The loss of between 20-30% ANSI contrast, reported from users of the Panamorph and Prismasonic seems unavoidable due to optical scattering.
However, there is some confusion about the total lumen loss with anamorphic lenses, aided by a few intemperate posts based on improper measurement protocols.
When you diverge a light beam with a brightness of a particular lumen value, using an HE anamorphic lens (the simpler case to explain vs the case with a VC lens) over a larger area of coverage, of course the foot lamberts measured at the screen must go down compared to that measured with the smaller area of coverage.
But comparing apples to apples, by measuring the foot lamberts with the *lens in place* in both cases, and comparing full panel usage, using a scaler to use the complete vertical extent for a 2.40:1 film, to partial panel usage, where vertical pixels are trimmed to make the film fit the native 1.78:1 panel width, there is a gain of around 20% in brightness, simply from more of the panel radiating light versus radiating its black level.
From quality optics the transmission loss should be around an imperceptible 1-2% or so for anamorphic geometry warping, made up for by the additional pixels used by scaling to use all the vertical resolution you paid for.
The other confusion arises when people compare the brightness of a wide image, say 2.40:1 with the *lens in place*, which uses the full panel, to a native 1.78:1 image, with the *lens removed* and again using the full panel.
Of course, the same lumens over a smaller total area will produce a higher foot lambert reading in that case, as expected. This brightness gain is the primary reason some people prefer to move the lens away for 1.78, not the loss of some horizontal resolution, IMHO. Brightness rules the list of the eye's preferences.
Given that broadcast, satellite and cable HDTV feeds are generally already compressed down from 1980 to a range around 1440 horizontal resolution for distribution, the trimming of the panel width in a CIH, fixed lens setup, to a similar horizontal resolution when displaying 1.78:1 HDTV, is not losing significant information, when using a good scaling engine.
For those using a CIH, anamorphic-lens-in-place-at-all-times setup, the foot lamberts stay near constant for the narrower aspect ratios, as the image is trimmed by shutting off pixels from the width, but since the light engine/optical path combined lumens remain in the same proportion to the area radiated, the foot lamberts remain constant at the screen.
The issue of whether the 20-30% loss in ANSI contrast is a problem or not is all about tradeoff's.
For Cinemascope movies, the larger, immersive image seems to be a good trade for that small loss in ANSI contrast. For native 1.78 images from a native 1.78 panel, the tradeoff of CIH convenience, simplicity of masking or not having to move the anamorphic lens, for that loss of ANSI contrast, especially when viewing HDTV sources as opposed to film sources, becomes more problematic.
However, all the years when CRTs ruled the high quality projection scene, with their native ANSI contrast significantly lower than the ANSI contrast of the JVC RS1, even after accounting for the loss from an anamorphic lens, should tell us that the perceptible impact when watching film sources will be minimal.
I hope that helps!