|
About Digital Cameras By Gary Gaugler, Ph.D. This discussion presents my views regarding the present state-of-the-art in digital photography. At this time, I don't believe that digital cameras (DIGICAMs) are ready for mass consumption--not toys or novelties but truly useful tools. This means that I would buy, keep, and use a camera that had the essential features I wanted and needed when priced at or below $1,300. While there are DIGICAMs and digital backs for medium and large format, these are significantly more expensive than my target price. Furthermore, some of the digital backs require a hardwired connection to a computer to use the back--you cannot shoot without the computer connected. There are a lot of technological advancements on the horizon that, in my opinion, will turn the digital photo market upside down. Happenings 1. At present, the control electronics are generally divided among at least 4 separate integrated circuit (IC) packages. These chips provide camera controls, compression, decompression, preview and replay. LSI Logic of Milpitas CA has integrated all of the essential functions into a single IC. The only other chips needed to make a digital camera is memory, CCD imager, and A/D converter. This new IC supports image resolutions from 640x480 pixels to 4 million pixels. JPEG codec is in hardware to speed up the processing of raw image data. Most cameras use a high performance RISC microprocessor and an application specific IC (ASIC). For example, Fujitsu uses a 66 MHz micro SPARC and an ASIC for a total cost of about $25. Fujitsu's chipset is used in digital cameras from Sanyo, Olympus, Agfa, and Epson. LSI Logic does not make cameras but will supply its new chip to traditional and new camera manufacturers. The new chip will be able to make a camera that can take pictures over 30 times faster than present cameras can perform. Even at megapixel resolution, this IC can help produce a camera that can take 3 pictures per second but at lower resolutions, it can take more than 10 images per second. [1] 2. Raw images are typically captured by a CCD. The normal problems of CCDs in digital cameras are blooming and poor resolution. Blooming is caused by a pixel receiving too many photons and causing the surrounding pixels to charge up, resulting in a dense white spot. Poor resolution is a direct result of the total number of pixels. The trade off is that more pixels can be contained in a given area if the pixels are smaller. Yet smaller pixels have less resistance to blooming. Furthermore, more pixels require a higher scanning rate to achieve the same image result. Higher scanning rates result in higher power drain. More pixels makes for a more expensive imager. A vicious circle, eh? 3. Newer imagers based on CMOS active pixel imaging technology are on the horizon. This type of imager will allow the inclusion of signal processing and high performance PIN diodes. Furthermore, they integrate the sensor, scanning, control and conversion electronics onto a single chip. The net result will be better performance and lower cost--in addition to higher overall quality due to the ability to include the signal processing functions right on the imaging chip and would not require the multiple supply voltages as do current imagers. [2][5] 4. Getting digital images out of the camera and into the computer and/or onto paper is another major issue. The Olympus D-300 for example is no better than a throwaway cardboard box camera. Why? Since it has only internal memory, once you fill up its meager storage with images, the camera is useless until you get connected to a computer and download the stored images. This is completely asinine and totally user-hostile in my view. Other manufacturers offer removable storage media. For example, PCMCIA is a standard portable media. Kodak uses a proprietary media of course. Look at Sony and their Micro Disk. It is a total flop since just as with their Beta format VCR, it was not an open standard. Contrary to company belief, consumers are not stupid. Now there is a new memory option--Compact Flash (CF). This promises to be a real winner. Look at your recent motherboard. You will notice a small header which is for the USB. This is the Universal Serial Bus interface. I think we will hear more about this feature. Why? For one thing, some cameras have a SCSI interface. This is nice and is fast but as you users probably found out, SCSI is not hot swappable. That is, you cannot have your computer running and then come along and plug in your SCSI camera and download its stored images. The reason is that the SCSI host adapter scans the SCSI bus at boot time to determine what peripherals are on the bus and records their IDs. Then, these peripherals are available via the operating system. Ironically, SCSI is a very good interface protocol and interconnect methodology. It is essentially device independent and allows addition and removal of devices without undertaking special procedures. The only primary consideration is that each device have a unique ID. The extension of SCSI to the Universal Serial Bus is poised to make major inroads to DIGICAMs. USB is uniquely positioned to provide a simple yet high speed (12 mega bits per second) device independent interconnect. Plus, USB supports hot swappable peripherals. Quoting from [4]: 3.3.4 Solid State Subclass A Solid State device is any linear memory device that has no moving parts and requires memory regions to be erased before individual bytes of data can be written. This device subclass is capable only of transferring data, so it does not require any isochronous endpoints. The interface to these devices borrows from the Direct Access Storage and Optical Memory Device protocol of the SCSI-2 standard. Solid State devices are random-access byte-oriented devices. By definition, seek time on these devices is 0. Read operations operate at the full