White Paper on Continuous Light Technologies for Digital Still and Video Work compiled Fall, 1995

by Gary Regester [GR], et al with contributors: Michael Collette [MC], BETTER LIGHT, (developer of Dicomed Digital Camera), Ron Tussy [RT], former Sales Manager, PHASEONE Camera.

Contents

  1. Preface
  2. Why Digital?
  3. "Apples, Oranges and Noise"
  4. CCD Concerns
    a. CCD vs Human Response
    b. CCD Requirements
    c. Exposure
    d. Noise
    e. EMI
  5. Lighting and Light Sources
    a. Filtration Corrections
    b. Color Rendition Index (CRI)
    c. Ballasts and line frequency
    d. Daylight
    e. HMI/MSR
    f. Tungsten
    g. Fluorescent
  6. Appendix:
    Definitions
    Manufacturers
  7. Bonus recipe: Build Yourself a Digital Light under $300 usd

1. Preface by Gary Regester

Just like every other frontier, the Great Digital Landscape is already filled with its cast of charlatans, carpetbaggers and snake oil peddlers preying on the fear and trembling among the new "cyberpioneers". This is especially true with the question of what light technologies remain appropriate and what new concerns arise. Well, here is a "dag gum honest" bit of a "hodge podge" trying to get at some answers about what will and will not work with the latest linear array cameras/ scan backs and video. It will also have application to area array digital cameras and the coming digital cine. This discussion about digital lighting choices will include HMI, tungsten, fluorescent and flash including some other new arcania such as spectral noise, wave frequency, and RF interference will be included.

There are probably three immediate possibilities available to the still photographer considering the move to digital work. Use film and/or instant material and a scanning bed. Most lithographers seem to recommend this simply due to the existence of the "match print." This is a good point, but outside this discussion. Next, area arrays of CCD such as Kodak DCS cameras allow exposure by flash which minimize any rethinking and "re-tooling" regarding lighting. Finally, the phase or linear array cameras which require continuous lighting. Phase [linear or scanning] array cameras are analogous with video and digital cine in their requirements for the particular kind of continuous light. This is the primary reason for this discussion and is new ground that most still photographs and many videographers have not yet crossed. The choice by the still photographer between area vs phase [or linear] arrays will probably remain a trade off between cost and need, obviously phase arrays [scanning] will give a better resolution for less money than the area array which can use conventional lighting from a flash tube because fewer CCDs are used to cover a larger image size.

About that "snake oil". There is a suspicion that the rather obvious question of what lighting requirements would be needed to satisfy and SELL their new digital cameras caught the CEOs and Sales Managers a little off guard. Digital still imaging, which itself is a rather obvious extension of video technologies, may have occurred too fast for such management types to have asked the lighting question. It was certainly known to and troubled their development engineers. Given the usual "just sell it!" mentality, it appears that the management reflex was to peddle the existing "cine" lighting solutions of tungsten and the older and very expensive HMI [metal halide arc]. After all, if you can sell a $30,000 camera, who is worried about a couple of $3000 lights. But does what works with film, work with the CCD? There was sufficient knowledge of known problems, specifically the high sensitivity of the CCD to heat [IR] that it is not unreasonable to question the motives of such immediate sales of tungsten and HMI. But then, how much money can be made by suggesting such humble, ugly and unprofessional light sources as cool fluorescent which someone could fabricate easily from the local hardware store and nursery [for plants!].

And having revealed my own prejudice, I have tried to keep this discussion on the side of the needs of the camera's developer rather than the lighting designer or photographer. My belief is the creator of the new animal has the best knowledge of how to care and feed that animal much better than the user or the corporate peddler. Specifically, you see the most comments coming from Michael Collette, developer of Dicomed's linear array camera.

[Also keep in mind that very little information is ever discussed which does not have at least some profit motive. How many workshops have you attended entitled "Lighting with Bed Sheets"? Yet, in truth, bed sheets are probably the most sophisticated means to diffuse lighting, but no one happens to make any significant amounts of money selling bed sheets to photographers and videographers.]

