We show that the diversity in the density slope of the dense wind due to non-steady mass loss can be one way to explain the spectral diversity of Type II luminous supernovae ( LSNe ) . The interaction of SN ejecta and wind surrounding it is considered to be a power source to illuminate LSNe because many LSNe show the wind signature in their spectra ( Type IIn LSNe ) . However , there also exist LSNe without the spectral features caused by the wind ( Type IIL LSNe ) . We show that , even if LSNe are illuminated by the interaction , it is possible that they do not show the narrow spectra from the wind if we take into account of non-steady mass loss of their progenitors . When the shock breakout takes place in the dense wind with the density structure \rho \propto r ^ { - w } , the ratio of the diffusion timescale in the optically thick region of the wind ( t _ { d } ) and the shock propagation timescale of the entire wind after the shock breakout ( t _ { s } ) strongly depends on w . For the case w \hskip { 3.0 pt } \raisebox { 1.72 pt } { $ < $ } \hskip { -7.5 pt } \raisebox { -3.01 pt } { $ \sim$ } % \hskip { 3.0 pt } 1 , both timescales are comparable ( t _ { d } / t _ { s } \simeq 1 ) and t _ { d } / t _ { s } gets smaller as w gets larger . For the case t _ { d } / t _ { s } \simeq 1 , the shock goes through the entire wind just after the light curve ( LC ) peak and narrow spectral lines from the wind can not be observed after the LC peak ( Type IIL LSNe ) . If t _ { d } / t _ { s } is much smaller , the shock wave continues to propagate in the wind after the LC peak and unshocked wind remains ( Type IIn LSNe ) . This difference can be obtained only through a careful treatment of the shock breakout condition in a dense wind . The lack of narrow Lorentzian line profiles in Type IIL LSNe before the LC peak can also be explained by the difference in the density slope . Furthermore , we apply our model to Type IIn LSN 2006gy and Type IIL LSN 2008es and find that our model is consistent with the observations .