The Fine Art of Inkjet Printing
Today's digital files need to be high resolution to hold crisp detail and smooth gradation. But did you know that the resolution in digital files and for digital printers isn’t always the same.
Taking inkjet printing to the next level
High resolution in digital files and printers is needed to hold crisp detail and smooth gradations, which have become hallmarks of fine photography. Resolution in digital files and digital printers isn’t the same. Resolution for digital files is measured in pixels per inch (PPI); pixels are square and surrounded by other pixels. Resolution for digital prints is measured in dots per inch (dpi); dots come in a variety of shapes and are surrounded by the colour of the substrate to which they’re applied-typically to the white paper.
Pixels come in millions of colours, while dots come in only a few (typically variants of cyan, magenta, yellow, red, green, blue, orange or black). Precise combinations of these limited colours are clustered to create the appearance of many more colours. So, it takes many dots to reproduce a single pixel. The print resolution then is always higher than file resolution.
Smaller dots usually produce finer detail and smoother transitions than larger dots. The dots produced by today’s printers are very small-smaller than the width of a human hair. Dot size is measured in picoliters (millionths of a litre) and droplet sizes today may be less than 2 picoliters.
Dot selection, structure and placement
Dot size isn’t the only factor in achieving a smooth and detailed visual appearance; selection, shape and placement are also important. Today’s inkjet printer drivers/RIPs use sophisticated software algorithms to determine what types of dots are generated and where they’re placed.
Selection: The choice of the dot colour is extremely important, not just for reproducing colour accurately but also for reproducing smooth detail. For instance, lighter inks hold highlight detail because lighter dots are less visible, so highlights appear smoother. To test this, make a print with black ink only and look closely at smooth areas and highlights; the image becomes coarse. Compare that to a print made with all of the ink colours; the image becomes smooth. To achieve the finest detail and smoothest transitions, you need to use most of all of the inks.
Shape: While most dots are ellipses, some are triangular, square, or star-shaped.
Placement: Larger well-placed dots can outperform poorly placed small dots. Screening frequencies determine dot placement. Offset presses typically use halftone frequencies, where the dots are placed in a regular pattern while their size varies (smaller dots for lighter values, larger dots for darker). To avoid moiré patterns, halftone frequencies, where the dots are placed in a regular pattern while their size varies (smaller dots for lighter values, larger dots for darker). To avoid moiré patterns, halftone frequencies are limited to four inks because the pattern of each ink’s dot is offset at different angles and there aren’t enough angles to accommodate more than four inks. When more than four inks are required, stochastic screening frequencies are used-dots are randomly placed while their size remains the same (fewer dots clustered for lighter values, more dots clustered for darker). The latest inkjet printer screening frequencies are a hybrid of both, using precise random placement and variable droplet size.
Ink-delivery systems also vary: Epson printers use piezoelectric technology while both Canon and HP use heat-based technology. Currently, piezoelectric systems are more precise, rendering crisper detail and smoother gradation; but will future advances in heat-based technology and increasingly small dot structures deliver similar results without such a high degree of precision?
File resolution for Epson printers
For Epson printers, fine line detail is best rendered when printing at a file resolution of 360ppi. Image quality doesn’t deteriorate significantly until file resolution drops to less than 180 PPI and quality is slightly compromised when printing at resolutions above 720 PPI. File resolutions that are even multiples of the printer’s resolution deliver incrementally higher quality than uneven multiples.
Resampling and sharpening
To achieve better results, print files set to the print size you desire at resolutions that are even multiples of the printer’s resolution. For example. 360 PPI is ideal for 1440x2880-dpi resolution and 300 PPI is ideal for 1200x2400-dpi resolution. When you do this, the printer driver won't resample (add/subtract pixels) the file before printing but if you don’t, the printer driver will resample. This results in only a slight increase in the detail a file content and allows for sharpening after resampling.