Videographers hopefully will forgive the bias of this paper toward high resolution use of CCD chips. Everything that can be now be learned about the particularities of the CCD in high resolution apply immediately to low-res television and, very soon, to digital cine. In a sense, the problems of digital still is a advance look at problems of video in the near future.

The following discussion was largely developed between the participants through the Kodak site on American On Line. This discussion is a beginning and hopefully raises as many questions as it answers.


2. Why Digital? Several Notes [MC] Shooting digitally definitely saves the client more than the price of a hi-res scan. The overall dollars-per-image may even be the same, but the client will see major savings in TIME, and improved CONTROL over the final image. Client execs identify with the former, and the creative types will appreciate the latter. As the photographer, you can no longer charge for film, but you are now doing the hi-res scan directly, having deftly adjusted both the lighting and the scanner controls to produce exactly the image you and the client had discussed. The client should expect to pay for this additional service, at least after you get to know what you're doing with the digital camera.

[RT] Being a photographer for 12 years and now in marketing and sales of digital products, it upsets me to know that this digital revolution is not being led by photographers, who are instead being dragged into it, not by their will. Wake up out there. Photogs should be the leaders here. We should create an organization and set standards within the industry to adopt to our existing industry order, on color, resolution, bit depth, terminology and direct this into our own industry. We should have representation within the Color Management Consortium. It is our medium and our industry that is being sold into and dictated to. We could be the organization that endorses a new product with a stamp of approval. This would force new technology and development to meet existing industry criteria, not desktop designers, not lighting manufacturers, and certainly not the pre-press industry. Those are my thoughts. End of Speech.


3. "Apples, Oranges and Noise" [GR] The difficulty of comparing the available light sources considered for use with linear array [scanning] digital imaging and video [HMI and fluorescence, halogen and daylight] is due to two factors. First, variations in respective spectral output trichromatic fluorescent sources have a discontinuous peaked RGB spectrum when compared with the continuous spectral characteristics halogen. HMI is a jagged approximation of a natural daylight curve with some unexpected infrared at 850 and 900 nm. Secondly, variations with CCD hardware and controlling software require their own specific peaked RGB sensitivity and is therefore color blind to or confused by [rf: "crossover"] certain parts of any continuous spectral output [daylight and halogen.]

Consider using lumen output specifications by lamp manufacturers to simply to compare light output. Such specs place a 200 Watt HMI/MSR lamp at 15,000 lumens, 200 watts of biaxial triphosporous fluorescent produces 14,000 lumen and a 750 Watt halogen lamp is about 15,000 lumens. However any comparison between the light output of hmi/ halogen/ fluorescent measured with an lumen [lux or exposure] meter would seem impossible as the meter "sees" only the total accumulated light energy of discontinuous spectrum of the fluorescent and HMI which is then compared with the full spectrum of the halogen "black body" and the somewhat bumpy daylight which, in turn is, to various degrees, unused by the color blind CCDs. It would seem that the only comparison possible is though tests with specific digital and video cameras. And one would assume that each camera would result in a different comparison due to differences in CCDs and control parameters.

(We can use the more familiar problem of CRI ratings as an example of the problem. Most photographers know that all fluorescent lights are the not the same on film and already know, at least experientially, about CRI. The Color Rendition Index for fluorescent lamps is based on the average color mismatch of 8 object colours [with 100 CRI being a perfect match]. "Average" is the operative word here. A lamp which has a single greenish phosphor coating might have a CRI average of 49, while the more expensive triple phosphor coating approach an average of 98+. The interesting point is that the exposure meter measures little difference between the light output of a single phosphor lamp or a lamp with tri-phosphors [assuming equal wattage and Kelvin temperature rating]. However, there would a very large visual difference on film.)

There is a third and fourth consideration first the apparently weak blue response of most CCDs, Kelvin balance of less than 5000 degrees are of no real interest to digital applications. In fact, CCD image capture would be much easier if the lamp manufacturers made some interesting lamp configurations in 7000 or 10,000 Kelvin (the color of open shade.

