Glow Is Evidence of Something Brighter Than the Picture

How it works

Three deterministic steps, no model.

Contrast is raised hard on a copy of the image. This is the isolation step: it pushes mid tones and shadows down towards black while the brightest areas stay bright, so what survives is roughly the highlights.

That copy is blurred with a wide radius, which is the spread. The width of the blur is what makes the effect read as glow rather than as a halo.

The blurred highlights are screened back over the untouched original. Screening brightens and never darkens, so the effect can only add light, and areas where the isolated copy went to black contribute nothing at all.

The original image is what the blend starts from, so everything outside the highlights keeps its own detail.

Isolation is the whole design

If you take one idea from this page, it is that bloom is defined by what it excludes.

Real optical bloom is a threshold phenomenon. Light scatters in every lens, all the time, but it is only visible where a source is dramatically brighter than its surroundings — many times brighter, not a little. Everywhere else, the scatter is lost under the image itself.

An effect that ignores that threshold and spreads everything reproduces the physics of a dirty lens, not of a bright light. A dirty lens does exactly this: lifts the shadows, lowers contrast, and makes the picture look like it was shot through a window that needs cleaning.

Which gives you a diagnostic that does not require any judgment about taste. Look at the darkest part of the image before and after. If black is still black, the effect is behaving like light. If black has become dark gray, it is behaving like haze, and the cause is always that the isolation did not isolate.

Headroom, and why some photographs cannot glow

There is a class of image this simply cannot help, and it is worth recognizing before running it three times with different expectations.

A photograph carries a fixed range of values. When it contains a genuinely bright source — a lamp against a dusk sky, sun through leaves, a specular flash off water — that source sits at or near the top of the range while most of the frame sits far below. That gap is what the effect uses.

A flat, evenly exposed picture has no gap. Nothing in it is dramatically brighter than anything else, so isolating the highlights returns most of the picture, and the result is the fog described above.

The upstream version of this is exposure. A photograph deliberately exposed for the shadows, letting a light source clip, has enormous headroom for this kind of treatment. One exposed to keep everything in range has almost none. That decision was made at capture and it cannot be made again later.

When the effect is doing work and when it is decoration

Bloom is one of the few effects that can make an image more accurate rather than less.

Your eye does this. Look at a bare bulb and it spills over its own edges, because your eye scatters light too. A photograph of a bright source with hard, clean edges is in one sense more literal and in another less true to the experience of having been there. Adding the spill puts back something the sensor removed.

Beyond a certain strength, it stops doing that and becomes a style — a soft, dreamy, slightly nostalgic register that is perfectly legitimate and no longer about accuracy. Both are fine. Knowing which one you are doing is what stops the setting drifting upward every time you use it.

Publication gate

This page ships once the workflow has been run against a bright point source at dusk, a flat overcast scene with no highlights, a high-key image that is mostly bright, and a backlit portrait, with the black point measured before and after in every case to confirm the shadows have not lifted.