You can achieve this easily, any ranges are fine, it's usual inverse dependency. But you probably have issues because adaptation is not purely applied as divider to each pixel of screen in enbeffect.fx, but there are some other parameters. For example, EAdaptationMinV1, EAdaptationMaxV1 also cropping like in enbseries.ini, but they work in non linear way and not actually min and max, but called such way for simpler understanding. Here is the code from enbeffect.fx:
Code:
grayadaptation=max(grayadaptation, 0.0);
grayadaptation=min(grayadaptation, 50.0);
color.xyz=color.xyz/(grayadaptation*EAdaptationMaxV1+EAdaptationMinV1);//*tempF1.x
grayadaptation is adaptation texture, internally computed with parameters from enbseries.ini applied. max and min in this code is clamping exactly the same way as in enbseries.ini, just mathematically min(grayadaptation, 50.0) is actually maximal level clamping AdaptationMax. If you want to have very high ranges of adaptation, then increas 50 to much bigger (i suggest to not use more than 32768) or you can get black areas.
If you replace
color.xyz=color.xyz/(grayadaptation*EAdaptationMaxV1+EAdaptationMinV1);
by this:
color.xyz=color.xyz/grayadaptation;
the result will be pure inverse dependency, producing normalization and no matter how bright screen is, the result is same in average.
Another thing, don't forget later tonemapping applied, it compress colors and brightness a lot, if you want to test adaptation only, then disable tonemapping and write output of color.xyz after adaptation code to temporary variable and in the end of shader use it as output directly. You know, it's not that easy, you should learn to tweak shader too or select some which you like and adjust it's parameters, because custom shaders by users can be very much different.