Reading the correction curve

A correction curve is not a mathematical decoration. It's a contract with the image: "for every tonal value you bring me, I'll give you back this other value on output". If you know how to read this curve, you know in advance what your correction will do to your image before you even look at the render.

The graph serves two distinct purposes. While you adjust — it's your visual feedback, you see how each slider movement deforms the transformation. After you've finished — it's your documentation, the mathematical signature of your calibration that you can compare to others or archive.

#What you see

A graph with two axes. The horizontal axis carries the input values — on the left, the darkest values of your image before correction; on the right, the lightest. The vertical axis carries the output values — at the bottom, how dark these values will become; at the top, how light they will become.

If the correction is null (nothing to change), the curve is a perfect diagonal running from the bottom-left corner to the top-right corner. Each pixel comes out exactly as it went in.

Any deformation from this diagonal means the correction changes something. Three typical shapes:

S-curve: dark in the shadows (drops below the diagonal in the left part), light in the highlights (rises above it in the right part). This is an increase in contrast — the shadows sink, the highlights brighten.

Curve bowed upward: the whole curve is above the diagonal. This is a general lightening. Often used to recover an underexposed image.

Curve bowed downward: everything below the diagonal. General darkening.

Curve with a knee: a sharp break somewhere. It's the sign of a threshold in your correction — typically, your black point or white point forcing an abrupt compression.

On the graph, twenty-five sampling points are placed at regular positions along the curve. They are not used to change the correction (changes go through the three sliders), but to visually read where the curve passes at different places. It's more accurate than estimating by eye on a continuous line.

#Why it matters

To anticipate the render before looking at the image. If you read on the curve that your black point starts very low and your white point ends very high, you know your image will come out contrasted even before the preview updates. When you calibrate in series, this quick glance replaces the visual inspection of each render.

To compare two calibrations. Display two superimposed curves — one from your May chemistry, the other from June. You immediately see where they diverge. If the June curve is pulled lower in the shadows, your chemistry has become more contrasted — that's measurable, not an impression.

To pass on a calibration. If you send your

.acv

#When you don't need it

For quick adjustments by instinct. If you correct three images in two minutes with no intent of reproducibility, the direct visual preview is enough. The curve is a tool of precision and memory, not a mandatory step.

For trivial corrections. A simple brightness increase doesn't need the graph — you push the gamma, the image lightens, that's it. The graph shines on more complex adjustments.

If you work only by applying existing presets. You load your saved calibration, you see the result on your image, you accept or you decline. No need to read the curve each time.

#Worth remembering

#The test

Load an image, open the analysis. The curve should be a perfect diagonal at startup (no correction). Now push only the black point to the right — the curve should flatten in the left part, taking the shape of a curve that starts at zero with a plateau then rises to rejoin the diagonal. You physically see how your adjustment transforms the tonal values. If the curve doesn't move, either the preview is frozen, or you're moving the wrong slider.