An overview of some
Professional Digital Cameras when evaluated for use on a light Microscope
originally published, Journal of Biological Photography 1998
updated Aug 2003
Introduction
This article is the outcome of the talk that was given at BioComm 97 in
Reno, Nevada. Writing about digital cameras seems almost impossible with the
continual release of new cameras almost weekly in conjunction with the dynamic
state of affairs the industry is in. Specifics on this topic are almost meaningless
and what is much more relevant are general trends and considerations for choosing
and using high end digital cameras. In the end analysis, your organization's
goals and services need to drive such an acquisition.
Background
The majority of my work is accomplished at the light microscope.
For this reason, many of the concerns I have, revolve around problems found
when working there. Some of the obvious concerns found in this application are
chip resolution, contrast, image brightness, as well as the chip's spectral
response to name a few. Given the difficulty in creating high quality silver
halide photomicrographs, problems inherent to direct digital cameras might seem
to only compound the challenge.
In the time since Reno, I would speculate that no less than 100 new articles on this topic have been written in the contemporary magazines. Additionally, I can share that in the time since that meeting, no less than 100 new cameras have been released and there are industry projections that no less than 70 new models will be released over the next year. Consequently, writing this piece has become like trying to catch an elusive dream. It's right there but always moving. This feeling is absolutely consisitent with the entire field of electronic imaging.
Overview & a brief history
When the abstract for the original talk was submitted, my issues were primarily
focused on Technical issues. My interests resided in how different cameras functioned
as well as how they performed under controlled circumstances at a light microscope.
As the work evolved, what became more relevant to me was the issues that confront
a person who might consider purchasing a camera for use in their work. What
camera would serve as the best choice and why.
I am quite lucky in that as a faculty member at RIT, many of my former students work in positions where I can borrow cameras for demonstration and/or evaluation. For this reason, I can evaluate the pro's and con's of many cameras in side by side comparisons. Not so long ago, when direct digital cameras were "still video" technology, their resolution was so poor that the immediacy of image capture had minimal value to the Biomedical Photographer. It is also important to recall that this was not so long ago. It has been said by many that electronic photography has done in 7 years what it took silver halide technology 150 years to achieve. As it pertains to purchase, this causes a dilemma because there is never the right time to purchase such a powerful tool. As each month brings something new and better and cheaper, one needs to use their best guess as to when to jump in. These cameras do not come cheap either with the least expensive "professional" camera starting out at no less than $6K.
One of the most important considerations for acquiring this technology is not the capability of the cameras anymore, but rather what you are trying to do with the camera. In fact, pictures made from Kodak 35mm Ektachrome 64T film and those created on quality direct digital cameras are very hard to differentiate. Last year at Biomm 96, I won an award in the Photomicrography category using a silverless process. The final print was a dye sublimation image produced from a high end scan back camera. It is an absolute a fact that these new cameras are very good and so the real motivation for purchase needs to be the type of product that is being produced. Simply the need to have a direct digital camera should not drive this type of a purchase. If this is the case, the Kodak DCS 210, a high end consumer camera is quite acceptable. In the end analysis, products need to drive the hardware needs. As consequence, one's entire computer operation needs to be integrated. As an example, which output device that will be available will influence the possible camera options. How the pictures will be incorporated into documents, at what resolution as well as how they will be stored all factor into the analysis.
Storage and Digital Files
One important factor in considering whether to acquire a digital camera
revolves around image storage and retrieval issues. The storage and accessing
of digital files continually is a concern for photographers. With 35mm 2 x 2
color slides, storage has always been achieved through the use of pages or storage
cases with built in illuminators. More often though , the slides left the photographic
department and went with the customer. Consequently few departments were overly
concerned with storage beyond areas that were located in clinical areas such
ophthalmology, pathology or dermatology. With the ability to create filmless
images, the how to, the where to, and the who should store the files seems to
be problematic and needs serious consideration.
