For mid-wave IR and long-wave IR edge and wide bandpass filters commonly produced at DSI for military uses, a HPP (or CWL) tolerance of ±0.5% (of the wavelength value) or higher is relatively routine, while tighter specifications tend to drive up cost. Similarly, slopes of 2% or greater are standard; smaller (steeper) values are more difficult. Thus, if there is a large separation between the wavelengths to be passed and rejected in a system, it's important not to put arbitrarily tight constraints on these values.
The out of band blocking specifications can also drive cost. Here it's necessary to state whether the blocking specification is an average value over the entire blocking band, or a minimum threshold value. In many military systems, what is really required is blocking just at certain key wavelengths or wavelength bands. In this case, these are what should be called out, and it should be made clear that blocking at other wavelengths isn't as critical. This can simplify the coating design and lower the fabrication cost.
Another important — and often misunderstood — specification for coated optics is flatness. First, it's important to realize that virtually all optics vendors nominally specify component flatness prior to coating. The mechanical stress inherent in optical coatings (which, again, is typically higher in denser, harder coatings) can easily degrade the surface figure of a nominal λ/10 flatness component by an order of magnitude.
There are a couple of additional important considerations when defining flatness specifications. First, if an optic is being used solely in transmission, the surface flatness shouldn't be specified at all. Rather, the relevant specification is wavefront distortion on transmission, a parameter which is much less sensitive to degradation caused by coating stress.
The other factor to be aware of is that flatness tends to be much more of an issue with certain specific types of coatings. In particular, it's most pronounced when there are coatings with substantially different thicknesses (and there fore stress characteristics) on two sides of a thin optic. So, a window which has the same antireflection coating on both surfaces usually does not experience a problem. But dichroic beamsplitters, which have a thick long (or short)-wave pass filter coating on one surface and a much thinner antireflection coating on the other surface do often exhibit significant flatness degradation due to the differential stress effects between the two sides of the optic. Here it's not uncommon to add layers to the antireflection coating which change its mechanical stress characteristics without altering its spectral performance. However, producing such components in practice typically requires an iterative series of test runs in order to optimize the performance, and this adds cost.
Military optics have a lot in common with commercial optics. However, although their performance and environmental specifications are similar, the requirements for military applications are generally more stringent and the need to rigorously hit the target with those parameters is more critical. An understanding of how optical coatings are made and how they are best specified can reduce the impact these more stringent requirements have on the cost of the optics.