VARIABLE DOT OPTIMIZATION
Variable dot inkjet printers produce dots of variable dot size, typically three or more discrete sizes, by dynamically varying the pulse waveform to each nozzle to jet droplets of different volumes. When applied properly, this technology can produce smoother tone gradients and improved dynamic range. Unfortunately, it also adds a significant layer of computational complexity to the halftoning process.
With a single dot size, the halftone algorithm must decide if a drop should be jetted based on the current input pixel value combined with the quantization error accumulated from already processed pixels. But when multiple dot sizes are possible, the halftone algorithm must decide not only if a drop should be jetted, but which drop size should be used to produce the smallest possible error. This is typically achieved by setting static input thresholds; if the input pixel value exceeds a certain threshold value, the next largest drop size is used.
However, there are many factors that must be considered when determining these threshold values, factors that involve the physical printer, ink, media, environment, and the physics that define how all four interact to produce a dot on the page – or more specifically, hundreds of millions of dots. Moreover, not all drops are created equal. The pulse waveform used to jet a single droplet is often complex. It determines not only the volume of the droplet, but its velocity and trajectory as well. Precise timing offsets are necessary to ensure droplets of each color land in alignment.
But, when the waveform is altered to produce a smaller or larger droplet, the velocity and trajectory change along with the drop volume. In many cases, dot alignment can suffer when multiple dot sizes are used. In areas of the image with high-frequency detail, this may not be obvious; but, in low-frequency areas such as mid-tone color fills, dot misalignment can result in graininess and banding, showing patterns of light/dark bands or low-frequency chromatic oscillation. This can often be minimized, however, by printing these regions with a fixed dot size, or by limiting (or even eliminating) one drop size to favor others. This can be further enhanced by injected a small amount of stochastic noise, with an amplitude inversely proportionally to the input frequency of the image data, into the diffusion algorithm feedback loop to further disrupt these low-frequency artifacts.
As an added benefit, proper optimization and tuning of the variable dot halftoning algorithm can very often eliminate the need for setting ink limits and linearization curves during calibration, helping to ensure the printer achieves its maximum gamut and tonal range for the selected print mode. Even in situations where the available drop size is too large for the selected resolution, variable dot optimization can avoid over-inking without resorting to the setting of artificial ink limits.
With a single dot size, the halftone algorithm must decide if a drop should be jetted based on the current input pixel value combined with the quantization error accumulated from already processed pixels. But when multiple dot sizes are possible, the halftone algorithm must decide not only if a drop should be jetted, but which drop size should be used to produce the smallest possible error. This is typically achieved by setting static input thresholds; if the input pixel value exceeds a certain threshold value, the next largest drop size is used.
However, there are many factors that must be considered when determining these threshold values, factors that involve the physical printer, ink, media, environment, and the physics that define how all four interact to produce a dot on the page – or more specifically, hundreds of millions of dots. Moreover, not all drops are created equal. The pulse waveform used to jet a single droplet is often complex. It determines not only the volume of the droplet, but its velocity and trajectory as well. Precise timing offsets are necessary to ensure droplets of each color land in alignment.
But, when the waveform is altered to produce a smaller or larger droplet, the velocity and trajectory change along with the drop volume. In many cases, dot alignment can suffer when multiple dot sizes are used. In areas of the image with high-frequency detail, this may not be obvious; but, in low-frequency areas such as mid-tone color fills, dot misalignment can result in graininess and banding, showing patterns of light/dark bands or low-frequency chromatic oscillation. This can often be minimized, however, by printing these regions with a fixed dot size, or by limiting (or even eliminating) one drop size to favor others. This can be further enhanced by injected a small amount of stochastic noise, with an amplitude inversely proportionally to the input frequency of the image data, into the diffusion algorithm feedback loop to further disrupt these low-frequency artifacts.
As an added benefit, proper optimization and tuning of the variable dot halftoning algorithm can very often eliminate the need for setting ink limits and linearization curves during calibration, helping to ensure the printer achieves its maximum gamut and tonal range for the selected print mode. Even in situations where the available drop size is too large for the selected resolution, variable dot optimization can avoid over-inking without resorting to the setting of artificial ink limits.