Storage Media
It also is interesting to reflect for a minute on the evolution
of storage media. High end digital cameras generate large files sometimes 5-to
25mB if not more. Consequently an integral part of the system needs to be the
storage media the image will be archived on. Additionally the hard drive space
in the computer itself needs to be a factored. Computers not so long ago were
coming packaged with 40mB hard drives, while now a 2 -5 Gigibyte hard drive
is common. Similarly storage media has also come full circle. One of the very
popular original leaders in storage media was SyQuest® which was capable
of saving up to 44mB per disk. With uncompressed 25 mB files, only one image
could be archived on this type of disk. Compression of course is an important
step that should be used. Remember the Bernoulli Drives®, like so many products
from this very recent past, many have all but disappeared. Now Syquest sells
88 and 200mB drives. Many other products from the very recent past such as the
magneto optical(mo) systems that stored 128 mB find themselves coming out with
larger capacity storage abilities in the neighborhood of 650mB. An extremely
popular new storage media is the Zip Drive which will hold 100mB while the Jaz
drive will hold 1 gBs. The Zip drive which sells now for less than $100 is excellent
for this purpose because of its very low price. Writable CD's are another very
useful product for archiving and they will hold approximately 650mB. Storage
devices are an important consideration because the archiving of these images
takes up storage space. Producing 3 files that will be printed as 8 x 10's will
require 60 - 70mB. If this activity were done three times a day for 5 days a
week, the storage dilemma becomes quite obvious.
Before analyzing cameras, it need to be considered that the creation of the digital image can be achieved several ways. 35mm film can easily be scanned using any of the very good and moderately priced desktop scanners. In this fashion, quick, high res digital files typically in the 5 -6 mB range are produced for an investment of only $1,000. These files can be quickly be enhanced and utilized as necessary. Obviously using a scanner does not replace the need for the camera but can create a digital image. Also over the last few months many new intermediate quality cameras have been introduced that provide reasonable results. These type of applications might include but certainly not limited to head and shoulders portraits as well as quick press release work. The Kodak DCS 210 as well as the Fuji DS-300 are quite adequate for many of these applications. These cameras have found niche markets in the real estate business as well as in the insurance claims area. Their acceptance within the scientific community has been slow but I speculate that this is changing as I write.
The Professional Digital Cameras
The Digital Rev(s)olution is obviously
here! The look of the pictures, the quality of the image resolution, and the
capabilities of the computer to work on the images continues to mature exponentially.
As a result, the serious photographer that has need for direct digital cameras
will have countless choices and issues to weigh before making a decision. Cost,
features, speed, studio or mobility and many other subtle operational issues
will be part of the decision making. High end cameras can be categorized into
3 groups. The first group would include cameras that work like scanners and
need to be driven by a computer. The second group of cameras would be instantaneous
capture while the third group being multiple exposure instant capture. I have
found that each group has advantages and disadvantages based on the type of
work being produced.
Linear Array Cameras
Scanning cameras produce the highest resolution
of the three groups. They work similarly to any film or flatbed scanner. The
camera is composed of a single row of image sensors that form its CCD or charged
couple device. The sensor moves across the image area from one side of the frame
to the other. The CCD is moved by using a very fine stepping motor. There are
many cameras that work like this such as the Dicomed Studio Pro, Phase One or
the Leaf MicroLumina . Each of the backs is slightly different. Some fit into
a 4 x 5 camera back similarl to the way a cut film holder works while others
have a bayonet mount in the front that accepts a variety of lenses. The individual
differences can easily be learned by referring to any distributor of digital
products such as Logix or others. Since the camera is comprised of a single
row of pixels over a fixed length(width), the camera is capable of very high
resolution. The Leaf MicroLumina for example, has an array of 2700 pixels over
a travel distance that would comprise 3400 pixel points(length) which produces
an image of 9 million picture elements and a file of 26 mB. Additionally most
of these cameras produce images that are a minimum of 36 bit depth. Unfortunately,
Photoshop 4.0 still only acquires 24 bit images. As a result, some of the camera's
high detail is lost even before it has ever been used.
