Be it in spectroscopy or in photometry, one identifies the SED as a series of wavelengths and associated fluxes. In both cases, this is only a simplification of the fact that the measurement process convolves the true SED with a spectral response curve, yielding a transmitted flux at an effective wavelength. In spectroscopy, the response curve is almost invariably assumed to be Gaussian, with a σ determined by the slit width and the dispersing device. Therefore, in practice the distinction between the instrumental broadening and the broadening due to the intrinsic velocity dispersion of the astronomical object is not very sharp. To mimic the instrumental broadening, one should first convolve with the appropriate Gaussian and then resample onto the spectral bins. Care needs to be taken that the wavelength calibration of both models and data are better than a tenth of a pixel over the full wavelength range (this is not always the case, see e.g. Koleva et al. 2008).
In photometry, the response curve is much broader and therefore needs to be represented with more care, i.e. tabulated as a response function. The response function in turn depends on the detector quantum efficiency, the instrument transmission and the filter in use. Photometric calibration and response characterization is a vital task (see Koornneef et al. 1986; Landolt 1992, for just two prominent examples).
The signatures available for determination of the physical properties of galaxies of course depend on wavelength and on the achieved resolution. For example, in the optical many of the strongest features of galaxies can be adequately resolved at a resolution of R=λ∕Δλ~2000, while the low-resolution part of the Spitzer IRS can easily resolve PAH features at R~100. However, spectroscopy is more expensive in terms of telescope time, making photometry very attractive for obtaining large samples. In the last decade, successful use of narrow-band filters have blurred the distinction between spectroscopy and photometry, see for example COMBO-17 (Wolf et al. 2003), COSMOS (Scoville et al. 2007) and NEWFIRM (van Dokkum et al. 2009). Narrow-band filters have even been used to directly measure emission line equivalent widths (e.g. Kakazu, Cowie, and Hu 2007).