A satin under a clear sky is dull across the sun and bright along it
Worth reading first: A satin mirrors the top of the sky · No cloth shines under a sky · A lamp off the mirror lights a satin only across its floats.
A satin mirrors the top of the sky, and under cloud that is enough to make it the brightest weave in view. An overcast sky is three times as bright overhead as at the horizon, a satin’s long flat crowns face nearly straight up and mirror the bright top of it, and a plain weave’s crowns, curving over every crossing, mirror a wide ring that reaches down to the dimmer sky and the ground. Seen from above under cloud, an eight-end satin sends the eye 1.37 times what plain does.
That essay chose cloud for a reason it gave: an overcast sky has no disc and its brightness depends only on how high a direction is. A clear sky is different in both respects. It has a sun, which is a lamp of half a degree and needs the lamp essays’ arithmetic; and its dome is brightest round the sun and along the horizon and darkest about a right angle from the sun — nearly the reverse of the overcast gradient.
The prediction left open was that under the clear dome a satin seen from above might mirror a darker part of the sky than plain does, and so look duller everywhere except near the sun’s own glint. The calculation says something less tidy and more useful. A satin under a clear sky is a mirror of the sky’s pattern. It is duller or brighter than plain depending on which patch of sky its crowns face, and that depends on where the viewer stands relative to the sun.
The sky a clear day actually has
A clear sky’s brightness is not a guess. The CIE and ISO publish fifteen standard general skies, from heavy overcast to cloudless, each a formula for the radiance in any direction as a multiple of the zenith’s. The one used here is type 12, clear with low turbidity — the standard clear sky — and it has two factors. One falls off from the horizon towards the zenith; the other rises steeply towards the sun and a little again opposite it.
With the sun 40 degrees up, the sky five degrees above it is 6.6 times as bright as the zenith. The horizon is bright all the way round, 2.31 times the zenith even on the side away from the sun. The darkest part of the sky, two thirds of the zenith, lies 58 degrees up on the far side — a band about ninety degrees from the sun. The zenith itself sits between that band and the bright sky round the sun, neither the brightest nor the darkest place overhead.
That is the whole difference from cloud. Under cloud the zenith was the brightest patch of the dome and a weave that mirrored it won. Under a clear sky the zenith is middling, and the bright patches are off to the side — where the steeper facets of a plain weave are more likely to look.
From overhead the two weaves come out nearly level
Seen from directly above, a satin’s crowns mirror the zenith and a plain weave’s mirror a ring round it, as under cloud. What changes is what the ring contains.
With the sun low, the eight-end satin seen from above sends 0.93 of what plain does at a sun of 15 degrees and 0.99 at 25: the zenith is the dim middle of a sky whose bright parts are low, and plain’s tilted facets reach them. With the sun at 40 degrees the two are level, 1.02. Only with the sun high does the satin win clearly — 1.07 at 60 degrees, 1.14 at 80 — because a high sun brightens the sky round the zenith, which is what the satin mirrors.
None of those is the overcast sky’s 1.37. From overhead on a clear day the satin and the plain weave differ by a few per cent, and for most of the day, at middle latitudes, by less than the difference between two lots of the same cloth.
The ground decides the overhead comparison
The steepest facets of a plain weave, seen from above, mirror not the sky but the ground. Under cloud the ground was dim — a fifth of the overcast sky’s light sent back — and plain lost by seeing it. On a clear day the ground is lit by the sun as well, and the sun’s direct light on a horizontal surface is several times the whole dome’s.
On a black ground the satin seen from above is 1.12 times as bright as plain, which is what the clear dome alone does. Light the ground as a clear day lights it and the advantage goes: on grass or a pavement of albedo 0.2, with the sun’s direct light on the ground four times the dome’s, the satin is at 1.02; with eight times, 0.98. On snow or white stone the plain weave wins outright, 0.91 and 0.88. The overhead comparison on a clear day is mostly a statement about what the cloth is lying on, because the weave that sees more ground is the one a bright ground favours.
How strong the sun is on the ground is not part of the sky model: the standard skies give the distribution of the dome’s radiance and say nothing about the disc’s own strength, which depends on the air’s clarity and the sun’s height. The default here, four times the dome, is chosen, a diffuse share of a fifth of the day’s light, and the figure shows both sides of it.