Fourthly, continuous spectral output can create "confusion" or "crossover", if the "red" chip senses the "green" output; the "green" might see both "red" and "blue" information, etc. Here there is an operation advantage to creating a light source with discreet but discontinuous spectral output that occurs only within the sensitivity range of specific CCD response. Again, a task for lamp manufacturers.

A final important question for discussion related to the comparison between the fluorescence, HMI and halogen is whether any digital "noise" is created on the CCDs or associated electronics [computers, etc] by non visual parts of the spectral output of these light sources specifically, IR from halogen, UV or IR from HMI, or any Radio Frequency or electromagnetic line interference from the HMI or fluorescent lamps or their ballasts cycling at 50 Hz to 40 kHz. Our concern would be with such "noise" as might degrade [or enhance] the final image. One should also question any "noise" [RF or EMI] from the camera itself and its influence on associated equipment.


4. CCD Concerns

CCD vs Human Response [mc] Color CCDs are usually no more "color-blind" than humans, although they certainly may respond differently than we do. Graphs of the color response of the human eye, obtained through various test methods, show a "hole" in our response to luminance between blue and green, at around 480 nm. Also, human vision does not have neat, symmetrical Gaussian response(s) to three equally-spaced wavelengths, but instead appears to have a rather narrow response to blue (with a peak at 440 nm.) and broader, mostly overlapping responses to green (peak at 540 nm.) and red (peak at 570 nm.). Note particularly that the red visual "peak" is actually well within what we often think of as green (or yellow, actually) it is the DIFFERENCE between the green and red visual responses that helps us discern red colors (this is also true of the other color pairs, to a lesser degree). Never underestimate the amount of (brain) processing we apply to our color vision (and to all vision, actually the eye is a fairly simple "camera" all by itself). I recall seeing graphs of human color response, including the effects of the brain's processing, that have effective "peaks" at 440 nm., 560 nm., and 620 nm., for whatever that's worth. Of course, since the eye-brain is the ultimate judge of color, that's the way it SHOULD be...

CCD Requirements [MC] How does CCD exposure work?; similar to film, each pixel's electrical signal is the product of light intensity and time of exposure. Light intensity is controlled by the subject illumination and lens aperture; exposure time is controlled by the CCD electronics. With a (tri-)linear CCD, thousands of exposures are made across the frame; a line (trio) may be read out of the chip while the next line (trio) is being exposed. Most CCDs are designed for very fast pixel readout, but other considerations such as analog-to-digital conversion time, pixel processing time, and data transfer time often determine the overall maximum system data rate. Regardless of the actual CCD exposure time per line, the scan can only proceed as fast as the data can be handled.

[MC] Phase One literature states a maximum data rate of 11 MB per minute, yielding a minimum full-res (100 MB RGB) scan time of about nine minutes. A full-res (84 MB RGB) scan on the Rollei takes about twelve minutes, producing an effective maximum data rate of 7 MB per minute (I don't think this includes the conversion time to CMYK). Given enough light, the Dicomed insert can sustain a data rate of 52 MB per minute, allowing it to capture a 129 MB RGB file in just two and a half minutes. As the line time is made longer, the CCD exposure time necessitates slowing down the scan and data rates, to as long as 15 minutes overall (still an effective data rate of 8.6 MB per minute). Retrieval time presently adds a couple of minutes to this, but may not always be necessary in the future (how much RAM do you have?). Scans at lower resolutions take proportionally less time with the Dicomed unit; a 32 MB RGB image can be scanned in as little as 75 seconds (or as long as 7.5 minutes). The shorter times are fairly common outdoors, even at f22; in the studio, nearly everybody operates near the longer end of the exposure scale, taking any "extra" light as additional depth of field. For reference, I have seen top-quality tabletop illustrative photography scanned with the Dicomed using a single 2000 Watt tungsten Fresnel source at f16.5. Using one of North Light's luscious 5 kW tungsten softlights allows tabletop operation to f32. HMI or fluorescent sources will use less wattage to produce similar results; alternatively, use more light and work even faster.

