The Secret That Makes the Creamiest Bokeh in Photography and Why It Remains a Niche

Sony sells a 100mm portrait lens with an f/2.8 aperture that meters like an f/5.6. That is not a manufacturing defect or a marketing rounding error. Sony built a disc of gray-graded glass into the middle of the lens, on purpose, and gave away roughly two stops of light to buy what might be the smoothest out-of-focus blur any lens can produce. The trick works, it has worked since the 1990s, and the lenses built around it have never been more than a niche.

What an Apodization Element Actually Does

An apodization element is a radially graded neutral density filter placed at or very near the aperture. It is clear in the center and grows steadily darker toward the edges, so light passing through the middle of the lens comes out at full strength while light near the rim gets dimmed. The sharp, in-focus part of your image is left essentially alone. The principal visible effect lives in the blur.

To see why, picture a single bright point in an out-of-focus background: a streetlight, a speck of sun on chrome, a fairy light on a tree. A lens does not render that point as a point when it falls out of focus. It renders it as a disc, the familiar bokeh ball. On a conventional fast lens, that disc usually has a defined edge, and often a brighter rim, a signature of how the lens's spherical aberration is corrected. Other artifacts pile on separately: molded aspherical elements can leave faint concentric onion rings inside the disc, a flaw that turns up in expensive glass as readily as cheap, and stopping down turns the circle into a polygon that echoes the aperture blades. Stack hundreds of those hard-edged, textured discs across a busy background and the blur turns nervous and cluttered, fighting your subject for attention.

The apodization element attacks exactly that rim. Because the filter is darkest right where the disc's edge forms, each bokeh ball fades smoothly from a bright center to nothing at its boundary, with no hard outline and no bright ring. The discs melt into one another instead of stacking like coins. The name comes from optics and signal processing, where to apodize a signal means to shave the "feet," the small sidelobes at its edges, off the peak. Do that to the pattern a lens draws from a point of light and the background stops competing with your subject and starts dissolving behind it.

F-Stop, T-Stop, and Why the Lens Meters Two Stops Dark

The blur is free. The light is not. To understand the bill, you have to separate two numbers that most photographers treat as one.

An f-stop is pure geometry. Divide the focal length by the width of the entrance pupil and you have it, a number that describes how wide the opening is and therefore how shallow your depth of field will be. What it cannot tell you is how much light survives the trip, because it assumes the glass passes every photon that enters. Real lenses never do. A T-stop measures the light that actually reaches the sensor after the elements and coatings have taken their cut. Cinematographers mark their lenses in T-stops because an exposure has to match from shot to shot, and a typical fast still lens rated f/2.8 usually transmits closer to T3.2.

An apodization lens blows that small gap wide open. The Sony FE 100mm f/2.8 STF GM OSS has a true f/2.8 entrance pupil, and the big wide-open blur circles that go with it, but its graded element blocks so much light near the margins that it meters at T5.6. You get the background separation of f/2.8 and the exposure of f/5.6 at the same time. That is about two stops gone, traded for the blur. On Fujifilm's APD lens the math runs the same direction but smaller: a bright f/1.2 that transmits like T1.7 wide open, with the penalty shrinking as you stop down because the darkened outer zone only matters at the widest apertures. By roughly f/5.6, the diaphragm is admitting light only through the filter's nearly clear central region, so the apodization effect and its transmission penalty effectively disappear and the lens behaves like any other.

The depth of field mostly stays put, and that surprises people. Depth of field follows the physical width of the pupil, and the apodization element does not shrink the opening, it only tints it. The entrance pupil and the largest blur circle the lens can draw are still those of an f/2.8 even as the meter reads two stops darker, though the graded filter dims the outermost rays enough that the perceived falloff from sharp to soft can differ slightly from a conventional f/2.8's. That darkness is a real cost. It dims the viewfinder on a mirror-and-prism camera, it pushes your ISO up in exactly the dim, moody light where you most want a fast portrait lens, and it narrows the window where handheld shutter speeds stay safe. You are paying, in light, for a look.