Scanning cameras are typically slow in capturing images. As an example, when at a light microscope that would require a 1/15 second using Kodak 64T film, using a scanning type camera, the exposure might take 8 -10 minutes. Because the sensor is moving, this type of camera cannot record moving subjects. It also needs to be tethered to a computer. Computers with large hard drive space are required as well as a minimum of 32mB of Ram is also needed. One recent studio camera requires a machine with a minimum of 256mB of RAM. These cameras are low in sensitivity(response) and require high brightness to record them. The use of flash is also not possible because of duration while some tungsten filaments will experience flicker during the course of a long exposure. As a result, studios utilize HMI lights for this type camera because there is no flicker to create different exposures(brightnessess) as the sensor moves across the frame. The camera is solely a light sensor with the camera controls in the computer as a software interface. Often camera access can be found as a PhotoShop plug-in. Almost universally, these cameras have a pre-scan for initial set of the sensor based on the subject's requirements as well as the lighting. Image contrast can be adjusted, scan speed adjustments can be made as well as many other features such as color response are available. Some of the cameras allow fine focusing to be executed during the prescan by having NTSC out that is concurrent with the still digital signal. The image sensors also responds best to flat light. Ratios that would be similar for transparency film or less seem to be ideal for CCD's.
At the microscope, this type of camera provided exceptional results. Because these cameras do not use mirrors that would be associated with SLR type cameras, the image is less susceptible to vibration. Determining critical exposure was very important . With too much exposure, the sensor caused blooming or exposure was created in adjacent picture areas. This was evidenced as a color glow(exposure streaks) into adjacent shadow areas. As a result, making pre-scans that created a brightness value below 230 in the clearfield helped resolve this so long as the sample had adequate contrast. Flat subjects are really problematic regardless of capture media. With the scan speed determined, a white point could be set and the image corrected by the software. Another interesting discovery was that if the exposures were too short, the camera sensor often moved across the image field so quickly that vibration was introduced. Vibration in this type of camera is observed as bands or density differences across the width of the image. In the clearfield areas of a light microscope, this banding is very evident as density changes. At very low magnifications, where achieving Kohler is difficult regardless of image capture methods, the cameras produced obvious localized exposure differences as a result of the lamp filament. It was recorded as different pixel values as a result of the sensors being very responsive to localized brightness differences. The pixels really enhance edges naturally so through the use of 2 diffusion filters , the lack of uniformity in the background was removed.
Accurate color seemed initially challenging but very accessible. The sensors from different cameras, like slide films, have different color responses. It was interesting that the most accurate color was often achieved using lamps at approximately 3400 K at maximum light intensity. Through the use of neutral density filters to remove significant light, it could be speculated that the spectral responses of chip might be optimized at this color temperature. As it relates to brightness, it was also experienced that if a typical film exposure would be longer than 1/4 second with 100 speed film, these cameras did not operate.
Area Array cameras
The majority of cameras on the market seem to fall
into two large areas, scanning and area array. Probably the largest group of
cameras on the market are the area array type. Area array cameras see the entire
image simultaneously as do film cameras. Consequently all the camera controls
are the same on the digital camera that would be found on a film camera. These
types of cameras are sold by many camera manufacturers such as Kodak, Nikon,
Fuji, Canon and Minolta just to name a few that should be familiar names. So
from the front, all the digital cameras look and operate the same as any film
camera such as the Nikon N-90. Spot versus center weighted metering is but one
of the many functions available as with any fim camera. Behind the lens is where
the huge changes occur. Rather than a pressure plate for film, a Charged Couple
Device is located there. As a detector, it has a vertical and horizontal pixel
array as compared to film format. The physical dimensions of the chip determine
its resolution. A small chip might be 640 x 460 pixels while the highest resolution
chip on the market boasts a image resolution of 2K x 3K. Many intermediate resolution
cameras have a chip approximately of 1K x 1.5K. Chips also have a sensitivity.