[MC] Most CCDs are less sensitive to blue light, which makes the use of higher color temperature sources desirable. At around 5500K (true daylight), the Kodak CCD in the Dicomed Digital Camera has nearly equal response to all three colors when used with the "daylight" version of the infrared filter available for this device. Since the Dicomed unit has a very straightforward way of adjusting its color balance, the relative sensitivity of this camera to various light sources is easily quantifiable by examining the three color balance numbers after neutralizing a gray scale. This ease of comparison will be clouded somewhat if a different infrared filter is used HMI and tungsten should use a thicker filter, which blocks more infrared, but also changes the response of the red channel significantly. Nonetheless, the "usable sensitivity" of the camera with each light source can still be determined, which is what the photographer wants, anyway. Exposure Measurement [MC] Most light meters, including those used for photography, have a built-in filter that approximates the low-light level response of the human eye, with a strong green peak and much lower blue and red sensitivity. While such meters are generally useful for most photographic lighting and subjects, they will not always provide an accurate indication of the total energy available, especially with discontinuous spectrum sources like fluorescent and HMI lamps. For a given source, the meter (or the photographer) can quickly be re-calibrated to provide accurate exposure information, but source-to-source differences cannot be made as accurately, as you point out.

[MC] Equally important is the "usable color rendition" provided by each source, which may not be apparent from its effective color temperature, CRI, or usable sensitivity. As you surmise, this will also be a function of the camera used to make the image each camera vendor's CCD, color filters, and infrared filter combination will have a somewhat different complex response. Image processing by the camera or its software may also affect the color response of the system, by attempting to make up for known deficiencies. The Dicomed camera does not presently perform any color-specific data processing; color balance is determined by adjusting physical parameters of the CCD's operation.

NOISE [MC] Some of the sources of "noise" in digital camera systems include:

  1. shot noise from dark current in the photosensitive elements themselves
  2. errors in the CCD charge-transport mechanism
  3. readout noise from the charge-sensing amplifier on the CCD
  4. noise in the signal processing circuitry
  5. noise and quantization errors in the analog-to-digital converter(s)
  6. noise induced from external sources operating in proximity

...and that's before we even open the shutter!! The signal generated by incident light has its own shot noise component, and is also affected by all the sources above.

[MC] Regarding digital "noise" from non-visible energy striking the CCD I am not aware of any physical cause for such a "noise" being created by the CCD, although the presence of non-visible "light" will certainly affect the color balance, color rendition, and even the dynamic range of the light source and camera combination. For example, the Dicomed camera nearly always exhibits less dynamic range when taking an infrared photograph of a given scene (without the infrared filter), versus the same scene imaged in color (with the infrared filter); this may be due to many objects' inability to absorb as much infrared as visible light, creating brighter shadow regions. Also, the overall spectral balance of different light sources will require compensation to produce a neutral color balance; whether this is done physically (as in the Dicomed camera), or by manipulation of the CCD data (as in most one-shot cameras), the net result is usually more noise in the energy-deficient channel(s). With 3200K tungsten light and the recommended infrared filter, the Dicomed camera still requires around a 2-to-1 ratio of blue sensitivity to either red or green, inevitably causing slightly more noise in the blue channel than that observed with true daylight.

[MC] It depends on your definition of noise... If we are talking about the random "snow" sometimes visible in the dark areas of digital images, then I am not aware of any mechanism that would create (or increase) this kind of noise simply due to non-visible light energy striking the CCD. One source of this "dark noise" is the shot noise due to the CCD's dark current (an unwanted "leakage" current that creates a signal even in the absence of light). The dark current increases with increasing CCD temperature; if the digital camera is being used in a small room with large hot lights, an increase in the room's ambient temperature will cause a small increase in this noise, especially at longer line times.