From the Minolta 135mm to the Sony 100mm STF

The idea reached production in 1999, when Minolta shipped the AF 135mm f/2.8 [T4.5] STF. STF stood for Smooth Trans Focus, Minolta's name for the apodization design, and the lens remains a curiosity for a second reason: it is one of the very few autofocus-mount Minolta or Sony A-mount lenses ever built without autofocus. The graded element confused the phase-detection sensors of the day, so Minolta simply left focusing to your hand. The lens even carried two diaphragms: a conventional nine-blade automatic aperture, plus a separate ten-blade manual aperture that stayed nearly circular to keep the blur round, its ring marked in T-stops to reflect the light the apodization element swallows. It built a cult reputation for its rendering that still trails it on the used market. Sony inherited the design when it absorbed Minolta's camera business and reissued it as the A-mount 135mm f/2.8 [T4.5] STF in 2006.

In 2014 Fujifilm brought the technique to a mass-market autofocus body. Its XF 56mm f/1.2 R APD was a portrait lens for the APS-C X-mount that wore its apodization design right there in the APD suffix. The filter still defeated the on-sensor phase-detection pixels, so the APD version fell back to slower contrast-detection autofocus even on bodies that offered phase detection. Fujifilm has since discontinued it in favor of the newer XF 56mm f/1.2 R WR, which chases the same smooth rendering without the light-eating element.

The most polished version arrived in 2017. The Sony FE 100mm f/2.8 STF GM OSS was the first full-frame apodization lens that could autofocus, driven by a Direct Drive SSM motor, with optical stabilization and close focusing to 0.57 m for quarter-life-size macro. It shipped as part of Sony's flagship G Master line at $1,499, and its f/2.8 pupil still meters at T5.6. For under half that price, Venus Optics offers the manual-focus Laowa 105mm f/2 STF for full frame, an f/2 lens that transmits like T3.2 and, like the Minolta before it, pairs a second fourteen-blade diaphragm with the apodization glass to shape the blur. Canon reached the same feathered highlights on full frame with the Canon RF 85mm f/1.2 L USM DS in 2019, using a Defocus Smoothing coating vapor-deposited onto the elements to dim their edges rather than a discrete gray disc; the lens is rated around T2.2 wide open, roughly a stop and three quarters slower than its f/1.2 geometry suggests.

Nikon's Defocus Control Solved a Different Problem

Nikon chased the same goal by a completely different road, and the two are easy to confuse. The AF DC-Nikkor 135mm f/2 D and its 105mm f/2 sibling carry a Defocus Control ring, which is not an apodization filter. Turning it shifts an internal group to deliberately over-correct or under-correct spherical aberration, changing how the out-of-focus zones render. Nikon's own account of the design explains that you tune the blur in front of the subject or behind it, one improving at the other's expense, anywhere from f/2 to f/5.6.

The trade is different from apodization in a way that matters. Defocus Control costs you no light, since there is no dark filter in the path, and the lens autofocuses normally. But it does not perfectly feather every highlight the way a graded element does, and dial in too much correction and it softens the sharp subject you were trying to protect. It is a bokeh-shaping tool. The apodization element is a bokeh-perfecting one, and it charges accordingly.

Why Apodization Lenses Remain a Niche

Line the reasons up and the neglect makes sense. These lenses are often expensive and single-minded, built around one focal length and one aesthetic. Several are manual-focus only, and even the models that autofocus can hunt in low light because the apodization treatment has dimmed the very scene the sensor is trying to read. They cost you anywhere from one to two stops of the shutter speed and ISO headroom you bought a fast lens to get, depending on the design. And they answer a question most people can solve more cheaply.

That is the honest case against. For a working portrait photographer, a plain 85mm f/1.4 or a 135mm f/1.8 throws a background so far out of focus that the difference in the rim of each highlight disc rarely survives being resized for a screen. Those lenses cost about the same, gather several more stops of light, and focus fast in the dark. A phone will fake a shallow-depth look for nothing, computing the blur it cannot actually capture. Against all of that, an apodization lens asks you to pay more, in money and in light, for a refinement that only reveals itself in the right background, at the right aperture, viewed large enough to notice.

When those conditions line up, though, nothing else quite matches the look. Point one of these at a subject standing in front of dappled light through leaves, or a city street of distant signs at night, and the background does not just blur, it liquefies, every highlight fading edge to edge with no seam between them. If your work lives on that specific quality of separation, the transmission tax is the cheapest part of the deal. Better light will do more for most portraits than any exotic prime will, which is why a course like Perfecting the Headshot pays off before a specialty lens does. But the apodization element remains the one piece of glass built for nothing except making the blur behind your subject disappear as gently as physics allows, and it will keep doing it long after the phones have moved on to the next computed trick.