Live Picture

          By E. Sapwater
          Article published in PHOTO>Electronic Imaging, Vol.37
          No.4, 1994; reprinted with permission.

     Bruno Delean is the creator of FITS   technology and the innovative digital
     imaging software package, Live Picture. The following telephone
     interview spanned several thousand miles-from PHOTO>Electronic
     Imaging's offices in Atlanta, to Delean's French headquarters in
     Toulouse. The 34-year-old native Frenchman was very positive about
     America and what it has done for him and his product. The
     three-hour interview, transcending both language and time-zone
     barriers, focused on the very beginnings of Live Picture and its
     development. Below, readers can glean the highlights of those
     talks and catch a brief glimpse of the genius behind this truly
     revolutionary technology.

     PEI: Where did you get your start in digital imaging?

     BD: Shortly after graduating from engineering school, I    [Image]
     co-founded Diadem, a high-end imaging company located in
     Paris. My partners were Guy Lehenaff, now a vice president of
     Triple I (Information International, Inc.) and Germain Nadaud, who
     has since founded several computer-related companies. I designed
     the Carat 500 and 560 series retouching and page assembly
     products, running on DEC VMS workstations. These systems were
     innovative in their ability to combine image retouching with page
     layout. In fact, my responsibility was to find a new technology
     that would make Diadem more competitive in the high-end publishing market.

     I was intrigued by the Quantel Paintbox-a system that gave users
     smooth, real-time editing through the power of hardware. I began
     to question if this real-time editing and manipulation power could
     be taken out of expensive hardware and placed into software. My
     intuition was to move on the idea, based purely on a competitive
     nature. It was a little more than five years ago when I came up
     with the idea of Live Picture-a brand new way of working with
     high-resolution images.

     PEI: How was your idea of Live Picture received?

     BD: Even though my partners at Diadem agreed it could be done,
     they thought I was crazy to attempt it. I don't think they
     believed it could replace a big system. At the time, competition
     was high and profits were declining. Our company was competing
     with the same tools and technologies found on all high-end digital
     imaging systems, and they were expensive when compared with
     software products like Adobe Photoshop-which could do basically
     the same thing we designed our equipment to do, but on a
     Macintosh. We couldn't successfully compete any longer. I felt I
     could do better on my own, by completely rethinking the software
     approach. So I left the company. And, with each passing day, it
     became more clear to me that, if brute force was replaced by
     complex and very sophisticated software, my theory could work.

     PEI: How did you go about developing your idea?

     BD: I retreated to my grandmother's studio in the Pyrenees
     mountains in the small country of Andorra. From her apartment, I
     could look out over a large ski resort and watch the falling snow.
     It didn't occur to me at the time, but snowflakes are resolution
     independent because of their fractile nature. I had the most
     beautiful example of resolution independence falling at my
     doorstep.

     I worked completely undisturbed-writing the code or mathematical
     core to Live Picture in longhand. I wrote the bulk of the software
     mathematics in about a year-and-a-half without a computer. At this
     point nothing was tested or visualized other than in my mind. I'm
     not able to visualize complex solutions while sitting in front of
     a computer. In fact, it was several years before I actually saw an
     image on the computer screen. The basis of Live Picture is about
     60 pages of mathematics. During this time, I lived off my savings.
     Thankfully, living was cheap there.

     Eventually, I left the mountains and began making the program a
     reality in the computer. All the longhand mathematics were
     converted to C++ software, which is what it is today. As the
     program came closer to reaching product status, I founded a
     company, FITS Imaging, to deal specifically with Live Picture and
     other applications of the FITS technology. I now had a team of
     programmers to support my work, and I could focus mainly on the
     core technology and algorithms. There are five founders of FITS
     Imaging: Robert Blumberg, vice-president; Mark Kalow, financial
     officer; Anna Godfrey, graphic artist and trainer; Adolfo Vide,
     chief programmer; and myself.

     PEI: Weren't you involved in another company between Diadem and
     FITS Imaging?

