-
Products & Solutions
-
Support & Training
-
Follow Us
-
- SUBSCRIBE
How Infinity Enhances Display Uniformity
At the system-architecture level, the Infinity solution establishes a dedicated uniformity-enhancement framework through coordinated optimization of driver ICs, algorithms, and the control chain. Rather than relying on single-point compensation, it applies multidimensional dynamic correction to maintain full-screen consistency across gray levels, temperatures, and operating conditions. Infinity addresses uniformity from three dimensions:
l Low-gray correction: Addresses the dispersed luminance response, dark spots, and banding associated with conventional driving in the low-gray region.
l Full-grayscale response correction: Provides precise linear mapping across the entire dynamic luminance range, from black to white.
l Adaptive thermal compensation: Compensates and balances current drift and color-temperature shift caused by temperature rise in real time.
Together, these three technologies deliver a step change in luminance consistency, color uniformity, and visual stability, providing a systematic path to image-quality optimization for LED displays.
a) Low-Grayscale Correction
Ideally, a PWM-modulated display has a linear relationship between luminance and input gray level, and Mura has a consistent appearance across gray levels. In the low-gray region, however, limits in current-output accuracy and channel turn-on timing exacerbate non-uniformity. Low-gray uniformity is particularly challenging because conventional PWM is constrained by minimum pulse width and timing resolution; even a small current error can create a visible luminance-response difference, causing dark spots and banding.
Figure 47.
As illustrated, at low gray levels the one-LSB pulse width is extremely narrow. Variation in channel turn-on timing shifts the rising edge, falling edge, and effective on-time among ICs and LED packages. Together with variation in LED epitaxial emission characteristics, this widens the inter-pixel difference at low luminance.
Low-gray correction addresses this device-dependent non-uniformity. A single, non-blended subfield is measured at a low gray level to capture its Mura distribution. The measured luminance error is then compensated by adjusting the pulse width of that individual subfield. First, the luminance increment per unit compensation coefficient is calculated:
step = [L(c0 + N) - L(c0)] / N
Using the target luminance and the measured luminance of each LED package, the required low-gray correction coefficient is calculated as follows:
coeff = [L_target - L_LED] / step
With pulse-width compensation based on the low-gray correction coefficient, the luminance of each LED package is equalized at one pulse width. The compensation mechanism is shown below:
Figure 48.
As shown above, pulse-width compensation brings the luminance of different ICs and LED packages into convergence. As the gray level increases, subfield blending progressively reduces the effect of the low-gray correction coefficient on luminance and uniformity; its influence becomes negligible in the medium- and high-gray ranges.
b) Full-Grayscale Correction
Variations in LED devices, drive currents, package structures, and thermal environments can cause small luminance and chromaticity differences even when pixels receive the same input gray level. These differences may be inconspicuous at low gray levels but accumulate as gray level rises, eventually causing luminance non-uniformity, grayscale discontinuities, and color distortion across brightness ranges. Conventional luminance and chromaticity correction is typically performed at only selected gray levels, such as peak luminance or mid-gray, and therefore cannot ensure linear consistency across the full range. In addition, the LED current-to-luminous-flux relationship is inherently nonlinear, and response-curve shapes differ among devices. Single-point or sparse-point correction therefore cannot make the entire luminance-response curve conform to the target electro-optical transfer function (EOTF). The actual relationship between LED-display luminance and gamma is nonlinear, as shown below:
Figure 49. Ideal Response Curve (Left) and Actual Response Curve (Right)
The Infinity full-grayscale correction system incorporates two key techniques:
l Four-Point Calibration
l Piecewise Linear Interpolation
Four-point calibration uses a precision measurement system to acquire grayscale data and determine the display's actual luminance-output characteristic. Four key gray-level pivot points - for example, dark, midtone, highlight, and transition regions - are selected as calibration reference points across the full grayscale range. Driver-IC behavior is considered to avoid gray levels affected by rebound or subfield merging. A correction coefficient at each pivot quantifies the deviation between actual and target luminance.
