## Sample fabrication:

The FROC films were deposited on a glass substrate (Micro slides, Corning) using electron-beam evaporation for Ge (3 Å s− 1) and TiO2 (1 Å s− 1) pellets and thermal evaporation Ag (20 Å s− 1), with the deposition rates specified for each material. The silica capped FROC films were deposited on a glass substrate (2948, Corning) using electron-beam evaporation for Ge (0.5 Å s− 1), TiO2 (1 Å s− 1), and SiO2 (0.8 Å s− 1), and DC magnetron sputtering for Ag (2 Å s− 1). All deposition materials were purchased from Kurt J. Lesker. Deposited layer thicknesses were measured with spectroscopic ellipsometry (J. A. Woollam).

### Numerical Calculation Of The Reflection And Absorption Spectra:

Numerical reflectance and absorbance spectra were generated using a transfer matrix-based simulation model written in Mathematica and Python. Spectral optical constants for the multiple materials were obtained variously from the Brendel-Bormann model (Ag), fits to the experimental materials (SiO2, TiO2), and an amorphous experimental model for Ge. Transmittance was zero for all structures. Absorbance was calculated as the complementary to reflectance, or \(A = 1-R\).

**Reflection measurements**: Experimental angular reflectance measurements were performed using a variable-angle high-resolution spectroscopic ellipsometer (V-VASE, J. A. Woollam). Sample transmittance was zero for all angles and wavelengths.

### Color Analysis:

The reflectance spectra to CIE 1931-xyz colorspace conversation was performed in Python utilizing interpolations of the standard observer distributions.21,22 The XYZ tristimulus values are given as:

\(X = \frac{{\int }_{\lambda }^{ } S\left(\lambda \right) \alpha \left(\lambda \right) R\left(\lambda \right) d\lambda }{{\int }_{\lambda }^{ } S\left(\lambda \right) \beta \left(\lambda \right) d\lambda }\) , \(Y = \frac{{\int }_{\lambda }^{ } S\left(\lambda \right) \beta \left(\lambda \right) R\left(\lambda \right) d\lambda }{{\int }_{\lambda }^{ } S\left(\lambda \right) \beta \left(\lambda \right) d\lambda }\), and \(Z = \frac{{\int }_{\lambda }^{ } S\left(\lambda \right) \gamma \left(\lambda \right) R\left(\lambda \right) d\lambda }{{\int }_{\lambda }^{ } S\left(\lambda \right) \beta \left(\lambda \right) d\lambda }\),

with *S* as the illuminant spectrum, *R* as the spectral reflectance, *k* as a constant factor, and \(\alpha , \beta , \gamma\) as the standard observer functions. The integration is over the visible spectrum. The CIE 1931-xyz values are then given as:

\(x = Y/N\) , \(y = Y/N\), and \(z = 1-x-y\),

for \(N = X+Y+Z\). Note that at constant luminance, the chromaticity is defined by *x* and *y*.

Color swatch arrays were generated by transforming the calculated chromaticity values into their sRGB equivalents using a matrix transform calculated from reference primaries with the D65 reference white and sRGB companding (IEC 61966-2-1 standard). Colors were generated with matplotlib in Python.

Excitation purity is calculated as:

$$p = | s - w| / |d - w|$$

where *s*, *w*, and *d* are the CIE 1931 (x, y) coordinates for the measured spectra point, white point, and dominant wavelength point, respectively.

Total CIE x-y space coverage is calculated as the area of the smallest convex hull encompassing all of the desired CIE (x, y) points. Area is presented relative to the area of the full visible light color gamut. Sufficient resolution was obtained in the numerical simulations to approach a smooth hull.