This is a key point very of us are aware of ! In digital Photography, tonal distribution is not linear along the histogram axis. Here is what I mean : Let’s say we have 5 main zones along the X axis which represents the tonal distribution : from left to right we will have the Dark Shadows – The Shadows – The Midtones – The Light Greys then The Highlights. We will suppose you capture in RAW (12bits) then, you’ll be able to capture 4096 differentiated tones per channel (RVB). The secret is : the Highlight zone will hoard half of the tonal palette just for itself so it’ll keep 2048 differentiated tones in the last fifth of the histogram ! Then the light greys will take half of the remaining thus : 1024 tones per channel, the midtones half of the remaining : 512 , then Shadows 256 and so on… In the facts, you have understood you will always get better details if you push your histogram to the right taking care, sure, to prevent clipping.
Let us suppose your histogram base takes about 2/5 of the x axis of the histogram… When pushed on the left stop of the axis, you will obtain 128 + 256 = 384 differentiated tones per channel… If you push the distribution to the right of the axis ( posing for highlights ) you will be able to differentiate 2048 + 1024 = 3072 tones per channel… up to you, which image will have the best details ?