[MC] If we consider "noise" to be ANY unwanted signal, then INFRARED light is certainly a noise of the worst kind (CCDs are rather insensitive to UV light, although this can cause problems with certain subjects fluorescing, usually bluish). CCDs are usually more sensitive to infrared light ("near-infrared," from 700 to 1000 nm.) than to visible light (especially blue); most of the on-chip color filters (and even those in filter wheels) become transparent in this region, allowing the infrared to pass. Without suitable additional blocking, infrared energy will affect all color responses by adding large amounts of signal in common to all color channels, color saturation (the signal differences between color channels) is greatly reduced, resulting in washed out colors, sometimes with unexpected hue changes (if there's any hue left). The filter(s) available for the Dicomed are designed to remove nearly all of the infrared energy passing through the lens, resulting in purer, more saturated colors. Two different filters are available: one for tungsten lighting, and one for daylight (natural or fluorescent). HMI lights, although not nearly as "hot" as tungsten, have a significant peak in the infrared that is best handled by the tungsten IR filter, even though these lights are "daylight balanced."

By leaving out the infrared filter, the Dicomed insert becomes a very sensitive infrared imager (remember to adjust focus slightly for best results). With a couple of modest hot lights (lots of infrared), monochrome or verrry pastel images can be made at f32 and beyond, using the shorter line times as well. Just a thought...

[MC] Here's a thought regarding possible additional noise in the shadows with HMI lighting these lights have a very strong peak in the infrared (around 810 nm.) that can be five or more times the average output in the visible region. If this peak fluctuates more (is noisier) than the visible output, this fluctuation will show up in the signal channel(s) if the infrared filter does not suppress these wavelengths enough (any signal from a "blue" pixel is reproduced as "blue," regardless of its original wavelength). Another possible cause of noise when using HMI lights would be #6 above; HMIs and other arc lamps are rather noisy sources, both in light output and especially EMI (electrical interference). (Spark gaps were the first radio transmitters...) Any such source (or ballast) operating near a digital camera may couple some of its noise into the CCD electronics, creating a noise that would probably be more apparent in the shadows (lower numbers). This sort of noise should decrease rapidly with increasing distance between the noisy source and the camera electronics.

In my somewhat limited testing of the Dicomed insert with HMI lighting, I have not noticed any additional (shadow) noise produced by these lights when used within about four feet of the camera. The presence of the infrared peak was evident in certain dyed colors, which reproduced incorrectly (excessive red, in most cases).

EMI Certainly, electrical interference from nearby appliances (including lights) can also create noise in digital cameras. The CCD is very sensitive to charge variations (actually the basis for its internal operation), and the wiring between the CCD and the remaining analog circuitry (up to the analog-to-digital converter) is also susceptible to both magnetic and charge variations. Since the CCD must be able to accept light, it is impossible to totally shield the device from its surroundings. Proper design, with critical components mounted close to the CCD and as much shielding as possible, will minimize the susceptibility of the camera to outside interference (except from art directors). As we discussed, this sort of electrical interference also obeys the inverse-square law, diminishing greatly with increasing distance. Keeping as much separation as possible between the lights, ballasts, and connecting cables, and the digital camera "head" and connecting cables, will minimize this problem. In general, I would expect HMI lights to be the greatest producer of interference, followed by fluorescent lights (especially those with separate ballasts). Tungsten lights should produce very little electrical noise, as they are essentially resistors across a low-frequency sinusoidal source voltage.

 

5. Lighting concerns and Light Sources

FILTRATION Corrections [MC] [regarding changing or smoothing the spectral characteristics of the original light source] the reflective material...and the diffusion plate...filter out the spikes and add the missing portions of the spectrum << Filter out, certainly, but the only way I know to add (create) any missing portions of the spectrum are fluorescence (down-converting a shorter wavelength into a longer one) or resonance (up-converting a longer wavelength into a shorter one). Fluorescence is quite common the glow discharge inside fluorescent tubes is predominantly in the UV; the coatings on the glass convert this energy into visible light. Resonance is much less common (and much less efficient); an example would be a frequency-doubled laser, using lithium niobiate (or something similar) to produce blue or green light from an infrared laser. Neither of these are utilized in most of the reflective/diffusive materials used for light modification. What you see with the light box example is plain old filtering the glass removes some of the existing red and blue light, causing the green color cast.

