Epson Stylus Pro 4880 printer user experience report
Part six: 8–bit vs. 16–bit printing
As you know, in an 8–bit image each colour channel (red, green and blue) contains 256 tonal values, which translates into 16.7 million possible colours per pixel. In a 16–bit image, on the other hand, each colour channel boasts 65,536 tonal values, which results in 281 trillion possible colours per pixel. Therefore, printing in 16–bit mode should result in smoother gradations and improved transitions – at least in theory. Indeed, according to Epson the new 16–bit driver technology is one of the key features of the 4880. To be able to do 16–bit printing, however, you have to have each of the following:
16–bit capable Operating System. Mac OS 10.5.X supports 16–bit printing: "Leopard can use enhanced numerical precision to unlock the printing potential of pro–level color ink jet printers, using up to 16 bits per color channel." Note that both Windows XP and Windows Vista do NOT offer this feature.
16–bit capable printing application. For most photographers this will mean upgrading to Photoshop CS4 as earlier versions of the software, including CS3, do not support 16–bit printing.
16–bit capable printer driver. Check with the manufacturer of your printer if your current printer driver supports 16–bit printing.
In other words, as of mid–2009 one has to use Mac OS 10.5.X, Photoshop CS4 and one of the latest printers with a driver that supports 16–bit to be able to do 16–bit printing.
To ascertain whether there is any visible (as opposed to theoretical) difference between 16–bit and 8–bit printing I conducted the following test: I chose four photographs and printed each image twice, in 16–bit and 8–bit mode (converting 16–bit image to 8–bit mode and saving it as a separate file as the last step before printing). For this test I used the printer with the latest printer driver (v6.12) and Adobe Photoshop CS4 running on MAC OS 10.5.6. This combination of hardware and software provides a complete, beginning–to–end 16–bit printing pipeline.
I chose this, this and this photograph for the test. These images boast very subtle colours, delicate tonal gradations and, between them, a great variation of colours, all of which can potentially benefit from 16–bit printing. The photographs were taken with a Hasselblad V series system and Fujifilm slide film. Transparencies were then scanned with an Imacon 848 film scanner at 3000 ppi in 16–bit mode into Adobe RGB (1998) colour space, thus producing true 16–bit TIFF files. The files were further post–processed in Photoshop CS4, all along in 16–bit mode and AdobeRGB colour space, and then reduced in size to produce 40cm by 40cm (16 inch by 16 inch) prints at a printing resolution of 360 dpi. I printed on Epson Enhanced Matte paper (which I use most of the time).
After the prints dried, I carefully examined and compared each pair of images, as well as showed them to several friends (both photographers and otherwise). And so here is my final verdict: no matter how hard we looked, we saw no difference between 16–bit and 8–bit prints—indeed, they looked perfectly identical. What, you ask, no difference at all? Nope, no difference whatsoever. Nothing, zilch, nada.
I, nonetheless, would still like to give 16–bit printing the benefit of the doubt. There probably might be some unique kind of images, colours or paper–and–ink combinations where 16–bit printing would produce a discernibly better transitions and gradations. Again, though, given the fact that I can see no difference in my tests whatsoever, I would not expect any possible improvements to be readily noticeable or of any fundamental significance. And if you shoot digital, where most current cameras do not go beyond 12–bit capture, the issue is even muter.
So is 16–bit printing nothing but marketing hype? Technically speaking, the claim that 16–bit printing produces better results is correct; in practice, however, I am afraid that, yes, for the most part it is marketing hype.
Part six: 8–bit vs. 16–bit printing