     BD: Yes, Sunline (the predecessor of FITS) was a company I created
     with Germain Nadaud, after we both left Diadem. We developed a
     product called Captitaine Color, which is now called ColorTouch. I
     developed the kernel and Germain did the marketing. We both
     developed the user interface. Adolfo Vide, one of our employees
     who remained as chief programmer of FITS, developed the I/O
     (input/output).

     We spent about six months on the initial development. ColorTouch
     was a very fast program-based on classic pixel editing technology
     similar to Photoshop. Images were put into little squares or tiles
     on the disk, allowing them to be accessed very fast. It was
     actually faster than Photoshop because it organized pixel data
     more efficiently. But, the technology and concept were the same as
     every pixel editing software in the world.

     PEI: Do you continue to own interest in ColorTouch?

     BD: Not any more. Germain chose to focus on ColorTouch working in
     the U.S. I stayed in France with Live Picture. Now, neither of us
     has anything invested in the other's product.

     PEI: Let's get down the heart of Live Picture.

     BD: Live Picture is a dramatic and revolutionary new way of
     working on the Macintosh with high resolution and/or multiple
     images. FITS Technology-the basis of Live Picture-has three basic
     concepts: The first addresses speed, the second has to do with
     resolution independence, while the third deals with infinite,
     selective undos.

     Live Picture occupies very little hard drive space, under one
     megabyte. Users find the application opens in just seconds. Unlike
     other programs, FITS Technology works on a proprietary file format
     we refer to as IVUE. The word vue is French for "to see." So, IVUE
     means "image view" or "I see."

     You can begin with a standard image created as a TIFF, Scitex CT,
     or Photoshop 2.5 format. The change to IVUE is fast, converting
     anywhere from .33MB to .5MB of data per second. A series of
     reduced resolution images are created and stored in the IVUE file
     itself. Convolution is applied at each level to ensure the highest
     imaging quality. The size of the tiles and the number of
     sub-images are not key factors. However, currently we reduce the
     resolution by two in each direction, i.e., four overall with each
     successive sub-image. For this reason the IVUE file is one-third
     larger than the original size. It's during this IVUE
     transformation that the pixel-based information goes through a
     change, effectively converting pixel-based information into
     mathematical equations. So, in essence, the pixel images are
     stored in IVUE format. This IVUE format permits fast image access
     and viewing.

     Then a FITS file is created. Any transformations or manipulations
     applied to the image are stored there. The IVUE format is
     necessary for FITS, but it all works together very efficiently.
     FITS files are resolution and size independent. If one were to
     draw an analogy: PostScript is to bitmapped fonts as FITS is to
     pixel imaging.

     PEI: Can you expand on this analogy?

     BD: Vector shape is resolution independent. For example, if you
     draw with bezier curves in FreeHand and zoom in, the edge of the
     outline will be smooth-without the big pixels or visible aliasing
     that would be apparent if you were to do the same thing in
     Photoshop. That is because a vector image is described by a
     mathematical curve rather than with pixels. When the image is
     drawn on the computer screen, the vector shape is filled and you
     have perfect image quality.

     But, in FreeHand, there is no airbrush or color correction based
     on mathematical equations. You're limited to text manipulation.
     This is how PageMaker, QuarkXPress, Illustrator, and FreeHand work
     using vector technology. FITS is the first technology to
     incorporate both vector drawing and image manipulation. This way,
     in terms of pure technique, it makes ripping possible for images.
     I developed FITS because no one else wanted to-nobody believed in
     it. I thought it would take two years. It took much longer than
     that. The engineering is very complex.

     PEI: Please describe the FITS file in greater detail.

     BD: The FITS file is an interchangeable imaging database and rip
     engine. It contains all the image data in a way that allows the
     rip engine to provide a mathematical expression of each color in a
     point. Instead of storing all the pixels of the photo montage, it
     stores the mathematical expressions of the effect. What we claim
     to be resolution independent, is actually effect independent,
     based on the color of the image.

