Multispectral Sensing
At 15°, 45° and 110°, 31 spectral sampling points capture each color's spectral fingerprint with precision.

Accurate data capture is the starting point for algorithmic reasoning
At 15°, 45° and 110°, 31 spectral sampling points capture each color's spectral fingerprint with precision.
Four high-precision camera angles analyze grain, sparkle and color travel, reconstructing HDR paint-surface images and quantifying microtexture.

Three core algorithm capabilities make color accuracy practical
Our solution combines spectral and texture image multimodal sensing to reconstruct surface appearance, bringing color evaluation closer to human vision.
Meeting the metric does not guarantee visual accuracy

We design the entire color-matching process as an end-to-end differentiable optimization workflow. From color difference to each toner adjustment, every step is calculated and propagated by algorithms.
From manual trial and error to global algorithmic optimization

End-to-end modeling maps formulas directly to color appearance without intermediate Lab or LCH metrics. Its core capability is reconstructing unseen formula appearances with high accuracy.
Database scale is the foundation; generalization is the real technical barrier

Simple, accurate, efficient and portable for many application scenarios
Spectral + texture sensing captures original colors precisely.
Access formulas precisely from a 5M+ color database.
Easy to carry, simple to operate and ready whenever you need it.
Continuous self-learning evolution makes every use more accurate.
An integrated workflow makes color matching simpler.
Save color-matching time and reduce operating costs.

Simple, accurate, efficient and portable for many application scenarios