Most cameras seem to exhibit sensitivities (ISO potential) of 80 at the low
end to 1600 as the current most responsive. The sensitivity for most cameras
can be changed or is considered variable based on lighting requirements. As
the ISO is raised less quality is noticeable however even though proper exposure
is achieved. The loss in quality is described as noise. Typically the higher
the signal to noise ratio, the poorer the result. Conversely, when the ISO is
lowered, the results are superior.
Because chips have a physical dimension, they often see less of the field than film cameras. As a result, required focal lengths are often not identical as with film recording. To be able to capture what a 50mm lens sees on the Nikon N-90S with film, the Kodak DCS 420mm uses a 20mm lens. As a very general rule of thumb, the chips require 1/2 the focal length of the normal lens when the camera where to be used with 35mm film. On a microscope, this often required using lower power objectives than one would first select. A field that can easily be observed with a 10X objective would now require a 4X for imaging if the requirements to record the same field of view were necessary.
The CCD as a device can only respond to differences in brightness not color. When using film, one can choose from many emulsions based on project requirements, however the CCD's must be modified to respond to color. As a consequence, color separation strategies are often required to create color vision. This is also true for scanning type detectors. One solution that is used is to place a Red, Green and Blue filter over the sensor in a linear fashion. A long red filter, a long green filter and a long blue filter sometimes 7µm in width, are run the length of the sensor. In this fashion, the surface area is divided into three equal spaces that each render a different spectrum. The computer software can then reprocess the signal and create a credible color display.
Specifically within the area array cameras, the solution to the color vision problems varies. The Kodak cameras for example have colored filters on each row of pixels. So across the row of 1500 pixels, there are 3 filter stripes for each row. Through the use of software interpolation, the image is written and color created from image processing. These cameras allow photographs to made anywhere at anytime using a variety of light sources as the cameras have shutter speeds and reasonable responses expected down to 1 second. I have found that below these times, the cameras often fail to respond accurately because the chip warms and fails. Sometimes the exposure is actually terminated. Another method to acquire color is through the use of mosaic techniques, where every four pixels in a square is outfitted with a different filters. Often the filters are R, G, B and with the fourth being either magenta or cyan or yellow.
Because the color is the result of software interpolation, there are sometimes digital artifacts produced. These color artifacts are often referred to as aliasing and are frequently observed in linear subjects that have high frequency patterns that span across several pixels. Aliasing is most evidenced in the medium resolution stripeing filter type cameras and is easily correctable in PhotoShop. The highest resolution camera on the market, the Kodak DCS 460 which uses this stripe technology does not display this problem. Using various options in PhotoShop 4.0, specifically in Image and Color controls, one can use commnads to change the radius and threshold of the pixel display all but eliminating the digital noise.
Another method that is used to create color is accomplished through the use of 3 CCD's. The image produced by the lens is broken apart by a prism and relayed through Red, Green and Blue filters to its own sensor. One camera that operates in this fashion is the Minolta RD-175. Each sensor is smaller than the chip as a whole, but the achieved results are quite acceptable. As a basic response, the chip of the Minolta camera has an ISO of 800 which was very important in when working in low light situations such as with darkfield, fluorescence and ophthalmic work.
Most CCD's are inherently sensitive to Infrared radiation. As a consequence, the use of a "hot filter" or IR cut off filter is desirable. The use of this filter is not a requirement, but certainly when using tungsten lights has great value. The high response of the chip to IR energy often causes an unusual color response. In the older Kodak publications, where Kodak IR film was discussed, a term referred to as false color was used. Through the use of CCD's, false color is also possible without the filter. Tiffen sells such a filter and it is referred to as a "hot filter".
The Fujix HC 2000 is a very interesting area array camera but is very different from other area array cameras. The Fujix 2000 is a studio area array camera with a NTSC video signal out. This feature allows for precise composition, focusing as well as exposure determination directly from another monitor which is required in addition to a computer work station. This camera created images with file sizes of around 5mB and was quite interesting for many reasons. It was excellent for use at the microscope as well as on a gross stand or portraits. The camera is tethered to a computer so it is not portable in the least. The camera required an ENG mount for adaptation onto a scope while any small video lens mount worked for studio applications. The ENG mount needed to par focalized to the eyepieces but this was not difficult to accomplish.
