Smart Analyzer

A default image auto-correction always does the same thing: it takes the darkest value, pushes it to black, takes the lightest, pushes it to white, and stretches everything in between. That works great on an average, well-exposed image. It massacres everything else.

A Low Key portrait — a barely lit face, a black background, a hushed mood — run through a naive auto-correction becomes a broad-daylight portrait with a gray background. The photo's intent is gone. A landscape in diffuse light, light mist over the sea, becomes an over-saturated postcard. The Smart Analyzer exists to prevent that kind of mistake. It looks at the image before correcting it, understands what it's trying to say, and adapts the strategy accordingly.

#What happens on screen

You see nothing. That's the point. When you load an image into Calibration Flow, the Smart Analyzer runs in the background during the few hundred milliseconds the import takes. It reads your image's luminance distribution — where the darkest pixels are, the lightest ones, how the midtones spread. From this signature, it sorts your photo into one of the following seven profiles:

• Normal: balanced, standard distribution.
• Low Key: dominated by dark tones, intentional deep shadows.
• High Key: dominated by light tones, dominant highlights.
• Charcoal: very contrasty with deep blacks, charcoal aesthetic.
• Soft Mist: low overall contrast, foggy mood.
• Underexposed: the whole image bunched toward black, accidental.
• Uncertain: no profile stands out.

This classification then drives the entire correction pipeline that follows. Instead of a single target (the blackest black possible, the whitest white possible), the Smart Analyzer proposes adapted targets: on a Low Key profile, it doesn't push the black point toward absolute black and preserves shadow depth; on a High Key profile, it limits highlight compression; on Charcoal, it keeps the strong contrast that makes the image.

You can always go back to manual with the three sliders (black point, white point, gamma) if you want to push in a direction. The Smart Analyzer provides a smart starting point, not a final decision.

#Why it matters

For anyone calibrating alternative processes across very diverse images. A cyanotype corrective curve calibrated on a neutral target then applies to a Low Key portrait, a High Key landscape, a normal still life. If the auto-level standardizes all those images toward the same maximum contrast before applying the curve, you lose each photo's intent. The Smart Analyzer avoids that flattening — each image keeps its mood, and the curve applies on top.

For anyone who wants a fast workflow. You can import ten images in a row, see ten consistent automatic results, keep eight, rework two by hand. Without the Smart Analyzer, you adjust all ten manually. With it, you only step in on the special cases.

For anyone starting out in post-processing. Manual adjustments demand a grammar (knowing how to read a histogram, understanding thresholds). The Smart Analyzer replaces that grammar with automatic recognition. You get a correct result without having to learn the mechanics first.

#When you don't need it

You're working on a calibration target. The Smart Analyzer is built for photographic images. On a target with 25 discrete gray patches, it will probably classify Charcoal or Underexposed depending on the process, and its auto-level strategy will make no sense. For target measurement, what matters is the patch-by-patch measurement — turn off the smart auto-level and let the app measure raw.

You want to preserve a non-standard, unique look. The Smart Analyzer normalizes toward its interpretation of the intent. If you want exactly the distribution your source image has, with no interpretive adjustment, turn off smart analysis and work on the raw histogram with the three manual sliders.

You're working on an atypical image. An abstract photogram, a directly scanned negative, a color target — the Smart Analyzer may land on the Uncertain profile, and its auto-level reverts to generic. In that case, you might as well set things by hand from the start.

#Key facts

#The test

Load two images one after the other into Calibration Flow: a softly lit portrait with a dark background (typically Low Key) then a beach landscape in full sun (typically High Key). Let the auto-correction apply to each. The render should be visibly different — on the portrait, the blacks stay deep and the hushed mood is preserved; on the landscape, the whites aren't crushed toward mid-gray. If both corrections produce an identical, standardized render, either the Smart Analyzer is turned off or your source image has a tonal profile it couldn't classify — switch to manual adjustments then.