Two Monitors Showing Different Colours? Multi-Monitor Colour Mismatch, Panel Gamut, ICC Profiles & Calibration Explained

2026-10-25 ｜ GPU Repair ｜ About 21 minutes ｜ HK GPU Fix Editorial Team

重點摘要

Two monitors can show the same image in different colours because colour passes through content, colour management, driver output settings, monitor modes, and viewing conditions. This guide separates those layers, checks the free settings first, and explains calibration in order. The goal is displays that look close side by side, not measured perfection.

The Short Answer First

Two monitors side by side can show the same desktop, the same photograph, or the same video in visibly different colours, even when both monitors are working. That result does not, on its own, prove that a graphics card has failed or that a monitor is broken. Colour on a computer is not a single value that travels unchanged from a file to the eye. It passes through several layers, and each layer can shift the result. When two monitors differ at any one of those layers, the same input produces two different outputs.

The useful question is therefore not "which monitor is right" in the abstract, but "at which layer do these two displays diverge". Answering that question in order — comparison method first, then monitor mode, then driver output, then colour management, then content and application, then a swap test, and only then hardware calibration — prevents the two most wasteful moves: replacing a monitor that only needed a mode change, and buying a calibration instrument before the free settings have been checked.

This article assumes a specific starting position, and that assumption matters. Both monitors have a picture. Neither monitor flickers, shows no signal, or loses its picture after waking from sleep. The problem is purely a difference in colour appearance while both displays are running normally. If the picture itself is unstable, missing, or broken into bands, a different diagnosis applies, and the steps here should not be forced onto that symptom.

The Five Layers Between a File and the Eye

Colour reaches the eye through five layers. A difference at any one of them can make two monitors disagree, and the layers interact: a monitor mode can mask a driver error, and a colour profile can be correct for one application and ignored by another.

No single layer explains every colour difference. The diagnosis below works through them in the order that costs the least and changes the fewest variables at a time.

Panel Gamut: Wider Is Not Automatically More Accurate

The gamut of a display is the range of colours it can produce. sRGB is the most common standard for ordinary SDR content: web pages, office documents, and most games are produced with sRGB in mind. Wider gamuts such as DCI-P3 and Adobe RGB can show more saturated reds and greens, and that wider range is genuinely useful for content made in those spaces. The misunderstanding is to treat a wider gamut as automatically more accurate for everything.

When sRGB content is shown on a wide-gamut display without any colour conversion, the sRGB values are stretched into the wider range. The result is oversaturation: reds become redder, greens become greener, and skin tones shift toward red or orange. That is a mode mismatch, not a hardware fault. The display is faithfully producing a wider range; it is simply being given no instruction to limit itself to the range the content was made in.

An sRGB mode, sometimes described as an sRGB clamp, exists to solve exactly this problem. It limits the output of a wide-gamut display back to the sRGB range, so ordinary SDR content looks as its maker intended. The practical rule follows from the content, not from a preference for vividness: for everyday SDR use, an sRGB mode is the appropriate baseline; for HDR or wide-gamut work, the native or HDR mode is the appropriate choice. Neither mode is universally better. Accuracy and vividness are different goals, and pushing saturation higher is not the same as making colour more accurate.

Two further panel facts belong in any honest expectation-setting. First, two units of the same model can still differ. Backlight, panel batch, and ageing differ from unit to unit, so white point, brightness, and uniformity can differ even before any setting is touched. The assumption that the same model guarantees identical colour out of the box is false. Second, viewing angle matters. The same display can look different viewed from an angle than viewed straight on, and TN panels show this effect most strongly; IPS panels can show a glow effect at an angle. A comparison made with one monitor angled and the other straight on is not a valid comparison, however careful the settings are.

Driver Output Settings: The Free Layer Worth Checking First

Driver output settings deserve attention before any purchase because they are free, immediate, and reversible, and because they are set independently for each display. One graphics card driving two monitors can hold one display at one setting and the other display at a different setting, and the difference alone can explain the colour gap.

The central distinction is dynamic range. RGB Full uses the range 0 to 255 and is the normal range for a PC monitor. RGB Limited uses the range 16 to 255 in the sense that black is 16 and white is 235, and it is the normal range for television and video equipment. Both ends of the chain must agree:

• Graphics card output set to Full, monitor interpreting the signal as Limited: shadows are crushed to black and highlights blow out, because the monitor treats the darkest values as below its black point.
• Graphics card output set to Limited, monitor interpreting the signal as Full: black becomes dark grey, the whole picture looks washed out, and contrast falls.

