Close-up on color printing

More than three decades after he left his job in a traditional printing shop, John Meyer still has the knack.

"I'd say that blouse is about 50 percent cyan, 35 percent magenta, and then, because it's got a bit of grey in it, about 10 percent black," he says, and he's just about on target, still able to eyeball a color and break it down into precise amounts of ink -- Cyan, Magenta, Yellow, Black.

That skill has played an essential role in Meyer's career at HP, where he's helped engineer a radical transformation in color printing from a costly, time-consuming process handled exclusively by skilled professionals to today, when almost anyone can quickly and easily print a good-quality color photo.

Meyer, who was part of the team that invented and developed HP's landmark thermal inkjet color-printing technology, talks about some of the technical achievements in HP Labs that made this transformation possible. Now director of the Hardcopy Technologies Lab, Meyer also discusses some of his group's research into the next frontiers of printing and imaging.

How has color printing changed over the years?

What we've done is engineer a radical transformation of the world of color printing. The world that used to exist was a world of craftsmen, skilled operators, expensive machines – it was a personnel-intensive world.

In the past, if you wanted a color picture on a piece of paper for, say, a magazine, you'd start with the color original – a slide, artwork, color print, etc. -- go through a process of color separation which, after a lot of corrections and adjustments by skilled people, another skilled person would convert it to a Cyan, Magenta, Yellow and Black set of half-tone films. Then you'd make a color proof -- a prediction of the printed result.

That’s a lot of people. What else was involved?

Next was page layout and imposition, in which films must be registered to each other with great precision. This required more skilled people.

Then you'd make the printing plates, involving more skilled people. Finally, you'd go to the printing press. Extremely skilled people were needed there -- the press operators who knew how to control the offset press. Then you might put the materials through a bindery. Finally, you've got your color picture.

The process was very closed. You, as the customer, got to maybe see the proof and possibly do a press check.

If you had a negative, you could go to a photo lab. There you'd work with lots of other skilled people, and they'd do a print for you on photo paper. If you wanted something 8x10, it could get expensive very fast. Again, it was all closed. You, the customer, stood outside and waited for the result to appear. Your primary function was to pay.

How does that compare with today?

Today, we have a world where the customer can buy a camera, color printer, PC – and can do something like this: Take a photograph. Remove the memory card. Place it in a slot in the printer – and print.

All these highly skilled people have gone away. You can print on photo paper or office paper. You can drop it into a document and print that. We've put all this in the hands of consumers –for less than $1,000.

How did we get from the complex, labor-intensive method to now?

The old printing world thought in terms of color in amounts of ink. The reason for this is that offset presses specialize in putting down a very controlled, thin film of ink. If you don't control the ink density carefully, the image will look bad.

Because a press is a tremendous investment, everything is driven toward making the press efficient. It takes a lot of skill – for example, a skilled color separator can look at a sweater or article of clothing and say, this is 40 percent Cyan, 22 percent Magenta and 5 percent Black, and be very, very accurate.

We've taken these processes and opened them up. The customer is involved all the way along. We've moved from densitometry, which is the measurement of amounts of colorants, to colorimetry, measuring color based on human perceptions, the way in which human beings see color. That's more customer-centric.

This is what we've been doing in last decade-and-a-half.

Where did you fit into all this?

I had an unusual combination of skills. I’d worked in traditional photo lithography. But I also had a PhD in low-temperature physics. After my degree, when I went out looking for a job, HP was looking someone who knew something about physics and about printing.

It was one of these splendid accidents of life. When I joined in 1979, inkjet was just an idea. Early on, my work was nothing about color. It was all about inkjet -- how to make the device survive, understanding the fundamental fluid mechanics and thermal dynamics of thermal inkjet.

(In thermal inkjet printing, tiny resistors create heat, the ink vaporizes and creates a bubble. As the bubble expands, droplets of ink are forced out of the nozzle onto the paper. The collapse of the bubble retracts the meniscus, pulling more ink into the print head from the cartridge.)

My work was to do thermal modeling to understand what actually happened when we turned the heater on -- what energy went into the surface and what went into the substrate. Also to study the process of nucleation, which is what happens when you heat a fluid to the point where it vaporizes and generates bubbles.