     If an image is stored as pixel dependent, and you want to zoom in
     on any portion of it, you have resolution limitations. But, if all
     the painting and manipulations are described as mathematical
     formulas, which are position dependent, then any portion of the
     image can be represented at any resolution. Mathematical formulas
     can be calculated on all the pixels required to be seen on the
     screen or ripped to the output. This is how FITS works. It is the
     engine we have needed for years to make this technology usable.

     I wanted to make all the parameters as simple as possible by using
     a very special concept of interpolation. We don't interpolate the
     value of the pixels, we interpolate mathematical expressions.
     There are very few formulas stored in the database. Instead of
     thousands or billions, we store hundreds of mathematical
     expressions. We calculate one formula for each pixel seen on the
     screen. The whole concept is extremely complex.

     If you make a brush stroke or zoom in on a portion of an image in
     Photoshop, you'll see squares. If you want to smooth the squares
     you're going to have artifacts. In Live Picture, if you only do
     painting, you can zoom in infinitely. You can print a poster 100
     meters square and you won't see any problem or artifacts. FITS
     technology was necessary to achieve the effect, speed, and
     performance of a Quantel Paintbox on a Macintosh.

     In order to use a big airbrush with current pixel editing
     software, a very powerful computer is required to calculate all
     the brush strokes. In Photoshop, when you use brushes between 20
     and 50 pixels it slows down. On a Quantel Paintbox you can go up
     to 300 or 400 pixels before it becomes terribly slow. Barco
     Creator is similar. But with FITS, the airbrush can be 10,000
     pixels in size, because the limitations are not linked to the
     power of the computer.

     PEI: Before we get to the output side of Live Picture, let's dig
     deeper into what separates it from other pixel imaging packages.

     BD: The main innovations in Live Picture that are not found in
     other programs such as Photoshop are:

          Resolution Independence.
          All effects created in Live Picture, such as painting,
          filters, and masks, are resolution independent. This
          means that they can be output at any size without
          resorting to interpolation of pixels. The effects are
          mathematically regenerated at the time of output, during
          a processing we refer to as FITS raster image processing
          (FITS RIP) at the specified resolution.

          Selective Undo.
          Because of FITS' mathematical approach, any effect in
          any layer can be progressively modified or completely
          undone Layers.

          FITS layers can be reordered, deleted, resized, scaled,
          etc. Although other programs have "layers" or "objects,"
          they don't treat brush effects, such as painting,
          lighting, blurring, and color correction, as layers. The
          number of layers is unlimited and doesn't depend on RAM
          memory.

          IVUE Technology.
          IVUE is unique in that it permits both fast viewing and
          geometric changes. With IVUE, you can basically zoom
          into a 300MB image in just a few seconds,
          guaranteed-depending upon your equipment configuration.
          No special hardware or extra memory is required. The
          other important aspect of IVUE technology is that Live
          Picture can work directly on JPEG compressed images. It
          doesn't have to decompress the entire image before using
          it. Only the necessary pixels are decompressed. This has
          tremendous implications for IVUE use on networks,
          telecommunication lines, and CD ROMS.

          Brush Technology.
          In Live Picture, everything is done with brushes-not
          just painting, but image insertion, ghosting,
          distortion, color changes, blur, sharpen, etc. The brush
          size is unlimited. For example, you can sharpen a 100MB
          image of the moon with a single brush stroke. And the
          brush operates in real time, i.e. its speed is
          independent of the underlying image. Masking.

          I wrote the Live Picture masking technology to be
          semi-automatic. It integrates easily. For example, when
          you have a woman on a white background and want to
          insert her on a black background, normally there would
          be a white line around the edge of the girl. There is
          interference from the background. Whatever you do for
          the silhouetting, when this object is placed on a new
          background it is going to look wrong. It's not going to
          look realistic.