HDMI deserves a specific check. Some drivers default to Limited over HDMI, because HDMI originated as a television interface, while DisplayPort more often defaults to Full. A pattern in which the DisplayPort monitor looks normal and the HDMI monitor looks washed out is a useful clue, but it is only a clue. The actual setting on each display must be checked; the connector type alone does not prove the cause, and not every HDMI connection will be set to Limited.

The driver locations differ by brand, and each brand must be checked in its own terms rather than assuming one path covers all hardware:

• NVIDIA: Control Panel, then Display, then Change Resolution, then Use NVIDIA Colour Settings, where output dynamic range, output colour format, and output colour depth are set.
• AMD: Display settings, where the pixel format is set, for example RGB 4:4:4 Full, described as the PC standard.
• Intel: Display settings, then the advanced section, where the quantisation range is set to Full.

Three further driver checks belong at this layer. First, brightness, contrast, gamma, and digital vibrance settings inside the driver may have been set differently for the two displays; a vibrance or saturation change alters overall appearance without correcting white point or gamma, and it can easily push two displays further apart. Second, video colour settings and desktop colour settings are separate in the driver, so a correct desktop does not prove that video playback is correct. Third, a driver update can reset these settings. If a colour difference appears immediately after a driver update, the driver settings should be checked before any other conclusion is drawn. Menu wording varies by driver version, so the manufacturer's documentation and the actual on-screen labels for the installed version are the final authority.

ICC Profiles: A Description Is Not a Calibration

An ICC profile, a file ending in .icc or .icm, is a description of a display's colour characteristics: its gamut, its white point, and its tone response. Colour-managed programs use that description to convert colour for that display. The distinction that matters most in multi-monitor work is between two things that are often confused.

Calibration is adjustment: using the monitor's on-screen-display controls to set brightness, contrast, RGB gain, white point, and gamma so that the output moves closer to a target. Profiling is measurement: using a colour measurement instrument to measure the actual output and generating a profile specific to that individual display, which the system and programs then use for conversion. A profile without measurement is only a description of a model, not a correction for the individual unit on the desk.

Manufacturer-supplied profiles describe the typical characteristics of a model. They do not describe the measured characteristics of the individual unit in front of the user, with its own backlight and ageing. That limitation is not a defect in the file; it is the nature of a model-level description. Only a profile generated from measurement of that specific display describes that display.

Windows colour management is set per display, and that per-display structure is where a common multi-monitor error occurs. In Windows Colour Management, the correct display must be selected, the option to use the user's own settings for that device must be enabled, and the correct profile must be set as the default for that display. Attaching the profile for display A to display B, or setting a profile while the wrong display is selected, produces a colour error that looks like a hardware problem but is purely a selection error.

A final limit must be stated plainly. An ICC profile only affects programs that use colour management. Programs without colour management, and some games and video playback paths, may ignore profiles entirely. In those paths, the result is governed by the monitor mode and the driver settings, not by the profile. A profile therefore cannot be described as making colour accurate across the whole system in every program. It improves the programs that honour it, and it leaves the others to the monitor and driver layers.

A Calibration Workflow in Three Levels, in Order

Calibration work should proceed from free to instrument-based, and the order should not be reversed. Each level has a different claim, and confusing the levels is how a rough visual match gets mistaken for a measured match.

Level 0 deserves more attention than it usually receives, because most failed comparisons fail here. The monitors should be warmed up before comparison or calibration. Both displays should be set to the same baseline, for example the same 6500K colour temperature or the same sRGB mode, before any judgement is made. Colour-temperature tools that shift the screen toward warm tones at night must be off during comparison, or they will be mistaken for a monitor difference. And the comparison image must be the same image, in the same program, at the same time, on both displays. Comparing a photograph in one program on one monitor with a video in another program on the other monitor proves nothing about either monitor.

Level 1, the built-in wizard, is a genuine improvement over an unadjusted pair, but its limit is structural: it relies on the eye. The eye is affected by room light, by contrast with the neighbouring display, and by fatigue. A visual match is therefore a reasonable goal for everyday dual-monitor use, and it should be described as exactly that — looking close side by side — rather than as measured equality.

