That is correct, and why you sometimes also see the term output-referred used.
The only valid replacement term that I can think of is “relative”, and as it happens it’s also already in use. While scene/output-referred are the most common, I have seen scene/output-relative as well on several occasions.
The term is about the output medium, which could just as well be paper or something else. And indeed, both Filmic and Sigmoid let you set the black and white points of the output, which affect the dynamic range.
Yeah, I think that goes with the common “scene-referred workflow” and “display-referred workflow” monikers.
“Referred” is about the data. “Scene-referred data” is image data whose magnitudes correspond to the light energy levels at the scene, where math like ‘red x 2’ lines up with a corresponding 2x increase in the original light energy. The rubicon where this is disturbed is the first ‘non-linear’ tone curve applied, which changes that energy reference.
So, when folk use the term “scene-referred workflow”, they’re referring (oh, overload, don’t ask…) to a sequence of image operations performed on scene-referred image data.
I thought it was ok to say energy in this context.
Anyway… I simply think using “referred” is not plain English. Relative (or related) sounds better to me.
I think lowering the filmic/sigmoid white point to get a print of limited range (and of course it’s low to begin with) is a bit of a blunt tool? if I wanted that I’d probably do it with modules.
While true, since total light energy is proportional to photon count for any given colour of light, I’m not sure the difference has any practical significance in this context. The more as white balancing already adds a colour-dependant multiplier…
… light intensity is proportional to photon count for any given colour of light …,
Just a remark from a physicist who keeps noticing that the physical terms are mixed up among photographers. But I don’t want to highjack the thread, so no further comment on this topic from me.