          Live Picture has semi-automatic silhouetting. It takes
          the previous background color and, when the object is
          silhouetted, automatically calculates the difference
          between the previous and new backgrounds, and sets the
          edge. It makes what we call "Chrominance Compensation."
          This works nicely with hair, which is difficult to mask,
          and the effect is very realistic. This technique is
          very, very new, I think, in Live Picture. But it does
          require training.

     PEI: With resolution independence and the use of sub-images, is it
     possible to easily cut and paste images of different resolutions
     together without compromising resolution?

     BD: Absolutely. It's a totally new way of dealing with imaging
     technology. Most people think that speed is the best thing going
     for us. But, they must also understand resolution independence or
     multiple undos if they hope to begin to grasp what FITS technology
     is really about. Sometimes I wonder if the world is ready for this
     type of change.

     PEI: There have been statements that, with Live Picture, there are
     no pixels. Can you clarify this and give us an overview of how
     Live Picture actually approaches image editing?

     BD: That can be a little misleading. Within Live Picture all
     effects performed are resolution independent. You start off with
     pixel images. In fact, IVUE is a pixel image format. Pixel images
     are generated during output. But, with Live Picture, using FITS
     technology, we work within an "imaging space" that is resolution
     independent. This means you can actually merge images of varying
     resolutions. There is no fixed resolution until the image is
     output. At that time, the user determines dimensions and
     resolution-and Live Picture generates it. It performs convolution
     and anti-aliasing as necessary.

     The key to FITS technology is how image effects, such as sharpen,
     blur, coloring, painting, etc., are represented in a way that's
     resolution independent. We characterize them mathematically and
     store them in a FITS file. With standard pixel technology, these
     effects are calculated and recalculated for every pixel, and
     stored in each one permanently. Live Picture holds a series of
     layers containing a number of effects. This separates the creative
     process from the mechanical/technical process. This is critical.

     PEI: What tools are resolution independent?

     BD: All of them. The smallest brush can be reconfigured to any
     size you want. You can cover the largest poster with the smallest
     brush in Live Picture. You can zoom back and see the poster on the
     screen, then zoom in after you've applied your brush effect and
     see it covering the whole montage. This shows you the effect of
     resolution independence. Enlargement, cloning, distortion,
     silhouetting, rotations, and variations are all the same
     technology.

     The biggest difficulty was making all of these tools. Some were
     quite easy to do, like the brushes, rotations, etc. But
     silhouetting, cloning, and color corrections were more difficult
     to achieve. To include all the features that the market demanded
     in the Live Picture package-without going back to pixels-was quite
     difficult. Cloning took one year. Silhouetting took one year.
     High-quality color correction using this technology was especially
     difficult.

     Tim Gill, president of Quark, told me that, with future releases
     in digital imaging software, many will claim they can do the same
     thing as Live Picture. But nobody will be completely resolution
     independent unless they have FITS technology. They will all go
     back to pixels at some point.

     PEI: What are the minimum and maximum hardware requirements for
     Live Picture?

     BD: The minimum is a 68040 processor with 32MB RAM, and a 24-bit
     graphics card. The optimum configuration would be only to add or
     allocate 64MB of RAM and add a graphics tablet to Live Picture.
     Anything above this in RAM doesn't really increase the speed. This
     way, if your machine has more RAM, it can be made available to run
     other applications.

     PEI: Will Live Picture run on PowerPC?

     BD: We have PowerPC here and have ported an initial version of
     Live Picture. The native port took two days. But we haven't yet
     measured the performance, since our configuration is incomplete.
     It is certainly not a problem to run on PowerPC. On a Quadra, the
     screen rendering time is 10 to 30 seconds, depending on the
     complexity of the image. We'd love to bring this down to several
     seconds using the PowerPC in all cases.

     PEI: Let's continue to discuss JPEG compression. If Live Picture
     allows manipulations in virtually a real-time environment, then
     why is compression even needed?