Level 2 changes the basis from judgement to measurement. Both displays must be calibrated to the same target set. A commonly used target set is a D65 white point, a gamma of 2.2, and a luminance around 120 candela per square metre, with the important qualification that the target depends on the work: print and video work may use different targets, and no single target set is the only standard for every purpose. After calibration, the on-screen-display brightness and RGB gain should not be changed casually, because the generated profile describes the display as it was measured; changing the controls breaks that correspondence. Level 3, including three-dimensional lookup tables and displays with internal hardware calibration, is mentioned here only as a boundary: it is the appropriate territory when work demands measured consistency, and it is beyond what ordinary dual-monitor matching requires.

Diagnosis in Order: Narrow Down Before Changing Anything

The following sequence keeps one variable changing at a time. Changing several settings at once and then judging the result makes the result impossible to attribute, and it is the most common reason a colour problem appears to resist every fix.

1. Check the comparison method. Is it the same image, the same program, and the same time on both displays? Is one display in HDR mode and the other not? Is a night mode or a colour-temperature tool active on one display? Is one display viewed at an angle? If the comparison method itself differs, unify it first; no later conclusion is trustworthy until this step is clean.
2. Check the on-screen-display modes. Are both displays in the same mode? One display in a vivid or game mode and the other in an sRGB or standard mode will differ by design. Set both to the same baseline mode and compare again.
3. Check the driver output, display by display. Is the dynamic range, the output format, and the colour depth consistent? Check the HDMI display with particular care. Check both ends: the graphics card setting and the monitor's black-level or HDMI black-level setting must agree, because changing only one side leaves the mismatch in place.
4. Check system colour management. In Windows Colour Management, is the correct display selected? Is each display carrying its own correct profile, set as default? Has a profile been attached to the wrong display? Has a driver update reset the assignment?
5. Check the content and the program. Does the difference appear in every program, or only in one browser, one game, or one video player? A difference confined to a program without colour management points toward monitor mode and gamut, not toward a system-wide profile problem. A difference in every program points toward the monitor, driver, or profile layers.
6. Run a swap test, changing one variable at a time. Swap the two cables, and observe whether the problem follows the cable and output port or stays with the display. If the problem follows the cable or port, the driver, cable, or port setting is the stronger suspect. If the problem stays with the display, the monitor mode or the individual panel characteristics are the stronger suspect. Swap only one thing per observation.
7. Only then consider an instrument. If every setting above is consistent and a visible difference remains, hardware calibration is the appropriate next step. If calibration still leaves a large difference, or if the problem is confined to part of the screen, the boundary section below applies.

Boundary Cases: What This Method Does Not Cover

This article covers colour differences while both displays are running normally. Several neighbouring symptoms need different methods and should not be forced into this framework. A difference that appears only in HDR mode belongs with HDR tone-mapping settings. A problem that is specifically about banding in gradients belongs with colour depth and dynamic range, not with the colour-matching steps here. Blurriness from a non-native resolution belongs with scaling settings; this article assumes both displays are running at their native resolutions for comparison. Flicker, no signal, and failure to show a picture after waking from sleep belong with signal and standby diagnosis, not with colour matching. Naming these boundaries is part of the diagnosis: applying a colour-calibration workflow to a signal problem wastes the effort and hides the real fault.

Panel limits also set the expectation. Calibration narrows a gap; it does not promise zero measured error between two displays, and two units of the same model will not necessarily become identical. The honest target is that the two displays look close side by side under the same light and the same content. A promise of perfect equality would be false for the reasons already given: unit variation, ageing, and uniformity limits.

Consider professional inspection when more than one of the following signals is present, or when calibration has been done correctly and a large difference remains. The signals are: a colour cast confined to part of the screen, yellowing or darkening at the edges, lines, spots, or dark patches on the panel, a colour channel that is entirely missing so that the whole screen leans green or purple and cannot be pulled back through the on-screen-display or driver controls, or brightness and contrast driven to their limits without bringing the two displays close. Flicker, intermittent black screens, or loss of signal alongside a colour problem move the case out of this article's scope and toward signal or power diagnosis.

Readers should not open a monitor panel, replace a backlight, modify panel firmware, or overwrite system profiles with profiles from unverified sources. A wrong profile can make colour worse and is difficult to trace afterwards, because the error then sits in a file rather than in a visible control. Before seeking inspection, it helps to gather the two monitor models, the connection type for each display, screenshots of the driver settings, the on-screen-display modes in use, which programs show the difference, and the result of the swap test. That information allows an inspection to start from the layer already isolated, rather than repeating the whole sequence.

