In Jerkstrand+2012, model luminosities for optically thick MIR lines ([Ni II] 6.63 mu, [Co II] 10.52 mu, [Fe II] 26 mu) were used to constrain the volume filling factor for iron/nickel gas (Fig 11, 12).
The luminosity for optically thick lines in LTE is given by Eq 11.
L(t) = 8 pi k T(t) lambda^(-3) V(t) t^(-1)
The linear T dependency (compared to a usual exponential) means one can estimate V(t) to good accuracy even with only roughly known temperature.
V(t) can then be compared to the total expansion volume V_exp(t) = 4pi/3 (Vt)^3 where V is the maximum expansion velocity (inferred from the line widths), to give the filling factor. This in turn can be compared to 3D explosion models.
There is a complication - models typically show significant contribution by primordial iron and nickel in the hydrogen zones. Neglect of this causes an overestimate of f. In the modelling work of J12 this contribution was taken into account (Fig 11,12).
The filling factor for SN 2004et was found to be ~0.15. The mass of synthesized iron-group elements was ~0.06 Msun, which is only ~1% of the mass that resides inside the expansion volume (lots of oxygen, helium, hydrogen also reside in the inner ~1800 km/s). The reason the filling factor gets so large for this component is believed to be that it stays hotter due to its initially trapped radioactive decay and therefore expands (while compressing other components). The degree to which this happens depends on the fragmentation and overall mixing degree of the supernova zones.