     BD: Here's one overview. Clients may live a considerable distance
     from their service bureau and require scans to be delivered in
     different ways. Sometimes high-resolution scans exceed well over
     100MB. The TIFF scan could be converted to IVUE format. Then, with
     JPEG compression, an IVUE file could easily be stored on floppy or
     maybe sent to another location by modem, and the file could, be
     retouched in Live Picture. Then, the FITS file-not the IVUE-would
     be shipped or modemed back to the service bureau. The bureau would
     simply take the FITS file, apply all image changes to a working
     copy of the original file, then this newly modified image would be
     built to a TIFF, or some other format, for output. This works
     because the original FITS technology is built into the stored
     high-resolution data, and the FITS file contains all the
     information needed to create the final hi-res output.

     PEI: With image data being exported and imported, will there be
     any loss of quality because of this back-and-forth exchange?

     BD: There's no loss of data or quality when going back and forth
     with TIFF or any of the other formats we support when transferring
     files. It's a great advantage.

     PEI: This IVUE approach sounds like its working with a proxy. Or
     is it different?

     BD: It's different. With the proxy approach, there are two sets of
     data to deal with. A proxy is a fixed reduced-resolution image
     representing the full-resolution image. Proxies are smaller in
     size and, therefore, more manageable to most image editing
     software. When proxies are used, people perform their image
     manipulations directly on the proxy. Then, after the image
     processing is completed, they switch the proxy for the full
     resolution image, applying those effects in full.

     PEI: Is it the mathematical approach to IVUE that makes the
     difference between proxy and Live Picture?

     BD: As we've discussed, using the IVUE approach dynamically
     regenerates the image. The IVUE file contains not only the
     original image but also a series of reduced-resolution sub-images.
     Think of it as a pyramid, with the highest resolution file at the
     base and the lowest resolution sub-image at the top. You always
     see as much image data as you need, based on screen size and
     magnification level. If you have a 200MB image of a person, you
     can zoom into the eyeball, down to the point where you get to the
     grain of the image. In effect, you work on the actual
     high-resolution image.

     With this approach, zooming takes just a few seconds. You can zoom
     into whatever level you want-always hitting the actual data, as
     much data as you can possibly see on the screen. When you're
     finished with the manipulations, effects are applied to the
     high-resolution data, much like the proxy approach. But, unlike a
     proxy, you're working with hi-res data all along, and there's no
     cumulative processing error-all mathematical information is
     calculated once.

     PEI: In cumulative processing, a scripter memorizes all steps
     performed on a proxy and carries out each step during
     postprocessing, regardless of whether the effects performed are
     included in the final version. On the other hand, with the FITS
     Rip, there is non-cumulative processing. Can you provide an
     analogy of this non-cumulative process?

     BD: It's sort of like a particle and an anti-particle annihilating
     each other. After that, there's nothing left. These would be
     represented by equations. During ripping, you take the
     mathematical expressions and evaluate them, ignoring unwanted
     steps. (For example, if you perform a blur and then undo it, then
     you apply a distort and keep it, FITS will bypass the blur
     altogether-processing only the distort effect.) Suppose you play
     golf. You strike the ball and it lands 20 meters from the hole.
     Then you do it again and it gets closer. It takes six steps to get
     the ball in the hole. But what if you are 50 meters from the hole
     and you have to get it in with just one strike. This may be
     difficult to do. With FITS, you can hit the ball and get it close.
     Hit it again and get it closer, then hit it once more to get
     closer. Finally you succeed. If the ball goes too far you can go
     back from this position.

     With golf you can play as many times as you want, but you have to
     play it diligently to improve. With Live Picture it may take many
     attempts to get the ball in the hole, but in non-cumulative
     postprocessing the ball always goes straight in. In Live Picture
     you may rework many areas of an image several times, but the final
     rip only applies to the effects of all the desired changes.

     PEI: What about this build process or FITS RIP before output?

