Re: Measuring yarn samples
Re: Measuring yarn samples
- Subject: Re: Measuring yarn samples
- From: Garth Fletcher <email@hidden>
- Date: Sat, 23 Jun 2007 21:39:38 -0400
- Organization: JacqCAD International (Fletcher Applied Sciences, Inc.)
Eric Bullock wrote:
in Colorsync-users Digest, Vol 4, Issue 227
Subject: Measuring yarn samples
A few quick notes about simulating textiles.
Bear in mind that with textiles you are not looking at a "flat"
uniform color field. Instead you are looking at **textures**
which include colors.
The problem is especially acute with Jacquard woven samples which
inherently have a complex 3-D surface texture. If you watch
textile folks evaluate a sample you'll see them pick it up and turn
it this way and that to see how the appearance changes depending
on the angle of lighting...
The "standard" yarn sample is a small card around which the yarn
has been wound - think of a flat spool. When you try to measure
this with a standard spectrophotometer you get very different
readings as you rotate the sample. When the illumination is along
the yarns you get few shadows; when it is cross-wise to the yarns
you get a lot of shadows. A spherical spectro helps average
out this effect by providing diffuse illumination. Bear in mind
however that these reading will indeed be an average, and will
not match the visual appearance of the sample at different
orientations under normal lighting. Your readings may not change,
but the fabric's appearance will...
Your problem is hopefully a bit simpler in that you are looking at
the cut ends of a bundle of yarns - there still is texture, but
almost no orientation differences when rotated (assuming you are
measuring normal to the surface). You may find that a normal
spectro will provide stable readings which do not vary with the
sample's rotation. If not, then one approach is to create swatches
large enough to fill the spherical spectro's aperture.
However, you will still have significant color shifts when you
change viewing angle relative to the surface. While you may be
able to get stable readings, they cannot fully characterize the
surface's appearance.
The problem is that you are trying to define a single solid color
to simulate what is inherently a colored texture. Many swatches
when examined closely will be seen to not be a flat uniform
color, but instead will show "grain", highlights, etc.
A printed simulation using solid colors is likely to appear
"flat" and uninteresting when compared to the real thing...
One approach which might be worth investigating is to print images
of the swatches, i.e., colored textures, instead of solid colors.
The color control issues then concern maintaining the color fidelity
from swatch image through to print rather than simply reproduction of
a single color.
Two approaches come to mind -
1) use actual scans/photos of your swatches. To do this you need
to solve the problem of creating a repeatable tiling which won't
show joins.
2) analyze the scans/photos to extract metrics such as spatial
frequencies and variations in saturation/lightness/hue, then use
these metrics to create a synthetic tiling pattern.
In either case you would need to determine a "standard illumination"
(color temp, orientation to surface, source geometry) which will
be used for all samples, and which will define the simulation's
environment.
Finally, never forget that the printed simulation can never match
the fabric's behavior as lighting changes. Such simulations can be
useful for narrowing the range of choice, but in my opinion are very
dangerous when used for final approval - a recipe for swallowing a
lot of expensive product because "it doesn't match what I ordered"
(which indeed will always be demonstrably the case).
Cordially,
--
Garth Fletcher
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Colorsync-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden