When MegaVision went into business in 1983, our business was image processing: we made the world’s first commercial Megapixel real-time image processor. At a time when most real-time image processing systems were oriented about the likes of NTSC, we introduced a unit that pushed the state of image-processing art fourfold.
Our early customers were nearly all involved in some form of imaging R&D. In the early 1980’s, the only way to practically process lots of high resolution (“High” at the time meant about ¼ Megapixel…. MegaVision raised the “high” bar to 1 Megapixel.) digital pictures was to use either a super computer or a specialized computer designed specifically for processing pictures, such as MegaVision’s first product, the 1024XM image processor. The image processor was a kind of array processor; a big, dumb, but very fast slave to a much slower, but more general purpose, computer host. The 1024XM’s 32 megabytes of real-time memory and 350Mbyte/second processing speed were remarkable for the time.
Our customers were quite varied both in geography and application. Some were scientists observing the glow of bioluminescent creatures of the sea, others figuring out how a warhead with vision could tell the difference between a mosque and a missile launcher, others looking at how electrons flow in silicon, others looking for the odd cancer cell in a field of a million normal ones, still others trying to make 3-D maps of neurons in the brain…..a different application for nearly every unit we sold.
But a couple applications shared a number of customers: retinal imaging and image sensor development.
A number of our customers were looking at eyes---mostly human, with a smattering of other primates and mammals thrown in. Most of these scientists were doing research projects involving images of the retina: looking at how the cytomegalovirus spreads on the retina of AIDS patients, automating the assessment of various pathological markers, and evaluating retinal response to various drugs and stimuli.
At the time, most images of the retina were captured on film: the very first digital systems to directly capture digital images of the retina were just being developed. MegaVision explored this unpaved road for several years: by 1990 it was clear that the resolution, quantum efficiency and dynamic range of available sensors were insufficient. We could not capture digital retinal images that could rival the quality of those captured on film. Furthermore, a one Megapixel digital retinal camera system was not only less capable than film, it was too expensive for most potential users.
The other arena of common pursuit among our customers was that of electronic image sensors. During the late 80’s, image sensor development was a beehive of activity. MegaVision’s image processor included a very flexible input subsystem, which made connecting oddball imaging devices rather easy. Since the emerging image sensors were all oddball to one degree or another, we found ourselves supplying image processing systems to the likes of Kodak, TI, and Intel who were exploring the fabrication of imaging sensors and needed a convenient way to capture and evaluate the images from their odd little silicon eyeballs.
We saw firsthand what was being developed in the silicon labs and what would in the not-distant future emerge to become the basis of the digital camera revolution.
At the same time, it was becoming clear that normal, inexpensive computers were becoming fast enough to do image processing, and our specialized, expensive image processing computers were becoming obsolete.
We thus began shifting our attention to digital cameras.
By the late 1980’s, we were actively designing and building what was to become the world’s first professional digital camera system (the Tessera system), designed from the ground up for capturing digital images in a commercial photo studio.
As our image processor market was waning, the first Tessera system went into regular use in early 1989 at a commercial photo studio in Minneapolis (Photo Mechanical Services, Inc.). Shooting 4 Megapixel images in a production photo studio in 1989, while the rest of the digital world was way south of one Megapixel, was an accomplishment that does not diminish with the passing of time. In an era where technological progress is exponential; where computers improve with the inexorable regularity brilliantly observed by Gordon Moore (Intel co-founder) way back in 1965, it was to be another 5 years before a new generation of digital camera would match this performance; it would be 15 years before such performance became commonplace.
A few interesting metrics:
If you combine these factors into an “overall improvement”, you get something on the order of 10,000. If you look at computers over the same period, price/performance has improved about the same: 10,000 times.
A curious, but important anomaly to this trend, however, is that the total number of pixels in the best commercially available digital camera only increased by a factor of about 10, having gone from 4 Megapixels to a little over 40, while a garden-variety computer’s ability to process these pictures has improved by a factor of well over 10,000, so it is truly a thousand times easier to deal with digital pictures today than it was 20 years ago. Thus it is clear why the image processor became obsolete.
Through the mid 80’s, MegaVision’s image processing business grew. As the image processor became obsolete and we began working on digital cameras, our business evolved. The Tessera digital camera system system was large, expensive, and way ahead of its time; while few were produced, it truly ground-breaking and was used intensively to photograph products in some of America’s busiest commercial photo studios for some of America’s largest retailers such as Safeway and Rite-Aid.
It was not until 1993 that a CCD imaging sensor was commercially available that could match the performance of the Tessera system but for much less money. By 1994, MegaVision had built its first CCD digital camera back, and the digital photography revolution got seriously underway.
The late 1990’s saw digital photography mature from dawn’s early light to the full morning sun of a digital day. By Y2K, as the sun was rising on digital, the sun was setting on film. Silver’s luster quickly tarnished under the ascending shine of silicon. For the picture business, silicon photography has been as disruptive a technology as can be envisioned.
From some tests that we performed as early as 1995, it was clear that an acre of silicon could take a better picture than an acre of silver. Inch for inch, silicon would beat silver. And silicon was getting better fast, while silver halide was stagnant, so it wouldn’t be long before silver would start to seriously fade. It also meant that small digital cameras would be able to take better pictures than small film cameras.
Which was not good for manufacturers of large cameras.
MegaVision’s expertise was in the high-end professional medium and large format marketplace, and we projected that this market would be vulnerable to 35mm digital cameras, because the quality of digital 35 mm cameras would rival, and then even exceed, the quality of medium format film cameras. So 35mm digital cameras would increase in quality. And because of the economies of scale, they would be much cheaper than medium and large format digital camera systems. We felt that the medium and large format market would not grow, and in fact, would contract.
To augment our professional photography market, we pursued other markets where we could apply our expertise to add value.
MegaVision had spent considerable effort extracting as much image quality as could be squeezed from available sensors, so we explored quality-critical applications in medical and technical imaging fields.
In 1998, Philips had introduced a truly remarkable sensor which raised the silicon bar a very big notch. With 6 million pixels and 10 bit dynamic range at ISO 400, it could capture retinal images with better than film quality. In 1999, we began development of a digital camera system for capturing retinal images of ophthalmology patients. This became the first digital camera back for retinal photography to equal and then exceed the quality of film captured retinal images. We supply these digital backs and supporting software libraries as the critical digital imaging component to manufacturers and integrators of retinal camera systems. These systems are sold to leading ophthalmic institutions worldwide---they are instrumental in the treatment of diabetes, macular degeneration, and other diseases that affect the retina.
Substantial increases in pixel counts opened another market that to digital imaging: aerial photography. A digital camera, integrated with sub-one-arc-second accurate inertial monitoring unit, GPS, and highly corrected optics could provide automatically geo-referenced data for geographical information systems in near real-time. This was another good fit for MegaVision’s expertise, and we began supplying digital imaging products to OEM’s in this market in 2002.
During the period from 2003 through 2007, the digital revolution in photography largely completed; digital photography became the norm of both consumer and professional photography. Kodak laid off tens of thousands. Film sales dropped precipitously. Polaroid declared bankruptcy. Some of the biggest names in professional medium-format photography went out of business. Film labs world-wide shuttered in startling numbers.
And most makers of digital backs for professional photography were consumed by the enfant terrible we so assiduously midwifed. MegaVision, meanwhile, grew growth and created new value for our customers.
We are now engaged in exciting new developments, including developing multi-spectral technology and providing imaging services which promise to bring new dimensions to imaging.