jobs. Perceptual intent looks better on photographic content but introduces unpredictable brand colour shifts for logo elements — the gamut compression moves differently depending on where the brand colour sits in the source gamut.
This is genuinely an area where practices vary across converters. Some pre-press houses set Perceptual as the global default and handle brand colours by exception. Others use Relative Colorimetric universally. We use a hybrid: Perceptual for photographic panels, Relative Colorimetric for structural graphic elements, set at the object level in the RIP. Not every RIP supports this cleanly — Kodak Prinergy and Esko Automation Engine both handle it well; some lower-cost RIPs require a workaround via PDF layer separation.
The Variable Nobody Checks — Spectrophotometer-to-Profile Geometry Alignment #
The overlooked factor in ICC profile integration is measurement geometry consistency. ISO 13655 defines M0, M1, and M2 measurement conditions. M1 (D50 illuminant, UV-included) is the current standard for print characterisation. M0 (legacy, UV content uncontrolled) was the default in most instruments pre-2010.
If the characterisation data was built under M1 conditions but your incoming QC spectrophotometer is set to M0, your verification measurements will consistently read lower L* on OBA-brightened substrates. The profile isn’t wrong. The measurement condition mismatch is creating a phantom delta.
We encountered this in an audit of incoming characterisation data from a brand partner in early 2023 — 14 out of 18 submitted profiles were built under M0 but the partner’s approval lab was measuring under M1. The apparent ΔE deviation on paper white was 2.1 units. The press was holding specification perfectly.
All our incoming spectrophotometers (X-Rite i1Pro 3 and Konica Minolta FD-9) are configured to M1 as the fixed measurement condition. We treat any profile submission without documented measurement condition metadata as requiring re-characterisation before production use.
Commissioning Steps After Profile Installation #
After the profile is assigned in the RIP, we run the following commissioning sequence before releasing for production:
- Print 3 consecutive IT8.7/4 or ECI2002 characterisation chart runs at full production speed — no reduced-speed validation
- Measure all patch sets under M1 conditions; compare to profile prediction using profile validation software (we use Profilemaker 5 or i1Profiler depending on substrate class)
- Calculate average ΔE2000 across patch set — our pass threshold is ≤2.0 ΔE2000 average, ≤4.0 maximum on any single patch (aligns with ISO 12647-2 press conformance requirements)
- Validate grey balance specifically: 50%C / 40%M / 40%Y combination must land within ±2 units on Lab* of the profile’s predicted neutral
- Archive the commissioning run sheets under the PC-04 record number linked to the profile file version
If the average ΔE2000 across three runs falls above 2.0, we do not iterate on the profile — we investigate the press condition first. A profile cannot compensate for an unstable press; trying to use profile adjustment as a substitute for press calibration is a common shortcut that creates a stack of “corrected” profiles with no clear lineage.
Timeline expectation: from press stabilisation hold to signed PC-04 record, allow 3–4 working days for a new substrate class. A re-characterisation on an existing substrate after an ink lot change runs 1–2 days.
Specification Notes for Brand Partners #
When you brief us on a new packaging project that requires colour-managed output, the two pieces of information that matter immediately are: the source profile your design files were built in (typically ISOcoated_v2 or GRACoL 2013 for North American brands), and the target substrate class (coated SBS, uncoated recycled board, or specialty laminate). Without both, we will flag your brief at the PC-04 gate and request clarification before beginning press preparation.
The brief gap that causes the most sample iterations is when brand files are built in an RGB colour space (sRGB or AdobeRGB) without a documented conversion to CMYK. We receive RGB packaged artwork and discover the conversion at the point of proofing. At that stage, colour decisions that should have been made in design are being made under time pressure in pre-press. Provide CMYK artwork with the source profile embedded, or explicitly brief us on which CMYK conversion to apply.
Our standard sampling timeline for a new profile-qualified substrate is 10–14 working days from receipt of approved artwork and substrate confirmation. That compresses to 5–7 working days if we are reprinting on an already-commissioned substrate with an existing valid PC-04 record.
FAQ
What file format should I send the ICC profile in, and does version matter?
Send the .icc file directly — .icm (Windows format) works identically but confirm the file is ICC specification v4. We do not accept v2 profiles for new projects; the rendering model differences create consistent grey balance shifts on our v4-configured RIP pipelines.
Can we use a profile built by our previous printer as the starting point?
It depends on whether the profile was built on the same substrate class you’re ordering from us. If your previous printer used the same SBS caliper and coating weight, the source profile is useful as a reference target. We still run our own characterisation and build our own press profile — we cannot put our press into conformance with a profile built on a different press.
Your commissioning checklist requires 3 print runs — does that mean 3× the substrate cost at sampling?
Yes, the characterisation runs consume substrate. For standard SBS board jobs the characterisation sheet count runs to roughly 150–200 A1-equivalent sheets per run, so 450–600 sheets total. We absorb this cost for production orders above 5,000 units. Below that threshold it is quoted as a pre-press line item.
How often do ICC profiles need to be revalidated?
Our practice is annual revalidation for all active production profiles, and immediate revalidation after any of these events: ink supplier lot change, press maintenance affecting impression or blanket, substrate supplier change, or RIP software update. Some converters only revalidate after formulation changes. We’ve found annual cycles catch creep that single-event triggers miss — particularly for long-running SKUs printed 4–6 times per year.
What happens if our brand colour falls outside the press gamut?
We flag it before production, not during. During profile commissioning we map your specified brand colours (in Lab* from your brand standard, or from a measured Pantone reference) against the profile gamut boundary. If any brand colour falls outside, we present the closest in-gamut match and the ΔE2000 deviation, and ask for written sign-off before proceeding. Gamut boundary tolerance we work to is ±3.0 ΔE2000 for secondary brand colours, ±1.5 for primary brand colours per our internal colour approval form CAF-12.
Planning a packaging project? Contact our team to request a complimentary specification review and sample quote.
Ran into exactly this with a flexo supplier in Wenzhou last year — they’d been running the same ISOcoated_v2 profile across both their coated and uncoated liner stocks, and we were seeing TVI errors closer to 10% on the uncoated runs before anyone flagged it. Took a full recharacterisation on the uncoated substrate and a new commissioning record tied to that specific press configuration before the neutrals came back in line.
The silent delta accumulation point is real — we caught a 4 ΔE neutral shift on a Darjeeling tin label run that had been drifting for three weeks before anyone flagged it, and the RIP hadn’t logged a single warning the entire time.
The rendering intent misassignment issue caught us badly on a pearlescent laminate carton run last spring — 47,000 units into a 60,000-unit job before QC pulled it, because Absolute Colorimetric was left on from a previous tissue substrate profile and the paper white compensation was quietly flattening our mid-highlight gradients across a cream base.