Templates from this work compared to the most popular IR SED templates normalized to L_{8 − 1000}. Left panel: low-z templates. In black: all of the templates computed in this work (Fig. 10). For better comparison, we plot only the templates of the different libraries whose properties (L_IR, α) match those of the HRS. In turquoise: templates of Dale & Helou (2002) with α ranging from 1.5 to 4. In orange: library of Chary & Elbaz (2001) with log (L_IR/L_⊙) < 10.9, in magenta: median templates of Smith et al. (2012a) binned in dust luminosity for log (L_IR/L_⊙) < 11, and in green: templates of Rieke et al. (2009) with log (L_IR/L_⊙) < 11. Right panel: high-z templates. In black: mean SED from HRS (Fig. 9) as a reference for z = 0 galaxies. In purple: MS templates and in red the starburst template from Magdis et al. (2012). In blue: main MS template from Elbaz et al. (2011); and in cyan: starburst template from Elbaz et al. (2011). In green: selected templates from Berta et al. (2013). At low-z, there is a large discrepency between the templates between 20 and 100 μm.