[MC][regarding use of IR filters]I have used the Dicomed back with both the Kaiser high-frequency fluorescent light box and copy lights with superb results. Note that an appropriate infrared filter is ALWAYS needed for good color rendition, even with daylight-balanced fluorescent sources that have very little infrared. Dicomed provides a "daylight" version of the infrared filter for fluorescents (and true daylight) that produces a more efficient color balance with these sources. (Even though they are daylight-balanced, HMI lights require the "tungsten" version of the filter, due to a strong infrared peak that must be attenuated.)

CRI The Color Rendition Index (CRI) is a measure of how closely a discontinuous spectrum source approximates a true "black body radiator" at the claimed color temperature. Since a glowing tungsten filament is essentially a "black body" (to a physicist sure looks bright to me), tungsten lamps have a CRI of 100. In other words, a 3200K tungsten source will reproduce (all) colors with 100% accuracy compared to a "black body radiator" at 3200K. In that sense, you are quite correct in stating that CRI ratings between different color temperature sources are less accurately compared. However, even at the same color temperature, two different sources with CRIs of 90 may reproduce color somewhat differently, because of the "average" color mismatch rating you mentioned. CRI ratings do not take into account the non-visible response of most CCD cameras, either, but are again tuned to human visual response. Perhaps the only truth to CRI is that higher is better, with at least 90 required for accurate color reproduction.

Ballasts Regarding (HMI) ballasts, I presume that magnetic (line-frequency) ballasts are used because of their low cost and simplicity (reliability), although the newer electronic (high-frequency) ballasts may soon reach parity on these points, and should offer increased efficiency, too. Being an arc lamp, HMIs will generate a considerable amount of noise, both audible and electronic, regardless of the ballast used. Like fluorescent lamps, HMIs have no "lag" and will readily show the frequency (and waveform) of their excitation source (the ballast). This makes high-frequency, square-wave ballasts highly preferable for scanning digital photography, where adjacent scan lines may show variations in light output as faint striping. This is less of an issue in video and film (movies), where adjacent frames flick by rapidly. Even with a regulated high-frequency ballast, HMIs may exhibit small fluctuations in light output due to the arc "wandering" on the surfaces of the electrodes (another generic property of arc lamps). I tend to agree with Jeff that HMIs produce about 3x the usable light per watt over tungsten (between 1 and 2 f-stops more light).

Daylight [GR] If we all still had our northlight studios and head clamps all would be perfect. Interesting full circle of course, in a portable suitcase! Daylight seems to the the unspoken standard continuous light source to which CCD manufacturers design their products. And the standard by which all other sources are compared. A few millienia of life under the sun creates some basic presuppositions.

HMI/MSR [MC] First of all, HMI lamps do not have a continuous spectrum they are a special form of metal halide arc lamp, with many individual spectral "lines" combined to form a quasi-continuous spectrum of energy. Osram, the maker of HMI lamps (HMI is a registered trademark of Osram), shows this discontinuous spectrum in its product brochure, and claims a CRI of "more than 90" for these lamps.

[MC] HMI lamps do have a strong output in the UV, which is typically attenuated quite effectively by the common glass "shells" placed around the bulbs in photographic usage (a borosilicate Fresnel lens will also cut down this UV output significantly). Even so, some additional fluorescence of certain subject materials may be evident from the residual UV.

[MC] More importantly, these lamps have a VERY strong peak in the near-infrared, around 800 nm. This wavelength is conveniently not shown in the Osram literature (their graph stops at 780 nm.), but is the dominant peak in all other continuous arc lamps I have investigated during my scientific instrumentation career. Certain objects, such as dark green or blue dyed synthetics, gray open-cell foam, and even the black border on the Macbeth color chart, all reflect strongly in the infrared; although the human eye cannot see this "color," it often affects the color rendition of a CCD imager, since nearly all of these devices respond better at 800 nm. than at any point in the visible spectrum. Even photographic film is not immune to this infrared output Kodak uses the term "anomalous reflectance" to describe the unexpected color shifts observed in these subjects. The presence of this strong infrared energy makes HMI a poor choice for some kinds of product photography, even though the lamps have a "daylight" color balance.

