Meilian New Materials Patent and AI Colour Matching

👁️ The Old Way: Master Dyers "Gambling" with Their Eyes

A plastics factory gets a customer request: "I want this mango yellow, same as last time."

The traditional process goes something like:

1. The master dyer looks at a color card.
2. From experience, he says: "Add 0.8% yellow, 0.15% orange, 0.02% red."
3. Make a sample.
4. Check under the light box: Under sunlight it looks like a mango; under fluorescent light it looks like a pumpkin.
5. Adjust again, make another sample, waste another batch of material.

It's not uncommon for a single color to need a dozen rounds of adjustment. It wastes ink, wastes plastic, wastes time — and everything depends on the master dyer's "feel." When that person leaves, the knowledge is gone.

🌈 Color Isn't Just "Red, Yellow, Blue" — It's a Fingerprint Curve

Here's a counterintuitive point:

Your eyes see "yellow" not because the pigment is literally called yellow, but because at certain wavelengths it bounces back a lot of light, and at other wavelengths it bounces back very little.

What instruments measure is called a spectral reflectance curve:

• For each wavelength from 380 nm to 780 nm, what percentage is reflected?
• This curve is the physical ID card​ of a color.

Two things can look the same shade of yellow to the human eye but have completely different curves. That's when the classic disaster happens:

Metamerism​: They look like twins under daylight, but strangers under a different lamp.

Eco-friendly pigments are especially vulnerable to this.

Chrome-free yellows, cobalt-free blues, bio-based pigment pastes — they're not the same old heavy-metal pigments anymore. Their optical personalities are different:

• Some are more transparent
• Some are duller
• Some turn gray after outdoor exposure
• Some look greenish under LED but reddish under sunlight

So "the color looks right" doesn't mean it is right. You need the whole curve​ to match.

🤖 What Does Meilian New Materials' Patent Do?

Patent number CN122692660A, with a long title:

A Deep Learning-Based Method for Predicting Pigment Formulation Reflectance Curves.

In plain English:

Instead of having the master dyer guess first and then make samples,

let a neural network first "imagine"

"If I mix this formula, how will light bounce back at every wavelength?"

It uses a multi-branch neural network:

• One branch learns "which pigment"
• One branch learns "how much was added"
• One branch learns "plastic substrate / concentration / processing conditions"
• Finally, all branches merge to directly predict the reflectance curve

It doesn't just predict Lab values (the three-dimensional coordinates humans use). It goes all the way to the spectral level.

Benefits:

• First sample is much closer to target
• Less color shift under different light sources
• Fewer samples, fewer reworks, less waste
• Small batches of green formulations become economically viable

♻️ Why Is This a Game-Changer for Eco-Friendly Pigments?

Why have eco-friendly alternatives been slow to catch on? Not because they look bad. But because:

The customer wants "exactly the same as before,"

while the new pigment has "same color, different spectrum,"

and after three failed attempts by the master dyer, purchasing gives up.

For example:

• Chrome-free yellow: wants to replace chrome yellow, but brightness, hiding power, and heat resistance may differ
• Cobalt-free blue: wants to replace cobalt blue, but depth and outdoor stability are hard to copy
• Bio-based pigment pastes: great story, but batch-to-batch variation, lightfastness drift, and compatibility issues with resins

What AI color matching does is:

Instead of trying to perfectly replicate one old pigment with one new one,

it uses a bunch of compliant pigments plus an optimization model

to find a curve that looks the same to the eye, stays the same under different lights, and still fits the budget.

It's like going from:

"A master bartender mixing a cocktail from memory"

to: "A navigation app choosing your route based on traffic, gas prices, and road restrictions"

🎈 One-Sentence Summary

Meilian New Materials' patent didn't invent a new color.

It turned color matching from:

Master dyer's eyes + sampling machine + luck

into:

Spectrophotometer + neural network + reflectance curve + cost constraints

because in the future, selling pigments isn't just about selling a color card.

It's about saying: "I can legally recreate your original color using algorithms."