The correction coefficient at any gray level of a given pixel can be determined by linear interpolation:
C(g) = C_i + (C_{i+1} - C_i) * (g - g_i) / (g_{i+1} - g_i)
After the four correction coefficients have been acquired, the piecewise-linear interpolation algorithm divides the response curve into multiple short linear segments. Within each segment, LED luminance is approximately linear with the input signal. The algorithm accurately extends the measured deviation information to every gray level, enabling precise compensation across the full grayscale range.
For each measured input gray level, the algorithm dynamically derives an independent correction coefficient from the neighboring calibration pivots. This ensures that each gray-level output closely conforms to the target EOTF, substantially improving low-gray accuracy and uniformity while optimizing overall visual performance.
This approach overcomes the limitation of conventional one-point or sparse-point correction, which can ensure accuracy only at selected gray levels. It allows the LED display to maintain stable, balanced grayscale performance in dark regions, transition regions, and highlights, thereby achieving true full-grayscale correction.
c) Adaptive Thermal Compensation Technology
Adaptive thermal compensation analyzes the factors influencing LED temperature and builds a thermal-distribution model using heat-transfer theory and mathematical solution methods. Combined with sensors and a dedicated algorithm, it forms a complete adaptive thermal-compensation solution.
The system calculates the thermal distribution of the full display in real time and dynamically adjusts its compensation strategy. It maintains accurate cabinet-level color, color temperature, and luminance while reducing temperature-dependent Mura variation.
Fundamentally eliminating thermal effects requires coordination among LED packages, PCB design, mechanical design, and power components. This section focuses on how Infinity's automatic thermal-effect compensation mitigates residual thermal effects in the current system.
Using real-time display-state data, adaptive thermal compensation predicts thermal distribution and applies thermal compensation to pixel gray values. This reduces thermal Mura and maintains display quality equivalent to a no-heat-accumulation condition: uniform luminance and chromaticity with no perceptible color difference.
l Operational Flow of Adaptive Thermal-Effect Compensation:
Figure 46. Adaptive Thermal Compensation Solution
The system uses colorimeters, cameras, temperature sensors, and other devices for calibration-data acquisition. Bayesian estimation, fuzzy-state estimation, and model predictive control are employed to construct luminance/chromaticity degradation models and per-pixel thermal-compensation models. Together with the thermal-distribution prediction model, they form the complete adaptive thermal-compensation system.
The system calculates the thermal distribution of the full display in real time and dynamically adjusts the compensation strategy, ensuring accurate presentation of cabinet-level color, color temperature, and luminance.
l Construction of the Thermal-Distribution Prediction Model
First, collected calibration data are used to build the thermal-distribution prediction model. Real-time per-pixel temperature is then estimated from the currently displayed content. An LED package's temperature is governed principally by the heat it generates and the heat it dissipates.
The LED thermal balance can be expressed as:
where Q_in and Q_out denote the heat generated within the LED package and the heat dissipated to the environment during operation, respectively.
The heat generated within an LED is approximately linear with display luminance: a brighter display produces more heat. Because display luminance is, in turn, approximately related to the displayed gray level, analysis of image gray levels enables estimation of internally generated heat.
Q_in = kL
L: current luminance of the LED package (nit).
k: LED self-heating coefficient, representing heat generated per unit luminance (W/nit).
Heat dissipation to the external environment depends on the physical structure and materials of the LED display. Some regions have better thermal paths, while others are thermally constrained.
Q_out = eta A Delta T
eta: effective heat-transfer coefficient of the LED package [W/(m^2 K)].
A: effective area normal to the heat-flow direction (m^2).
Delta T: temperature difference between the LED package and its surroundings (K).
l Thermal-Effect Compensation Model
L_out = L_in + (T_t - T_i) * RatioV
The thermal-distribution prediction model yields the per-pixel temperature T_i. T_t is the preset target temperature; RatioV is the luminance attenuation coefficient as a function of temperature; and L_out and L_in are the compensated and uncompensated luminance, respectively.
This compensation drives luminance at any operating temperature to the luminance that would occur at T_t.
After compensation, the display maintains its original uniformity, stable color temperature, and stable luminance throughout factory commissioning, customer acceptance, and field operation. At commercial exhibition and virtual-production sites, it consistently reproduces natural color and accurate images.