[MC] HMI and other arc lamps also share another characteristic that may make them less desirable for (scanning) photography the arc tends to "wander" slightly over time, causing small fluctuations in the lamp output which may be detectable with a good line-scan camera operating with a fairly high contrast. Although I have no direct data on Osram's HMI lamps, literature from Hamamatsu (a vendor of scientific photonic equipment) shows "stabilized" arc lamp intensity fluctuations of about 1% at a rate rapid enough to show up as line-to-line differences in an image. For example, if a digital camera is using a five-stop lookup table, each f-stop is divided up into about 50 parts (5 x 50 = 250 out of 256 available gray levels per color). One-fiftieth of an f-stop is about 1.4% luminance change, so the camera may be able to resolve some of the lamp fluctuations. With a tri-color array, the three colors are scanned at slightly different times, within a second or two; if the lamp changes during this time, very faint (1- or 2-bit) color "striping" may be evident in the image data. I have seen examples of this striping when high-frequency HMI lamps (Broncolor, in this case) were used with the Dicomed Digital Camera no such striping was evident when the same unit was used with tungsten lamps. HMI lamps also change color temperature with ambient temperature changes, much more so than fluorescent or tungsten lamps.

TUNGSTEN [MC] Nobody I'm aware of, but most of the Dicomed users I've met don't seem to mind waiting several minutes for their typical scan, and occasionally longer for the big ones. I agree that modeling lights are totally inappropriate (but think of all the flash heads you could sell), especially when so many dedicated tungsten sources, both soft and hard, are readily available. With the Dicomed back, scan time behaves just like shutter time half the light requires twice the line time, and therefore twice the total scan time.

[MC] Actually, many photographers still use 250 watt tungsten sources; Dicomed even uses a couple in its show setup (along with some bigger, diffuse sources). Another benefit of inexpensive sources is that a photographer can use multiple lights in a fashion similar to light painting (without all the hand-waving). Often, a light is used only to highlight a small part of the overall scene being captured. The art and craft of "sculpting with light" will still differentiate photographers from service bureaus, but don't be surprised to see the service bureaus taking over more of the plop-and-pop (ahem, plop-and-scan) "ordinary" work.

[MC] Although I really like most Lowel lights, the Tota is perhaps my least favorite, mainly because of the small reflectors, which are both inefficient and ineffective in tailoring the quality of light delivered. I appreciate your budget concerns, but spend a few bucks more and get a set of DPs (like I did); even with a similar (750W) bulb, the DP puts out almost twice as much useful light on full FLOOD as the Tota does, and it has an eight-to-one focusing range that can really throw photons where you need them. Put in 1000W bulbs if you've got several circuits available, and you'll get an extra 1/3 f-stop of light. For a softer, larger source, use a THIN fabric scrim or a spun diffusion material between the light and the subject; some materials can be clipped directly to the edges of the barndoors for a compact, efficient soft light. For a much larger soft source, Lowel's Rifa-lights work well; for more light, take off the front diffuser (somebody should make a little blip to reduce the direct light from the bulb for shooting this way, to eliminate the "hot spot"). Remember that you will have digital controls for globally adjusting the contrast of your images later (and even doing local corrections, if desired); use lights to get the surface detail and shadow (edge) qualities you want. It's still a good idea to practice measured photography and adjust your lighting to a consistent range, but you should have a little more latitude than 4 f-stops with digital photography.

FLUORESCENT [GR]With all the "bad" news about other continuous lighting, here's some good news. There seems to be a surprisingly close affinity between the discontinuous RGB output of a tri-phosphorous fluorescent lamps and the needs of now sacred CCD. Certainly, the peaks in the spectral output of fluorescent could be improved if this market can grow to the size to catch the interest of the large lamp manufacturers. Of course, the humble fluorescent tube is not as glamourous as the snappy HMI. Nor is it magical. But it can be built simply using readily found material.

Primary importance is the tri-phosphor lamps be balanced to daylight or better [higher Kelvin] and a high frequency ballast [Motorola, Osram] cycling at 25 to 39 thousand cycles per second. The configuration is up to the user. Fluorescent is already a "broad" source, so the problem will be how to concentrate the light, rather than, as with flash and HMI, how to diffuse and spread the light.