     BD: Live Picture builds or rasterizes data as a TIFF file for
     output to an imagesetter, film recorder, or the like. Users
     basically tell the program what dimensions and resolution are
     needed. There are three basic settings: Standard, Fine, and Extra
     Fine. Any one of these may be chosen, depending on the output
     quality desired. The Standard setting produces output that's at
     least the resolution of the original scan. Fine and Extra Fine use
     over-sampling to produce the smoothest blends and curves, better
     than anything else.

     Let's say we're working on two separate graphics: One is a full
     page, 8.5x11 inches, with a resolution of 300 dpi (25MB), and the
     other is a 100MB poster or transparency output. For the 25MB
     image, some typical build times are: Standard-2 to 5 minutes;
     Fine-10 to 15 minutes; and Extra Fine-40 to 60 minutes. While the
     100MB file would be: Standard-8 to 20 minutes; Fine-40 to 60
     minutes; and Extra Fine-2.5 to 4 hours. Realistically the Standard
     build is more than acceptable for most files.

     PEI: What are the pros and cons of this type of postprocessing?

     BD: With Live Picture we implement FITS RIPing. The advantages
     are:

          Quality.
          All image edits are resolved to a small number of
          mathematical operations that are applied at one time.
          For example, as explained earlier, if you rotate an
          image five times during the course of image editing,
          Live Picture only applies one cumulative rotation at the
          end; thus you avoid the cumulative errors normally
          incurred in step-by-step image editing.

          Time.
          Because all image editing is performed in near-real
          time, there is a tremendous gain in operator
          productivity. You not only minimize processing time but
          defer any delays until the end when you are free to let
          the computer work on its own.

          Server operation.
          For production environments where even the final time to
          build the image is not acceptable, the FITS RIP can be
          executed in a separate server.

          The only disadvantage is that the image is not available
          immediately at the end. But even this is not obvious,
          because saving a large file in Photoshop can take longer
          than building an image in Live Picture.

     PEI: Let's step back a minute and take a look at HSC Software's
     involvement with Live Picture. Isn't the company publishing the
     package in North America?

     BD: That is correct. The way everything came together was
     remarkable. After working for years, we finally showed the initial
     prototype for the first time in Paris in February 1993. One of the
     persons who attended was Andreas Pfeiffer, editor of the major
     Macintosh magazine in France, SVM Mac. Andreas was stunned by what
     he saw and immediately sent out a special edition of what he
     called the "MicroCosmos Design Fax." Kai Krause at HSC heard about
     this within 24 hours. A few days later, Kai and HSC President John
     Wilczak stopped in Paris on their way to CeBit and met with
     Andreas and Robert Blumberg, vice president of FITS Imaging.

     From there it's history. Kai was the only individual who truly
     seemed to not only understand what I was trying to do, but he also
     was in a position to offer assistance in a business sense. He
     originally tried to find a distributor, since he only wanted to
     write plug-ins for us. In the end, Kai and his company were the
     most excited. They decided to take the package and go with it.

     PEI: Did other companies not want Live Picture?

     BD: The first time we showed Live Picture was at Seybold Boston in
     April 1993, and we got a lot of people interested. We had contact
     with companies like Quark, Aldus, Adobe, and Apple. But, due to
     their own business directions, it just didn't happen with them.

     PEI: Did you ever feel discouraged that this package wouldn't get
     distributed or sold?

     BD: When you first show Live Picture to top-level people, like
     John Sculley, former CEO of Apple Computer, there is very, very
     high emotion. You can just feel it. When people see the product
     and understand it, they love it. But for those who had just heard
     about it-it became a hard sale. So, I was very discouraged that,
     in the end, nothing was happening. Since last summer we were
     thinking about going with Apple. I heard Adobe got very, very
     upset when Apple began demonstrating Live Picture on their
     machines instead of Photoshop. They were probably strong enough to
     influence Apple to forget about their interest in us.

     PEI: Kai has also implied he is interested in producing his own
     image editing package with FITS. Is this something you can talk
     about?