Tilt the view and the sky’s pattern appears
The overhead view is the one the overcast essay featured, and on a clear day it is the least interesting. Bring the viewer down and round, and the satin’s crowns start to mirror particular patches of a sky that has very different patches to offer.
Facing the sun across the cloth, the satin’s crowns mirror the sky just beside the disc as the viewer comes down, and the satin brightens: 1.23 times plain at a 60-degree view and 1.38 at 30. With the sun behind, they mirror the low sky on the far side, which the clear sky brightens towards the horizon: 1.07 at 60 degrees and 1.37 at 30. Across the sun’s line, they mirror the dark band a right angle from the sun, and the satin dulls: 0.86 at 60 degrees, 0.80 at 30.
A plain weave cannot do any of that, because its facets point every way at once and average the sky they see; it changes with the view far less. The satin does not average. It sends the eye one patch of sky, so it is exactly as bright, relative to plain, as that patch is relative to the sky’s average.
Every direction at once
Three lines of sight are a sample. The whole of the question is every place a viewer could stand.
The map is a cross. Along the sun’s line, towards the sun and away from it, the satin is the brighter cloth, by up to 1.39 at a 39-degree view. Across the sun’s line it is the duller, down to 0.76. Near overhead the two are level. With the sun at 40 degrees, the satin is duller than plain from 57 per cent of the directions a viewer might stand in.
The sun’s height moves the balance. With the sun at 25 degrees the satin is duller from 68 per cent of directions, and both extremes widen, 0.72 to 1.42; with it at 70 degrees, from only 28 per cent, 0.86 to 1.22. A low sun makes the satin mostly the duller cloth and a high one mostly the brighter, which is the overhead result spread over the dome — but at no hour is it uniformly either, as it was under cloud.
Dull with a glint, or bright with a glint
The question the overcast essay left was whether a satin on a clear day is dull-with-a-glint or bright-with-a-glint. It is both, depending on where the viewer stands, and the answer has a shape. Along the sun’s line the satin is bright — brightest towards the sun, where it passes into the glint of the disc itself. Across it the satin is dull.
The glint is not drawn. A facet that mirrors the disc sends the eye a radiance tens of thousands of times the zenith’s, and the share of a satin’s surface that does so is decided by how wide a float’s highlight is and which way the floats run, which the lamp essays priced. Here a facet that mirrors the disc is given the sky beside it instead, so that the map shows the dome without the sun in it. The marked point on the map is where the glint is, at the viewer facing the sun at the sun’s own elevation. Round it the satin is already bright before the glint begins.
That also says why a satin on a sunny day looks so changeable as it moves. A plain weave’s brightness is an average over the dome and changes slowly with the view. A satin’s is one patch of a dome whose patches differ by a factor of ten, and a draped satin presents every orientation at once, so the sky’s pattern is laid across its folds.
The twill sits between, and lower
The 2/2 twill does not behave like a weaker satin. From overhead it is below plain until the sun is 55 degrees up — 0.90 with the sun at 15, 0.97 at 40, 1.08 at 80 — and below both satins at every hour. That it falls below both is not what a middle weave would be expected to do, and the averages do not explain it: its facets’ slopes lie between the other two weaves’, but the patches of sky they mirror are a different selection, with less of the bright low sky than plain’s ring and less of the zenith than the satins’ crowns. Under cloud the twill was most of the way to the satin, at 1.25 against 1.37; under a clear sky it is the dullest weave of the four from overhead for most of the day. The five-end satin tracks the eight-end within a few hundredths throughout, consistent with lustre being a length times a width: both satins’ floats are long enough to carry a flat top, and under a sky it is the flat top that decides what a crown mirrors.
What a finish would do, not computed
Two finishes change the facets, and the mechanism says which way each should move the pattern without the arithmetic having been run. A calender buys the width of a crown’s flat top, and under cloud a calendered cloth of any weave mirrored the zenith; under a clear sky every calendered weave should therefore become a mirror of the sky’s pattern, the plain weave included, and the difference between weaves should shrink while the difference between viewing directions grows. A hair layer veils a highlight by scattering what the crowns would have mirrored, and under a clear sky it should flatten the cross towards the plain weave’s average. Both are predictions about the cross’s contrast, and neither number is computed here.
Why this is the same physics as cloud
Nothing here is new about the cloth. The facets, their slopes and the weights they carry are the same height field the account of shine has used throughout. What has changed is only the sky, and that is the point worth carrying: whether a satin outshines plain is not a property of the satin. No cloth shines under a sky found that a uniformly bright hemisphere makes every weave alike, and the overcast essay found that a sky brighter overhead makes the satin win. A clear sky, bright and dark in patches, makes the satin a picture of the sky.
