Why a render can never disagree with itself

The situation

3 May. The pre-production sample for TW-4180 is on the table at Sundargarh Apparel. The overshirt is Rushmere Taupe: a cotton twill body from Chandori Mills, and a woven nylon tape from Larkhall Trims down the placket and around the inside of the collar stand. The tape is meant to read as the same taupe.

In the virtual sample the studio approved in February, body and tape are the same colour. In daylight at the factory they are the same colour. Under the office lights they are not. The tape goes pink and the body stays neutral.

This is a metameric pair: two materials that match under one light and separate under another. Course 3.3 explains what that is and how a dyehouse fixes it. This lesson is about why nothing in the digital process could have found it, and why that will still be true next season unless a field changes.

Two colours, one number, one blind spot

Rushmere Taupe is one of 11 of the card's 32 colours mixed from three or four dyes: the greys, taupes, olives and browns. That is where metamerism lives, because many different recipes reach the same answer under one light and separate under another.

Both materials were entered in the library the way lesson 1 described: from the same three coordinates, because that is what the colour field held and what the feed carried. So the body cloth and the tape carry identical colour data. Not similar. Identical.

Now think about what a rendering engine does with that. It takes the material's colour, combines it with the light in the scene, and works out what the surface sends to the camera. Give it two materials with the same colour value and one light, and it produces two identical pixels. Change the light in the scene and it produces two identical pixels again, because the same change is applied to both.

A render cannot disagree with itself. Two materials carrying the same colour value match under every light the software can put them under. Not because the fabrics match, but because the data has no way to differ. The pair is non-metameric by construction.

What the data would have to carry

The fix is a field, and it is the same field lesson 1 asked for.

Store the curve for each material and the picture changes completely. A curve describes what the surface does at every wavelength, so two materials that reach the same coordinates by different routes have visibly different curves. They cross rather than sit on top of each other. A system holding both curves can work out each material's colour under a second light and report that they no longer agree. That is arithmetic on data you already own, and it costs nothing to run.

Note carefully what is and is not being claimed. Storing the curves does not make the render show metamerism. Most production rendering pipelines work in three channels, and they will still light both materials the same way. What the curves buy you is a check that can fail: a comparison run outside the render, on the library, that flags the pair before anybody looks at a picture. The render stays blind. You stop relying on it.

The last row is the commercial one. With curves, the physical cross-check becomes targeted instead of universal, and a targeted check is one that actually gets done.

The mistake nobody made, and how it was caught

The taupe problem was found on 3 May by a quality supervisor at Sundargarh, who put the sample in the light box before the fit session, because she puts every sample in the light box before the fit session. She was not checking the trim against the body. She was doing what she always does.

Trace the chain backwards and there is no bad decision in it.

• The studio picked a nylon tape because it wanted a smooth, dense edge on the placket, and a cotton tape would fray. Correct.
• Larkhall Trims matched the tape to the coordinates it was sent, and it matched them. Correct.
• Chandori Mills matched the cloth to the same coordinates on cotton. Also correct.
• Nylon takes a different dye class from cotton, so the two recipes could not have been built the same way even in principle.
• The virtual sample showed them matching, under three lights, and was signed off. Correct against the data that existed.

Nylon and cotton dyed to the same coordinates by different dye classes is close to a definition of a metameric pair, and that is exactly the situation a trim tape creates. So here is the pattern worth remembering. The risk is not "muted shades". It is one colour made on two substrates by two suppliers.

Of the 32 colours on the card, 8 appear on a garment where a second supplier's colour meets the body colour on a seam or an edge. Those 8 are the physical cross-checks that have to happen whatever the library or the render says. At Sundargarh, comparing two swatches in the light box costs about USD 24 in somebody's time — USD 192 for the season, and 20 minutes each.

What the taupe cost

The tape could not be rebuilt in nylon to hold under both lights in the time available. Larkhall supplied a polyester tape instead, from a dye class that could be matched on the curve.

USD 682 is not the interesting number. The interesting number is 9 days, and where those days landed. Lesson 6 shows the season had 20 days of float — spare time before a deadline — and lesson 5 spends another 11 of them.