The problem was that something in the process was damaging the resistor. I was trying to understand how the damage mechanism worked and how we might get around that. We had very high thermal stress. We were heating something up very rapidly and that generated stress on the materials.

I did an analysis to determine if those stresses were actually the source of the damage, and what I learned was that they were not. Instead, the bubble collapse -- a process known as cavitation -- was the source of the problem.

The collapsing bubble formed a microjet of fluid that impacted upon the resistor, almost like a jack hammer. This was the source of the damage that eventually led to failure of the entire resistor. We solved the problem by creating a fluid flow architecture that moved the bubble off the resister.

What happened next?

After we introduced thermal inkjet, we began experimenting with the imaging capabilities of the device. Eventually, we discoveed a mode of operating the device called multidrop, where you could burst the nozzle at a very high frequency and send out up to 16 droplets at once.

That let you change the dot size, and if you could modulate the dot size you could change the level of intensity of the color. What is a picture but different levels of light to dark? At that point, we began to look at printing black and white pictures, and while doing that, it became evident that with simple steps we could make this a color printing device.

I went to a printing company in San Francisco and asked them to scan a color transparency (slide film) for me and give me the halftones that they would normally generate for an offset sheet-fed press.

(Halftone refers to the process of printing dots of different sizes to mimic the impression of different intensities of color. If the dots are spaced close enough we do not see them and the impression we have is of varying levels of color intensity.)

We then had to do a scanning process so we could get the color photograph into Cyan, Magenta, Yellow, and Black (CMYK) – the colors used for commercial printing. Once we had it in CMYK, the next task was to covert the offset-style halftones to ones that we could print with our multidrop ink jet mode.

The technique we employed is known as error diffusion. Instead of smoothly varying the dot size, it distributes the printer dots according to the amount of ink needed at a given location, while at the same time imposing a randomized, or “scatter,” pattern that makes it hard to see the individual dots.

The files were very large and had to be read from a tape drive, one block of data at a time. The tape drive broke in the process -- it was not designed for that type of data transfer. Once we made the conversions we printed the CMYK halftones, one at a time, using a single nozzle.

It was an incredibly painstaking process. We put the nozzle on an x-y plotter (a grit wheel-based technology invented in HP Labs) and printed out colors one at a time. We printed Yellow, then put it back on and printed the Cyan, then the Magenta, then the Black – and we printed out some of the first color pictures ever, right here at HP Labs.

Gradually we began to get into all sorts of color pictures. My life had come full circle. I had this background as a color separator and suddenly, here I was applying it to the world of inkjet. As time went on, we improved and improved, and that eventually led to our work in digital photography.

How did the work in the lab become incorporated into a product?

Some time after we printed those first color pictures in the lab, Dick Hackborn (an HP board member and veteran HP executive who led the company's move into printing) told the inkjet business guys to "go do color."

That was a big decision. The PaintJet (an early HP inkjet) had these huge dots, but we worked with it. It was 180 dpi -- you could count the dots -- but we did everything we could to make it work as well as possible.

The photos didn't look didn't look like much in the beginning, but often things that turn out good don't look like much in the beginning. That, to me, is one of the fundamental things that HP Labs has got to be about.

What role did HP and HP Labs play in the evolution of color printing?

HP drove the process. There’s hardly any company that matches us in mastery of the color processes we’ve developed for our printers.

HP Labs played a huge role. We did the early work on how to encode color for communication in computer systems. This led us into colorimetry -- measurement of color based upon how peple actually see colors. Ultimately, it is the combination of the human eye and brain that produces the sensation which we all call color.

By encoding color data in human visual variables obtained via colorimetric measurement, we moved from describing color in amounts of ink (CMYK) to a method that relates directly to our visual perception. With that as a foundation, we developed software called Architecture for Color Imaging. What it did was take RGB (red-green-blue) data, i.e. colorimetric data, and convert it -- through a series of algorithms and processes and look-up tables -- to data that would drive an inkjet printer.