A room is a sky of a kind too. A room lights a satin at the harmonic mean of its lamps, and the same reasoning applies there: the satin is as bright as the light its crowns face, and a room’s lamps are its bright patches.
How the sky’s reflection was computed
The cloth is the sheeting, with plain, 2/2 twill, five-end and eight-end satin woven at its setts, and its surface is this collection’s height field of crowns. Every visible facet mirrors one direction into the eye, for a view direction , and sends that direction’s radiance; a facet’s weight is the area it presents to the eye, . A weave’s specular radiance is the weighted average, and the figures divide each weave’s by plain’s.
The sky is the CIE standard general sky 12: radiance proportional to , with for an element at zenith angle and for one at angle from the sun, normalised to the zenith. The disc is left out: a facet that mirrors within the sun’s quarter-degree radius is given the sky’s radiance there. The ground sends its albedo times its illuminance over π: the dome’s horizontal illuminance, integrated over the standard sky, plus the sun’s, taken as a stated multiple of the dome’s.
Required of it, and checked: that the formula gives the zenith exactly one; that the sky beside the sun is several times the zenith and the sky a right angle from it darker than the zenith; that the dome’s illuminance is finite; that a sun overhead makes the dome the same in every azimuth; and that with a low sun, from overhead, the eight-end satin is duller than plain. A sun at or below the horizon, a sky direction below it, an albedo outside nought to one, a negative sun, a view from below the cloth and an unnamed weave are refused.
What the calculation leaves out
Only the specular part. Every number is the mirror-like reflection, which is what makes a satin look like a satin; the diffuse part, the same for every weave of one fibre, sits under it and lowers every contrast here. A white satin’s diffuse light may be the larger part, so the ratios are contrasts of the sheen and not of the cloth’s whole brightness.
No masking. A facet the view grazes is counted by its projected area and never hidden by a neighbouring crown, so views below about 45 degrees overstate the steep facets’ share — the plain weave’s more than the satin’s. The thirty-degree numbers are upper bounds on how far the pattern spreads.
One clear sky. Type 12 is the standard cloudless sky with clean air; hazier skies, types 13 to 15, have a broader glow round the sun and a less pronounced dark band, and would shrink the cross. The sun’s strength on the ground is an input.
And the glint. It is left out by construction. Near the sun’s mirror direction the satin is the brightest thing in the scene by orders of magnitude, and nothing above describes that.
Who found what
The standard skies are the CIE’s, published as ISO 15469 and CIE S 011, drawing on Kittler’s and Perez’s work on sky luminance distributions through the 1990s; that a satin’s lustre comes from the specular reflection of its long floats is old. What is added here is the satin’s specular reflection of a clear sky: that from overhead it is within a few per cent of plain and the ground decides which way, where under cloud it was a third brighter; that tilting the view makes it a mirror of the sky’s pattern, up to 1.4 times plain along the sun’s line and three quarters of it across; and that the share of directions from which it is the duller cloth falls from about two thirds with a low sun to about a quarter with a high one.
Still open: what a draped satin shows
Every figure here is a flat cloth. A satin dress or a curtain in sunlight presents every orientation of its crowns at once, and the map above becomes a picture laid across its folds: each fold faces some patch of the sky, and the satin there is as bright as that patch. The prediction is a pattern of light and dark across a drape that follows the sky rather than the folds’ depth — dark where a fold’s crowns face across the sun’s line, bright where they face along it — and so a draped satin under a clear sky should change its pattern when it is turned, while a draped plain weave keeps its shading.
Computing it needs the drape’s surface normals combined with the cloth’s own facets, a composition of two slopes that has not been computed. A photograph of one satin panel, hung in sun and rotated a quarter turn, would show directly whether its light follows the sky.
Shares its objects with
Essays naming at least two of the same things, that neither author linked.
- A damask is its own complement — both name float, lustre, satin
- The float in a knit — both name float, lustre, specular reflection
- A brocade weft floats as far as the next figure — both name float, satin
- A calender spends the compression for good — both name float, lustre
- A damask's edge floats further than its figure — both name float, satin
- A figured cloth has a step in its surface — both name float, satin
Named objects
A flat tag is an object no other essay names yet.