Resampling softens images more than imprecise file-resolution specification. Avoid minor resampling to achieve the desired print size; instead, redistribute file resolution to a less-than-ideal resolution (not less than 180 PPI). When major resizing is necessary, resample the file in Photoshop as little as possible, and sharpen afterwards.
Precise image sharpening is more important than precise file resolution. Sharpening makes image details appear crisper. Undersharpened images appear soft and oversharpening can make smooth gradations appear granular.
Resolution is important once again. Higher-resolution files sharpen more precisely than lower-resolution files. A compelling argument can be made for sharpening specifically for the substrate. Similarly, sharpening may be adjusted for print size. Don’t confuse the effects of resampling and sharpening with the effects of the file resolution.
Big vs. small
It's said that small prints require higher resolutions than big prints and it’s true that higher resolution is required to reproduce the same level of detail packed into a smaller space. But both scale and detail are important factors gained from making large prints.
The notion that large prints don’t require such high resolutions based on a theory that there’s an ideal viewing distance for artwork: two to three times the diagonal dimensions. To fill your field of view with a large image, you’d have to stand at a greater distance from it than you would from a smaller image - a distance at which you cannot see dots or lower resolution. (the human eye can make distinctions down to .005”, below which it fails to resolve differences.) In actuality, because people like to interact with the artwork, the viewing distance of artwork is variable: less than a few inches to more than a few feet. So I recommend you make prints that hold up at any distance, using the maximum resolution possible for every size. (Note: Sharpening also gets into this debate, as bigger files with more pixels sharpen more precisely than smaller files with fewer pixels.)
Substrate impacts print resolution
The type of substrate you print on significantly impacts print resolution. Glossy papers resolve more detail than matte. This can be overstated with very smooth matte papers but is quite significant with textured or fibrous substrates. Consequently, glossy surfaces can benefit more from higher resolution files than matte surfaces.
When it comes to resolution, you can end up splitting hairs unnecessarily. The differences between prints made with varying file resolutions within an acceptable range are quite subtle. The rules to follow are: Set file resolution as close to optimum as practical; resample as little as possible, and sharpen precisely.
The resolution of detail can have a profound impact on image quality and it's not just a technical concern; the level and type of detail in images is also an aesthetic consideration. Decide to use more or less detail in your images and to statement you wish to make. It can have a profound impact on image quality.
Photographic data in a test file should include textured areas with crisp detail and edges, smooth areas (preferably with gradients), very bright highlights, very dark shadow detail, and a full range of tones in between. It should also include the full spectrum; flesh tones and memory colours are particularly useful. As it can be challenging to find one image that has all of this information, consider using many images combined in a single document.
Synthetic data (rendered by software) in a test file should include neutral values. Extremely saturated values can also be useful, particularly pure CMY and RGB values.
Linear neutral gradient test files, with and without posterization, are particularly useful in determining whether a print has good neutrality and grey balance; whether a driver or profile is capable of rendering smooth transitions between closely matched values; to pinpoint values at which dot structure changes; and to identify the minimum and maximum printable values, given a particular printing combination (ink, paper, driver, ink limit, print speed and profile).
Synthetic gradient test file
Here’s how to make your own synthetic gradient test file. Create a new file, then use the Gradient tool (G) to draw a black/white horizontal (or vertical) linear gradient. Select half of the area and posterize it (Image>Adjustments>Posterize). The value entered for the posterization will determine the final number of intervals (we used 20 increments of 5% from 0-100). Next, make another selection and posterize it, generating greyscale increments in 1% increments from 0-10. Finally, in another selection, make a gradient with greyscale values from 90-100 and posterize it, again generating 1% increments.
If you want crisp edges, you can make this manually, filling hard-edged selections with specific values. It’s also useful to use the Type tool (T) to create and position numbers specifying the values of each patch. Here’s our grey gradient test file (test file not available).
