TLDR:
Friends don’t let friends view scene-referred imagery without a S-Shaped transform.
Long Story:
Basically, looking at scene-referred imagery without a display rendering transform that does some sort of tone compression is not really desirable.
As an example, let’s assume you shot outside and the scene peak luminance as measured was 20000nits and shadows about 0.0005nits. Now, you are back inside and want to display that on your screen that only goes from 0.01nits to 120nits, as you can see, there is a problem, the dynamic ranges are dramatically different.
You could naively compress everything to fit, but you would lose your imagery almost entirely as it gets pushed in the shadows. Alternatively, you could cherry pick a suitable window in the dynamic range but would clip blacks and highlights. The standard solution to that issue is a non-linear compression function that allows to fit as much possible of the dynamic range of the scene into your display dynamic range while maintaining a faithful appearance.
I have some images I can dig around if it does not make sense (but it should).
Cheers,
Thomas