PCMCIA Cards
The images that are captured in most area array cameras
are written to small removable hard drives sometimes referred to as film cards
in the amateur market. These removable hard drives are also referred to as PCMCIA
cards in professional circles and are available in variable storage capacities.
Cards are classified by their physical size either Type 1, 2 or 3. Often the
Type 3 card is the largest and has the largest storage capacity. Their storage
capability has been continually growing since the introduction of these products.
The Kodak DCS 460 often has been sold with at least a 230mB card. Referencing
any supplier of such cameras, this type of information is readily available
. It has been suggested as a rule of thumb that these cards sell for $100 per
5mB of storage.
Images are captured and then written to these miniature hard drives in this intermediate location for storage and retrieval. Once the camera's shutter has been activated, the hard drive of the camera also is activated. A term has evolved to describe this process as waking up the hard drive. Once activated, the camera can record the image and then write it to the drive. In most cameras, there is an LED or some other type of an indicator to reference that drive is writing. Some cameras are very responsive as to how long it takes to activate a system while others are quite slow. For this reason, shooting events that have the potential for pictures to be made at a moments notice such as in photojournalism, require the photographer to continually keep the camera awake(ready) so as not to miss any shots. Additionally, each camera has a different capability to hold several intermediates images before then writing to the drive. Each camera as a consequence will have a different burst rate or ability to record more than one image at anytime. The lower end cameras can only perform one activity at a time. Activate, capture, then write while the higher end cameras designed specifically for journalistic work, might be able to record up to 5 images before needing to write. Cards are removable from the camera and should never be removed while they are writing. The removal of a card during writing will permanently damage the card.
With the shooting completed, images from area array cameras will need to be moved from the camera/card to your workstation. This activity can be accomplished in several fashions. The camera itself can be tethered to the computer with a SCSI cable or the card can be ejected from the camera and the file accessed through the use of a card reader. A card reader is nothing more than a very small drive designed specifically for PCMCIA cards. These products vary in price and drives can be addressed through the PhotoShop Plug-in for that particular camera brand. Historically, each camera had its own file format which has been problematic. Formats such as DCS, HD, LDR MDR and others are a few of the many proprietary formats than have been problematic for users thus far. Recently though, manufacturers have been releasing cameras that save and export JPEG or TIFF image file formats which has greatly simplified working at various locations in the digital image chain
Conclusion
There are obviously countless cameras on the market, with each having their
own distinct advantages and disadvantages based on applications. I have also
worked with, but not extensively, with many cameras . It would be next to impossible
to use all the cameras that are available as it would be for film cameras. This
piece has been an attempt to lay out the many facets and details of selecting
and using direct digital cameras.
As with as all new technologies, there will always be waves of acceptance and our field is clearly in a phase where the majority of photographers are, or will be "adopting"very soon . I have laid out many aspects of digital cameras for consideration as the move towards adoption is made. Digital cameras will continue to be produced more cheaply and deliver higher resolution for less money. As the differences in computer platforms continue to diminish, so do the large differences between the cameras for now. The digital camera market will never be totally stabile for long as new technology such as active pixel cmos begins to change the way chips are manufactured and integrated into cameras. The comment made some years ago by the JBC editor Joe Ogrodnick at a RIT symposium still holds much truth, "the only thing for certain that we can say about the future is that there will be change". The advantages for the use of digital cameras in certain applications is becoming quite evident. Certainly it is easy to project many without great challenge. Prior to purchase, adequate research is essential to evaluate the many options available. Additionally long term goals should be considered in setting a direction away from film with an integrated approach. Buying a digital camera will cause consumable expenses to go down, but the initial investment to get there will be high. Also the life expectancy of such an investment is a factor. Remember when the purchase of a Nikon F -3 would last for 10 years? Well the acquisition of a digital camera and computer system will never achieve that life expectancy. Welcome to the time of rapid change.