A three-color, extremely high signal to noise, precisely pointed, sub-arcsecond photometer would enable us to distinguish Earth-like exoplanets from other rocky, gassy, or icy worlds - if we had the right three wavelengths, and the ability to block out the primary star’s glare. Color-color discrimination of Earth-like planets has been posited for quite some time. Broadband filters are not seen to be precise diagnostics of planets with life, but rather broad but critical similarity or difference to the one habitable planet we know. Visible wavelengths (Vis) are advantageous due to relatively greater abundance of photons of that wavelength range from planets around Sun-like stars. Near-ultraviolet (UV) wavelengths may be advantageous due to ozone and scattering properties in Earth’s atmosphere that make Earth truly stand out from other known and modeled planets. Near infrared wavelengths (IR) help discriminate potential biological color contributions. We conducted an optimization exercise to arrive at three broadband filters that reliably separate modeled Earth-like, nominally habitable planets from other possible exoplanets. Criteria we use included: - Atmosphere/clouds of habitable worlds - Sea/Land proportions (ocean, granite, basalt, other surface material) - Vegetation (chlorophyll spectra and fictional other photosynthetic materials depending on wavelengths of parent stars etc.) - Other factors that can be spectrally modeled (planet-covering cities, world oceans, etc.) The optimized bands resemble previous work for exoplanets and the solar system, but underscore the advantage of UV wavelengths and indicate their potential utility for exoplanet identification and/or discrimination in concert with other exoplanet observations. An exoplanet survey that could quickly identify such planets could then be followed up by more detailed, longer term study by more intensive campaigns with more capable, but more resource or time constrained.