Linear space → inversion → display

A camera sensor is a photon counter. Each photosite accumulates electrons in direct proportion to the number of photons that hit it during the exposure. The raw file records those electron counts as integers — no tone curve, no gamma, no colour matrix applied yet. The raw file is always linear regardless of what camera or scanner produced it. What varies is what software does to that linear data at the moment of import or development.

When you import a raw file into Lightroom with a profile such as Adobe Color, Camera Standard, or any manufacturer profile, Lightroom applies a DCP (Digital Camera Profile) that contains a tone curve and often a look table. The tone curve is an S-shaped function that lifts shadows, adds contrast in the midtones, and rolls off highlights. For direct photography this looks pleasing because it mimics the response of negative film. For a scan of a negative, it is the wrong operation — it applies a film-like response on top of data that already encodes film response, and does so before the inversion that is needed to convert the negative into a positive.

Inversion is the operation v → 1 − v. It is only mathematically meaningful when v is a linear representation of the scene — specifically, a value proportional to the transmittance T at each point of the negative. If a tone curve has already been applied, v is no longer proportional to T. It is a bent, compressed version of T. Inverting it with 1 − v produces a result that is the inversion of a bent curve, not the inversion of the underlying scene luminance. The toe of the original negative (dense shadows, low T) was lifted by the tone curve before inversion; after inversion it becomes a compressed highlight. The shoulder (thin highlights, high T) was compressed by the tone curve; after inversion it becomes a blocked shadow. These are not recoverable errors — the tonal relationships have been permanently scrambled. No adjustment in Lightroom after inversion can undo a non-linear operation that was applied before inversion.

Import the raw scan with a linear profile. In Lightroom this means selecting the built-in Linear profile in the Basic panel, or using a plugin such as Negative Lab Pro which handles this automatically. With a linear import, the data passed to the inversion operation is proportional to transmittance, the inversion is arithmetically correct, and the resulting positive has intact tonal relationships from toe to shoulder. Gamma encoding — the v^(1/2.2) step that prepares the image for display — is then applied as the final step, either by Lightroom's output transform or by the operating system's colour management pipeline. This is the only order in which all three operations (normalise, invert, gamma-encode) produce a mathematically correct result.

Switch between the two import modes using the buttons above. In the pipeline, stage 3 highlights red (wrong) or green (correct). In the result cards below, the dashed grey line in the left card shows the correct curve overlaid on the damaged one — the gap between them is the tonal information that was irreversibly destroyed by applying the tone curve before inversion. Notice that the damage is worst at the extremes: deep shadows in the scene (high density in the negative, low electrons, left side of the x-axis) and bright highlights (low density, high electrons, right side) both suffer more than the midtones. This is the characteristic signature of a premature gamma or tone-curve operation on negative film data.