Halftone Drilldown

Virtually “Wet Textiles” & Seeking Moiré

	Not all digital artifacts are bad. Some, like the ones you see here, may even be desired. Experience is the best guide I know, and the practical application of theory demands trials and (one hopes) error correction, therefore the images in this post are part of an investigation into the Moiré patterns produced with chosen sets of channel degrees employed by Photoshop Elements' “Color Halftone” filter found at the top of its “Pixilate” menu (i.e.: Filter→Pixilate→Color Halftone). The top right monochrome thumbnail was used to produce all the “test” images here. The table below aligns columns with particular sets of degrees of rotation in each color channel with rows containing increasing radii (in pixels) for each set. The first column's set of rotations are at 45 degree increments which generates a square grid Moiré. These angular values were also used to create the two larger images above the table: image top-left was produced using a radius of 7 pixels, the one to the right was metered to 20 pixels. Both are full size, whereas the images below have been scaled down to fit this layout.
		Color Halftone Channel Rotation
r↓		045   090     135       180	030   060     120       150	095   100     105       110	072   144     216       288	000   072     216       288	091   034     254       009	Ch. 1 2  3    4
	4
	8
	10
		.

The second column uses angles all divisible by 30°. Analog clock-faces separate each hour by this same angular distance, so we can visualize this set with the corresponding hourly marks of (moving counter clockwise) 2, 1 , 11 and 10 which correspond to 30, 60, 120 and 150° from a horizon running through 9 and 3 o'clock.

The third column uses a 5 degree step series producing a spiral-like pattern at larger sized radii. The next column divides 360° into 5 equal steps, though the fifth step from the last color channel (blacK) isn't actually made.

Still thinking of time, 72° (360° / 5) is equivalent to 12 minutes and we can trace the fourth set's angular values around the dial at (again, counter clockwise) 3 minutes after, 9 and 21 minutes before, and 27 minutes after any hour as shown here at right. The fifth column uses the same step size but decreases the first two channel values by 72°, so Cyan Channel 1 and Magenta Channel 2 would respectively align with 15 and 3 minutes past.

The final, sixth column uses a pseudorandom set of values that was generated using the Windows command shell environment variable %RANDOM%. Close enough for this experience ... or experiment.

Armed with these examples I have a better sense of which values might be best suited to various situations, like these.

Clock images created using a cleaned up “220px-Analog_clock_base.png” at wikimedia.org/wikipedia/commons