[MC]I am impressed with your interest in finding out the differences among the various light source types available and wish you the best of success. I think we will discover that fluorescent lighting systems will produce very good results in the studio (second only to natural daylight). My greatest concern is whether fluorescents will produce enough light to allow photographers to work at their accustomed apertures, since the "shutter time" of most digital cameras is restricted.


6. Appendix:
Definitions

Blooming [RT] a. A word used from the 19th century to present as an expletive, derived from English sailors describing the status of their furled sails. Modern example: "That blooming DCS CCD just bloomed on me!" or b. When oversaturation occurs in specular highlights of images shot with either area array or linear array CCDS, beyond the dynamic range of the particular CCD. The blooming in the specular highlight usually turns from white to yellow, to red if its really bad. This effect can also be seen with video camcorders when they pan the ceiling lights and streak to red.

With linear arrays, I experienced it in ultra reflective areas using singular light sources or pinpoint light sources on glass, crystal, glossy highly reflective material. This effect is why early- on DCS manufacturers used flat lighting, because many CCDS bloom quite easily.

Tip: When I would shot something that bloomed, I would locate the precise area on the monitor where blooming occurred with a un-defused light source. Next I would take a piece of tacky wax and rub some on the spot on the product, leaving a slight residue in this precise area. This would dull the area enough and prevent blooming. This method works very well. I passed it on to Robinsons*May and they use it every day in their studios.


Manufacturers:

Small HMI [200-1200W]

Bron Elektronik AG
Hagmattstrasse 7
CH-4123 Allschwil
Switzerland
fax +41 61 481 14 23

KOBLOD [Axel Fršmel]
Hans-Urmiller Ring 17
D-82515 Wolfratshausen
fax +49 8171 20367

K 5600, Inc [Gilles Galerne]
10434 Burbank Blvd
North Hollywood CA 91601
fax +1 818 762 6629

LTM of America [Laura Maurel]
11646 Pendelton Street
Sun Valley CA 91352
fax +1 818 767 1313

Tungsten

Lowel Lights
140 - 58th St.
Brooklyn NY 11220
fax +1 718 921 0303

Dedo Lights
Karl-Weinmair Str 10
D-80807 Munchen GERMANY
fax +49 89 356 60 86

Fluorescent

Videssence [Paul Costa]
980 David Rd.
Burlingame CA 94010
fax 415 697 7032

KINO FLO [Frieder Hochheim]
8824 Lankershim Blvd.
Sun Valley CA 91352
fax +1 818 767 7517

SCANDLES [Gary Regester]
432 Main St /PO Box Nine
Silver Plume CO 80476
fax +1 303 569 2932

 7. Recipe for a Digital Light under $300 usd

6 to 8- 20Watt to 32 Watt "Full" Spectrum "Grow" Lights from your [plant] Nursery or Tropical Fish store - 5500K or higher, CRI 90 or higher
6-8 x $20 $120-160
2 Motorola Ballasts such as 4-T8 lamps M4RNT81LL120 from a Motorola distributor, call 1 800 654-0089 2 x $80 $70
12 - 16 Two pin sockets from good hardware or electrical supplier about $12 to $24 for all

One commercial baking tray [used] from local baker. Bakers have a lonely early morning occupation, you [if you are a girl] or your girl friend can easily "chat" the baker out of a tray - but you may have to buy a sheet cake.

Misc. Wire [18 AWG solid copper], wire ties, wire nuts and an old grounded [earthed] AC (mains) plug. Altogether about $10..

Mix together and turn on.

Really, you need a electric drill to make a few holes in the baking tray to attach the lamp bases and ballasts and to pass the wires. You will also need some means to attach the light to a boom of light stand such as a Manfrotto "light tight" [you probably already have some thing similar]. You also will find that spray painting the baking tray WHITE [choose a neutral or slightly pink "white"] which will increase the out put of the light by almost a stop.

close this window