speed the USB can provide. Depending on the device type, write operations are typically preceded by an erase operation. For solid state devices such as flash memory, if a write operation will change any bit from a programmed state to an erased state, the entire block must first be erased before the write operation can occur. Unscheduled events, such as media change, are generated via the Interrupt endpoint. A single word packet identifies the event that caused the interrupt. Although the Solid State device will utilize many of the actual packet definitions from the SCSI standard, it will NOT use most other features of the normal SCSI Protocol. Thus there are no phases, no messages, no shareable bus (only one host), and no SCSI hardware. This specification will make use of many of the Standard SCSI Command Descriptor Block definitions and commands, but some of the commands that would normally be supported by a SCSI device will not be supported for various reasons. What this means is that if and when USB is offered in DIGICAMs, we will have the best of both worlds--device independence, high speed data transfer, hot swap. How fast will USB be for downloading images? Here are a few simple example calculations: This does not suggest that I recommend that the camera be USB-based. It could, but only if it included removeable media as well. The point of useage ergonomics is that the camera should be just like a traditional 35mm SLR or even an APS camera. When the film is fully used, it is replaced with another unexposed roll and shooting resumes. PCMCIA cards and drives are the electronic parallel to empty rolls of film. Consequently, the PCMCIA cards can be collected and taken home where they are connected to a download box peripheral that is connected to the USB. Since USB is hot swappable, you could perform downloads at any time--unlike traditional SCSI. Furthermore, the control electronics for USB is becoming widely available at super low prices. I can envision a $99 box that connects to a USB port and accepts Type I, II, and III PCMCIA cards. Digital versus Film There is much debate about whether digital will replace film. I can see that in some circumstances and applications it will, if not already. However, for most types of work, film is very much irreplaceable. First, consider that film has a lighting contrast range of about 22:1. Video cameras have a range of about 3:1 under ideal conditions. The advertising industry shoots film and converts it to digital. Movies are treated the same way. When film is transferred to video or digital, the film's qualities are essentially retained. Thus, it is unlikely that digital will displace video. A similar comparison holds for digital still photography. To render film's quality via digital means would cause immediate impact on two fronts. First, the final file size would be huge. Second, the imaging sensor would also have to be enormous. A film recorder operates at about 2,500 lpi minimum for high fidelity recording on film. For a document size of 8x10 at 2,500 lpi you would get a file of about 1.5GB--over one gigabyte! And the sensor would have to be about 25,000x20,000 pixels. This is 500 million pixels and is 500 times more than the current megapixel products that represent high end digicams today. Even so, the resulting image would be essentially a flat light image. The digital output from any camera is RGB and cannot compare to CMYK output. Furthermore, continuous tone color printing is rather expensive and specialized today. Kodak and Tektronix are two of the leading suppliers of dye sublimation contone printers ($3,500 and up). The Ideal Camera--Today
Conclusion So what does this all mean? Sit back and wait for the really good stuff to come out. I think that we need to let DIGICAMs go through one more generation to get the camera that I think we all are looking for....at least, the one I am looking for. They are coming. Competitive pressure is making them come faster and faster. Unless and until my ideal DIGICAM comes out, I will stay with traditional film for most of my photo work. Unless you or I only want to upload images to the Web, then a present-day DIGICAM would be acceptable--within the operational limitations of each manufacturer's offering. The rapid access to images via digital cameras is becoming a necessity for certain types of work. Especially work where large sized output is not required and where high resolution is not necessary. Working in a large and medium format for most high quality output, my 35mm is for other types of photography. Since I am fully Nikon-based, I prefer the ability to use my existing lenses and accessories. And I prefer not having to convert focal lengths. In this regard, I have been using the Nikon E2Ns digital SLR. I had considered the Minolta RD-175 but found that its lower ISO rating and non-Nikon lenses and accessories, and loss of TTL flash, was not acceptable. Furthermore, the Kodak DCS cameras were for the most part based on Canon camera bodies and they simply are not acceptable to me in any regard. That is a personal issue of course. The best quality solution today is an optical SLR and a decent quality 35mm scanner. I use my trusty Nikon and the Polaroid SprintScan 35 Plus. For medium format, the best solution is an appropriate MF camera and the Polaroid SprintScan 45. (Note that I said best, not cheapest). I found that in order to make the Polaroid SprintScan 45 work at all, I had to buy a Mac computer. The PC with Adaptec AHA-2940UW, 2940U or 1542 will not work with this scanner. Either the system fails to boot, locks up during boot or the scanner drops off the bus after boot. It works perfectly with the Mac (G3-266/384MB/9G). References [1] Digital camera's essentials crammed into
one IC. EE Times, May 19, 1997. 954, p. 14. |
All Photographs Copyright © by Gary Gaugler - All Rights Reserved - Use By Permission Only |