Since we started from a basis of what we would see, we could then convert the data for display on a computer screen or output on a printer, and there would be a consistent appearance between the two.

The thing that was remarkable was that there was no expert in the process to do anything. It worked automatically. We took a lot of what I knew, which was how to build colors by hand, and put it together with a lot of color science and imaging science and created the intelligent printer driver from that software program -- and that’s been the basis for the drivers for all of our color printers since then.

Your group also has had a large role in the development of HP digital cameras.

We developed the fundamental algorithms that go into the cameras' image-processing pipeline, the most recent of which is the adaptive-lighting technology that's been incorporated into the new PhotoSmart 945.

(Adaptive Lighting simulates the ability of the human eye to adjust for such high-contrast scenes as a person standing in front of a window or children playing in the snow. It lets point- and-shoot photographers easily replicate what they see with their eyes regardless of lighting conditions.)

I used to deal with the problem of adaptive lighting when I was doing color separation many years ago. I would get transparencies with very high contrast, and I had to have an effective way of reducing the overall range so I could produce a color separation suitable for printing on offset. In those days, you could use a technology known as area masking. A photographer would call it dodging and burning.

In the late 1990s I started a project in Labs called the “high dynamic range project” to find a solution for this. In the course of this, I came into contact with the Retinex algorithm for adaptive lighting by Polaroid founder Edwin Land. With quite a bit of work by the Labs team, engineers in HP's Hardcopy Division – particularly senior color scientist Bob Sobol – we developed the adaptive-lighting technology.

So you take your lead from the way people actually see?

Yes, and then we use that information to take the very large brightness range you see and fit it into the smaller brightness range that can be reproduced on paper. At the same time, we bring into play all the experience professional photographers have gained over the years.

The camera data is encoded in terms of what you see, i.e. colorimetric data. Visually speaking, the dominant human visual sensation is lightness to darkness. The second thing you’re aware of is hue. The third is saturation of colorfulness – the intensity of a red sunset, for example.

What goes back to your brain are three signals. Red minus green, yellow minus blue and the other is achromatic (without hue). The last signal, the achromatic channel consumes 95% of the communication bandwidth or information carrying capacity of the optic nerve. That means most of what you see is black and white information.

We learned to derive colorimetric data from the electrical signals produced by the sensors in cameras and scanners and then to build algorithms that manipulate that information intelligently. On top of all this, we’ve built algorithms that look at image data and make intelligent decisions about it. These went into HP cameras, scanners, printers and PCs.

How did your background in traditional printing influence your work at HP Labs?

Basically, it gave me the vision of automating the process of color reproduction, taking the painstaking process of doing it manually and using algorithms to automate that. We've succeeded in doing that. All the things all those people used to do are buried in those algorithms.

How does it feel to have been a part of this radical change?

For me, it's been a fabulous journey. As a color separator in the old world, I used to dream of a system that could do all this.

It's enormously satisfying from another point of view as well, because the process of color reproduction has an artistic dimension. We're opening up the ability of individuals to express themselves artistically. We have made photography much more accessible to the consumer. It is quite easy to "play" with your pictures and modify them to suit your taste.

What’s next?

Our group is in the midst of major initiatives in commercial digital printing, where we are taking control of color to next level, the professional level. It's called color management. We are advancing the technology we acquired by purchasing Indigo N.V. to provide robust color printing solutions, from creation of printed material to production on the printing press.

At the same time, we're looking at advanced displays -- paper-like displays with minimal power needs. Eventually these types of displays will be an important part of the way people will see information, so we intend to be there.

We continue to have a strong effort in digital photography, which is a fiercely competitive business. We believe digital cameras are ultimately a lens, a filter and a computer, and we intend to build on our expertise in computational image processing to take advantage of that component.

We are exploring pervasive image capture, wearable cameras, advanced imaging materials and we recently completed some work in ultra low-cost paper handling and finishing.

We're also doing advanced user studies to understand how people in separate locations might use new capabilities to interact via images, or how extra dimensions such as audio can enhance the overall photographic experience. Putting it another way, understanding how digital imaging might improve people's lives. It's not just a matter of putting technology in people's hands. It's understanding how they will use it.