     BD: I would like him to do it. But it's going to take us a little
     time before we can actually deliver the technology to him in such
     a way that he could build an application. We continue to discuss
     it and we're getting closer. We also expect him to develop an
     effects plug-in that will enable Kai and others to develop
     resolution independent image editing filters for FITS.

     PEI: Can you give us an idea of the speed of Live Picture compared
     to say, Photoshop?

     BD: Live Picture always takes one to three seconds on a Quadra to
     rotate an image of any size. If it takes Photoshop one minute, it
     takes Live Picture one to three seconds. If it takes Photoshop 10
     minutes, it still takes Live Picture only one to three seconds,
     etc.

     PEI: Is Live Picture something that can be ported over to
     Photoshop as a plug-in?

     BD: No, it's not a plug-in. We have tried to provide as many links
     as possible between the two products. As a stand-alone application
     it could be adapted to accept the Adobe plug-in standard. The main
     compatibility is at the file level: Live Picture can import and
     export Photoshop 2.5 files. Within Live Picture these are
     converted to IVUE format, which itself is a pixel format, to
     achieve fast viewing. We'll include IVUE plug-ins as well, so that
     other programs, including Photoshop and Painter, can import and
     export IVUE files. We will support acquire and export plug-ins.
     Additionally, Live Picture channel masks created in Photoshop can
     be imported; they are converted to resolution independent FITS
     masks within Live Picture.

     PEI: Will editing packages like Photoshop be left out?

     BD: It's very difficult to improve an aging product like
     Photoshop. If Photoshop wanted to become resolution independent,
     they would have to start from scratch. Any concept of image
     editing is out when compared to FITS, because nothing else works
     like it does. It would be great if Photoshop or PhotoStyler could
     directly create the FITS format. We hope they create products
     based on FITS technology and we'll provide a way for them to do so
     if they desire.

     PEI: With all the power Live Picture promises to bring to image
     editing why would anyone want to invest heavily in a high-end
     workstation?

     BD: It's true Live Picture is basically positioned as
     workstation-level retouching on the Macintosh. What tends to
     distinguish the high-end solutions, such as Scitex, Barco,
     Linotype Hell, etc., from the Mac solutions is that the philosophy
     on the Mac is the ability to hand-pick multiple programs made by
     different vendors. High-end systems, on the other hand, provide a
     complete set of functions. For example, if you want to do
     illustration and image editing on a Barco you need to buy two
     programs. They're only available from Barco and thus they cost a
     lot more.

     Live Picture competes directly with high-end systems on
     power/speed. Of course, on price/performance, Live Picture has no
     competition. But, we have to recognize that no other system has
     features equivalent to Live Picture (because it's built on FITS
     technology). Obviously, the workstation vendors are having a hard
     time now. A recent issue of UK's leading prepress magazine
     indicated that Quantel hasn't sold a graphics Paintbox in 18
     months.

     PEI: There's one question that seems to lurk in everyone's mind.
     With Live Picture, why would I even need Photoshop any more?

     BD: Because Photoshop, Painter, and others have their particulars.
     They have features and plug-ins people love to use. I'm not going
     to create the same thing that Photoshop does. The designer wants
     unlimited creativity. Adobe is already claiming that the brushes
     in that package are better than Live Picture's. The users are
     going to decide when and if they're going to discontinue using
     Photoshop. But this will take some time.

     We have to convince people to learn and accept a new product as
     best we can. Live Picture will not be Photoshop. It cannot be
     Photoshop. That's the point. Even if we take the same interface,
     they're not going to find Photoshop in there. It's going to be
     something new. They have to learn it, and we have tried to make
     that as easy as possible. It will be easier to use Live Picture
     than to ignore it.

     PEI: Many of the popular PC and Macintosh image-editing companies
     have included color management, either their own or someone
     else's. What is your approach?

