The effect, described as arithmetic
A double exposure is a photograph in which one frame’s structure appears inside another frame’s bright areas. On film it happened because the shutter opened twice on one piece of emulsion and light accumulated; the dark parts of the scene contributed nothing, the bright parts contributed everything, and the result was an addition rather than a choice. Screen is the arithmetic of that: one minus the product of the two complements, which is the only combination of two frames that adds light without ever exceeding white.
Everything else in the recipe exists to make a digital pair behave like that piece of film. Flattening the layer to luminance stops its colour arriving where the base’s colour already is. Lifting the black point stops the layer’s shadows being exactly transparent, which on film they never quite were. The highlight key restricts the whole effect to where the base is already bright, which is the step that keeps a face readable underneath a forest.
Four moves, in order
- Put the silhouette underneathThe base is the frame whose shape survives, so the portrait goes in slot one. A subject against a bright, plain background is what this effect was invented for, and the brighter the background relative to the subject, the less work every later control has to do.
- Put the texture on top, and fit itBranches, a skyline, a crowd, rain on glass. Cover is almost always the right fit: the texture has no subject of its own, so losing its edges costs nothing. Scale it up if its detail is too fine to read at the size of the face.
- Set the amount, then the black pointTake the divider up until the texture is unmistakable, which is usually somewhere past 70 per cent, then bring the black point up in small steps until the texture's dark areas have presence without the whole frame going pale.
- Key it to the highlightsRaise the key from zero and watch the face come back. The key is reading the base's own brightness, so it removes the texture from the subject and leaves it in the background — or, with the base inverted in your choice of frames, exactly the other way round.
When it will not work
The key is a brightness threshold and nothing more. It has no idea what a face is, so a subject wearing a white shirt will have the texture land on the shirt however carefully the key is set. The tool cannot separate a subject from its background — that is a different operation with a different owner — and a recipe that pretended otherwise would be lying about what the slider does.
The other common failure is a texture with no dark areas at all. A bright, evenly lit layer screened onto anything produces a pale frame, because screen can only add. If the result looks washed out no matter where the controls go, the layer is the problem: look for one with real blacks in it.
What people get stuck on
What makes a photograph a good silhouette for this?
One subject, a clean outline, and a background that is much brighter than the subject. A head and shoulders against a window or a pale sky is the classic case because the boundary between subject and ground is a single steep step in brightness, which is what the highlight key has to find. A subject photographed against a busy room has no such step, and no amount of adjustment here will invent one — the texture will land on the face as readily as on the background.
Why flatten the layer to luminance before mixing it?
Because screen adds light, and colour added to colour goes somewhere neither photograph was. A forest layer that keeps its greens will push every highlight of a portrait toward green, which reads as a colour cast rather than as a second exposure. Flattening the layer to a grey version of itself leaves the base in charge of hue and the layer in charge of structure, which is the division of labour the effect depends on. Turn it off when the layer's colour is the point.
The black-point lift looks like it is washing everything out.
It is, and past about 25 per cent that is all it does. The control raises the darkest value in the layer so that the layer's shadows stop being perfectly transparent under screen, which is what stops a dense texture from disappearing entirely into the base's midtones. Small numbers buy presence. Large ones flatten the layer to a grey sheet and lift the whole frame with it, so if the picture is going pale, this is the first control to check.
Can I use the recipe with a mode other than screen?
Yes, and lighten is worth trying first. Screen adds the two frames together in a way that is guaranteed never to clip, which is why it is the default and why it lifts the shadows slightly everywhere. Lighten simply keeps whichever frame is brighter at each pixel, which holds the base's dark areas absolutely still and makes the texture look painted on rather than exposed through. Overlay gives a harder, more graphic result and needs the black-point lift closer to zero.