One ratio explains almost all of it
Take the width of the lit surface and divide it by the distance from that surface to the subject. That number is the whole story. Above about 1 the source is very soft, shadows have no discernible edge and the light wraps around the far side of a face. Around 0.5 it is recognisably soft with a defined but gentle shadow. Around 0.2 the shadows have edges. Below about 0.1 you are looking at a point source, whatever the equipment cost.
A 36 inch box at four feet gives a ratio of 0.75, which is properly soft. The same box at twelve feet gives 0.25, which is not. This is why moving a light is a bigger decision than choosing one: the distance appears in the denominator and doubling it halves the softness, while doubling the physical width of the modifier is expensive, awkward and only does the same amount of good.
Expressed as an angle rather than a ratio the same thing is easier to picture. The source at four feet subtends about 41 degrees as seen from the subject, which is a large part of the sky in front of them. At twelve feet it subtends about 14 degrees. Direct sun is about half a degree, which is why sunlight is the hardest light most people ever work with despite being enormous.
The shadow edge is where softness becomes measurable
A point source casts a shadow with a sharp edge. A source with width casts an edge with a band across it, because from a point just inside the shadow you can see part of the source and not the rest. That band is the penumbra, and its width on a surface behind the subject is the source width multiplied by the subject-to-background distance divided by the source-to-subject distance.
Three things follow, and all three are used constantly on set. Widening the source widens the band. Moving the source closer widens the band. Moving the subject away from the background widens the band, and if you move them far enough there is no perceptible shadow at all, only a gradient — which is why the standard fix for an ugly shadow behind someone is to walk them forward rather than to buy a bigger box.
| Setup | Ratio | Look | Shadow on a wall 6 ft behind |
|---|---|---|---|
| 36 in box at 3 ft | 1.00 | Very soft, heavy wrap | 72 in of transition — no edge at all |
| 36 in box at 6 ft | 0.50 | Soft with a defined edge | 36 in of transition |
| 36 in box at 12 ft | 0.25 | Directional, edges visible | 18 in of transition |
| 7 in reflector at 6 ft | 0.10 | Hard, crisp shadow | 7 in of transition |
| Midday sun | about 0.009 | Hard, sharp-edged | Around 0.6 in of transition |
The cost of softness, in stops and in tolerance
Diffusion works by absorbing and scattering, so everything soft costs output. How much depends on the specific combination and there is no universal figure — a shoot-through umbrella, a single-diffused box, a double-diffused box, and any of them with a grid added, are four different numbers on the same head, and the grid is often the most expensive layer. The published figure from the maker is a starting point; a meter reading at a fixed distance is the real answer and takes two minutes.
The second cost is positional tolerance. A source close enough to be soft is also close enough that the inverse square law bites hard across the subject: the person nearer the light is meaningfully brighter than the person half a step behind them. Backing the light off makes the group even and makes the light harder, and there is no setting that gives both. Group shots therefore tend to use big sources further away, or several sources, and accept a firmer look than a single portrait would get.
Below about twice the source width, the inverse square arithmetic itself starts overstating the falloff, because different parts of a broad surface are at different distances from the subject and the light stops behaving like it comes from a point. That is a real effect and it is in the subject-lighting sweet spot, so a meter reading beats a calculation whenever a big modifier is used close. The light falloff calculator flags the same boundary from the other direction.
What the ratio does not tell you
It says nothing about the quality of the light within the shadow transition, which depends on how evenly the modifier is lit across its face. A box with a hot centre behaves partly like a smaller source, which is what internal baffles are for. It says nothing about spill, which depends on the depth and shape of the modifier and on whether it has a grid. And it says nothing about colour, which is a fixture and diffusion property.
It also assumes the surface you care about is roughly facing the light. A source that is very soft head-on is much less forgiving when it rakes across a surface at a shallow angle, because the ratio that matters is the one measured across the direction of the shadow. For a long horizontal rectangle lighting a standing figure, the vertical dimension of the box is what governs the shadow under the chin and the horizontal dimension is what governs the shadow beside the nose. Enter whichever of the two matters for the shadow you are trying to control.
Questions people ask
What size softbox do I need for portraits?
Ask the question as a distance rather than a size, because the two are inseparable. A source that is roughly as wide as its distance from the subject looks very soft, and half that is still clearly soft. If you shoot head and shoulders at three or four feet, a two to three foot modifier is in the soft range and a bigger one adds little. If your room forces you to work at eight or ten feet, the same modifier is no longer soft and you need something much larger to get the same look. Small studios get soft light cheaply; large ones pay for it.
Why does my softbox still throw a hard shadow?
Almost always because it is too far away. Softness is the source width divided by the distance, so a three foot box at twelve feet has the same ratio as a nine inch reflector at three feet, and it will produce a similarly defined shadow edge. The second common cause is a background right behind the subject: the shadow transition on a surface widens with the subject-to-background distance, so a person standing against a wall gets a hard-edged shadow no matter what the modifier is. Move them forward first, then worry about the light.
How many stops does a softbox lose?
It depends entirely on the modifier and the layers in it, which is why this page takes it as an input rather than assuming. A single layer of diffusion is a smaller loss than a double, and adding a grid can cost as much again as the diffusion did. The reliable route is to meter your own head at a fixed distance bare, then with the modifier, and note the difference on the bag. Manufacturer figures are a starting point but they are measured under conditions that are rarely stated.
Is a bigger modifier always softer?
Only at the same distance. Softness is a ratio, so a four foot box at ten feet is less soft than a two foot box at four feet. Bigger modifiers matter when the room or the shot forces the light to be far away, or when the subject is large enough that a small source would fall off visibly across it. They also cost more output, take longer to rig, and are less controllable — the spill from a big source goes everywhere unless it is gridded, and gridding it costs more light again.
Why are two people at the same light unevenly exposed?
Because the near one is closer to the light, and the exposure difference is the square of the distance ratio. A light four feet from the near person and six feet from the far one is a difference of about 1.2 stops, which is unmistakable. The fix is to move the light back so the ratio of the two distances shrinks, at the cost of the light getting harder, or to move the two people to the same distance from the light rather than the same distance from the camera. The light falloff calculator gives that stop difference directly.