Once you create this test file, save it to drop into other images whenever you feel it would be useful. If you make a conversion from one colour space to another, check to see that the values haven’t changed significantly after the conversion. The primary concern in pasting these values into an editing space with a different gamma (that is, 1.8-2.2 or vice versa) is that it will alter the appearance of the values. (Note: Dot Gain 20%, Colour match and ProPhoto are gamma 1.8; while SRGB and Adobe 1998 are gamma 2.2)
You can use any colour mode to generate and evaluate results (LAB, HSB, RGB, CMYK or greyscale), as long as you’re comparing numbers on the same scale: 0% is pure white with no detail; 100% is pure black with no detail. For this type of test, RGB and Greyscale are the easiest to use. Equal values in RGB are neutral. Greyscale is most useful for looking closely at luminosity values separate from the effects of hue and saturation.
Using your test file
After you’ve stored your test file, print it and use it to evaluate materials you’re considering using. Look at the grey and compare it to other greys. Is it neutral? Or does it contain a slight colour cast? If it contains a colour cast, evaluate your colour-management practices: how you navigate the printer driver and the light under which you view the print. Then re-evaluate the profile used by comparing it to another. This isn’t a difficult problem to address by adjusting the file in Photoshop before printing. For example, if the neutral values are only slightly inaccurate, make a small colour adjustment to each file before printing.
Now, look at the smooth gradient with posterization, particularly in the darkest values. Are the transitions smooth? Or do they move from one tone to another abruptly? If they transition too quickly, try using another profile. (Another brand of printer driver may also eliminate the problem, but this is usually costly and unnecessary.) This is a difficult problem to address by adjusting the file in Photoshop to compensate for inaccuracies in the profile or driver.
Next, look at the 5% intervals of grey. Are the intervals between the values equal in brightness? Or do some steps between intervals seem greater than others? If they appear unequal, you should also be able to see uneven transitions in the smooth gradient within the same values, although perhaps not as readily. Are the intervals between the values equal in hue? Or do some patches have different tints than others. Grey balance is a term used to describe uniform tint throughout the tonal scale and poor grey balance (sometimes called a colour cross) adversely affects both neutral and fully saturated images and image areas. If these problems exist, the cure for both may be found again by switching profiles (and as a last resort, the printer driver). This is a difficult problem to address by adjusting the file in Photoshop to compensate for inaccuracies in the profile or driver.
And now look at the lightest and darkest values rendered in 1% increments. At what value can you see detail? At what value do you first achieve maximum black? On the latest inkjet printers, a 1-2% dot will begin to reproduce detail, and maximum black is achieved with values in the range of 96-98%. Your results will vary depending on your choice of paper, ink, driver and profile. If you’re not achieving good detail in highlights and shadows, look first to the printer profile and second to the printer driver. The latest printer drivers have begun to incorporate sophisticated tools to help you customize the ink limit. Some offer shadow and highlight sliders that allow you to affect the amount of ink laid down in one or both ends of the tonal scale without dramatically impacting mid-tones. Classically, inkjet prints tend to over-ink shadows so some compensation is necessary. This isn’t a difficult problem to address by adjusting the file in Photoshop before printing. If reproduced highlights are too light, darken them before printing. If reproduced shadows are too dark, lighten them before printing.
A few tips
- A test file is the first thing you should print when you use a new substrate or profile. It will give you the data you need to objectively evaluate them.
- Reproduce this same test file on a variety of substrates to evaluate their differences. This will help you identify both technical and aesthetic aspects of different materials.
- Keep your test prints. Old test prints are excellent for comparison with new test prints you’ll make when you evaluate advances in drivers, profiles, inks and substrates. And there will be advances: Every year there’s something new.
- These are all characteristics that can only be identified in a print. You can’t soft proof them. The proof is in the proof.
A little testing upfront can ensure that you can get the finest results possible with the tools you’re using and it can help you decide when to replace the tools you’re using.