     BD: We have our own color management system embedded in Live
     Picture that produces separations on output. You can work in Live
     Picture and then output separations at the same speed, so there's
     no penalty for going out in CMYK. In fact, the IVUE Toolkit, which
     will soon be commercially available, works with RGB, CMYK, or any
     other color model. However, we believe that the designers should
     work in RGB because it is a larger color space and offers a better
     variety of colors than CMYK, and because it is the color space of
     the display monitor. Then, color management systems like
     KEPS-FotoFlow, and EFIColor will be used to guarantee color
     accuracy on output. We think it's the right approach to be
     consistent-to support color management, ColorSync, EFI, and KEPS,
     so that people can color match their scanner, display monitor, and
     output device. In the future, I expect we'll support Apple's
     ColorSync 2 and add more support for CMYK.

     PEI: Are you working on seven-color printing?

     BD: Yes. In fact, I think all the big industry players, including
     Scitex, Indigo, and Crosfield, are working toward developing more
     than just the CMYK approach; like five, six, or seven colors. We
     know that CMYK is limited. When you see the photo on the screen,
     you would like to print it as it is, with the same colors. That's
     not possible in CMYK. There are limitations. Seven colors is one
     answer. It's a possible way to guarantee that your screen values
     are going to be printed more closely. Because FITS describes
     everything as mathematical effects, it maintains the data in its
     original state, even if you go to CMYK output. But in the prepress
     industry, the scanner takes the original data input as RGB, not
     CMYK. The scanner itself makes the conversions. The CMYK file has
     lost colors that cannot be replaced, even when going back to the
     RGB mode. With FITS technology, when the image is finished, you
     can output it on any kind of output device.

     If you go to a prepress machine like the Scitex or Hell, the
     colors you get are always very fat, because data is stored in CMYK
     and viewed in simulated CMYK on the monitor. But on output, you
     lose the beautiful green or blue you see on the screen. You lose
     everything. This is why it is a limitation. It's important to show
     the world that whatever they do with this product, they keep the
     possibilities open to print their work on a film recorder,
     imagesetter, or printer-and eventually on the widest range of
     colors I know, which totals seven.

     PEI: In Live Picture you're working in a 48-bit color space. What
     occurs when viewing 48-bit graphics on a 24-bit monitor?

     BD: I must be honest. If I acquire the external image in 24-bit, I
     will keep it in 24-bit. But if I acquire the image in 48-bit, I'll
     only see it in 24-bit on the monitor. When I view it, I'm not
     seeing the actual data quality. The monitor screen is forgivable
     in terms of quality but the output is not. If I manipulate the
     data using traditional image editing software in 8-bit pixels (256
     colors per red, green, and blue pixel), then store it in 24-bit
     RGB or 32-bit CMYK, and then manipulate the data, each time I do
     an effect, I change the byte value. I begin to lose image quality
     due to compression and expansion of the colors. The more effects I
     perform the more deterioration I'll experience. In this kind of
     image-editing environment, where there are only 256 levels of
     color to describe the paint, stochastic noise is added to the
     paintbrush to avoid having flat steps. If you zoom in, you'll see
     many artifacts.

     When I manipulate the data in 16-bit pixels that have 65,536
     colors per red, green, and blue, the deterioration is much less
     severe. I have greater parameters to work within. I have more
     freedom to make changes without throwing away an inordinate amount
     of data, even when I experience greater compression and expansion
     of color. Not only will the data look good on the monitor, but
     I'll also be able to obtain the best quality output. This is
     important for color separations. If you paint in Live Picture and
     you rip in CMYK or RGB, you're result is going to be much better
     in terms of quality than if you do the same thing in Photoshop and
     print it. You won't have noise in the paintbrush and you won't
     lose any quality.

     PEI: FITS technology and Live Picture have really come a long way
     since you began writing code in your grandmother's studio. What's
     next?

     BD: I'm finalizing everything I originally wanted to do. Then I'll
     look at how to make people better understand what FITS is and how
     to use it in the most efficient way possible. This is something I
     must do.

          Copyright � 1994 PHOTO>Electronic Imaging & Inner Circle
          Technologies, Inc. Reprinted with permission.

    ---------------------------------------------------------------------