When you say “normalisation” here, are you referring to “making the image look normal”, i.e. “not flat and log”? Your normalisation process here is doing the job that a print emulation LUT or camera LUT would do if you used one, correct? I ask because the term “normalisation” is more commonly used for the “correction” step you refer to before your normalisation, to even out differences of exposure or white balance, to get every shot to the same “normal” start point.
In ACES, what you call “normalisation” is done for you by the combination of the RRT and ODT.
With your previous approach you then apply your look after this, working on a “display referred” image, i.e. one which is already a “normal video image”. You cannot do this with ACES, as the RRT + ODT is the last step, applied by the grading system, and you cannot add anything after it.
In ACES all your grading happens in what would be your first set of nodes in your previous approach. So you are not working directly on the pixel values, but rather seeing the effect of your changes through the RRT and ODT. This can be slightly counterintuitive at first for somebody who is used to a more “video” approach. It is more akin to a traditional DI grade, where a LUT emulated the look of a film-out, and you had to grade before this, because clearly you couldn’t add changes after printing to film.
It takes a bit of getting used to, but once you become familiar with how the various grading controls respond, it will become just as intuitive to you as your previous approach.