Misconceptions, Corrected in One Place

Buying two units of the same model does not guarantee identical colour. Unit variation, ageing, and differing on-screen-display or driver settings can all separate two identical models.

A wider gamut is not automatically better for every content type. Without conversion, wide-gamut output stretches sRGB content into oversaturation. Accuracy and vividness are different properties.

Installing an ICC file is not the same as calibrating a display. A manufacturer profile describes a model; only a measured profile describes an individual display, and only colour-managed programs use profiles at all.

Full range is not always correct and Limited range is not always wrong. The rule is agreement between the two ends. A television that expects Limited range should be fed Limited range at both ends, and that combination is correct.

Raising brightness or saturation is not calibration. Brightness and saturation change overall appearance; they do not set white point, gamma, or neutral grey steps, and adjusting saturation by eye alone can easily push a display further from neutral.

Calibration is not a one-time event. Backlights and panels change over time, and changing an on-screen-display setting, a cable, or a driver version is a reason to re-check the settings rather than assuming the previous calibration still holds.

An eye judgement is not a measurement. The eye is influenced by room light, by the neighbouring display, and by fatigue. Measured consistency requires an instrument; a visual match is a lesser, but still useful, goal for everyday use.

A colour difference does not, on its own, prove a graphics card fault. Most colour differences in this article's scope come from the settings layers — monitor mode, driver output, and profile assignment — and those layers should be checked before any hardware conclusion is drawn.

When Inspection Is Warranted, and What to Prepare

If the ordered checks above have been completed, the settings are consistent, and a visible difference remains that calibration cannot narrow, or if any of the boundary signals listed above are present, inspection is the reasonable next step rather than further adjustment by trial and error. Inspection in this context means a check of the display and the graphics output path against the information already gathered, not a promise of a particular outcome.

For readers in Hong Kong who reach that point, the workshop is at Shek Mun, Dali Plaza, Room 1209. The telephone number is 2186 8010 and the WhatsApp number is 6558 6806. Bringing the two monitor models, the connection types, the driver-setting screenshots, the on-screen-display modes, the list of programs that show the difference, and the swap-test result allows the check to start from the layer already isolated. Where a repair is relevant, fees fall within the published range of HK$400 to HK$2,800, subject to inspection quote. That range is not a fixed quote for any particular fault, and the applicable figure depends on what inspection finds.

Frequently Asked Questions

Why do two monitors of the same model show different colours?

Because unit variation, backlight ageing, and differing on-screen-display or driver settings can all separate two units of the same model. The same model is a starting point for similarity, not a guarantee of identical colour. Checking both displays at the same baseline mode and the same driver settings removes the settings portion of the gap, and calibration narrows what remains.

Should I set both monitors to their most vivid mode to get the best colour?

No. Vivid modes raise saturation for effect, which is a different goal from accuracy. For ordinary SDR content, an sRGB or standard mode is the appropriate baseline, because it keeps sRGB content inside the range it was made for. A vivid mode on one display and a standard mode on the other will differ by design.

My HDMI monitor looks washed out but my DisplayPort monitor looks fine. Is the HDMI monitor faulty?

Not necessarily. Check the dynamic range at both ends for the HDMI display first: the graphics card output and the monitor's black-level setting must agree. A Limited output read as Full produces exactly the washed-out, low-contrast appearance described here, and correcting the setting at one or both ends resolves it without any hardware change.

Does installing the manufacturer's ICC profile calibrate my monitor?

No. A manufacturer profile describes the typical characteristics of a model. It does not measure the individual display on the desk, and it only affects programs that use colour management. True calibration requires adjusting the display controls and then generating a profile from measurement of that specific display.

Why does the colour difference appear in one program but not in another?

Because colour management is per-program. A program that uses profiles will convert colour for each display, while a program without colour management sends values straight through, leaving the result to the monitor mode and driver settings. A difference confined to one program points toward that program's colour path and the monitor mode, rather than toward a system-wide fault.

How close can two monitors realistically get?

Close enough that they look consistent side by side under the same light and the same content, which is the correct goal for everyday dual-monitor use. Measured zero error between two displays is not a realistic promise, because unit variation, ageing, and uniformity limits remain after calibration. For work that demands measured consistency, instrument calibration to a shared target set is the appropriate standard.