PERFORMANCE TESTS
Cameras
We photographed ISO targets at noninterpolated resolutions and in an uncompressed format (when possible) under controlled conditions, with digital zooms disabled and lighting designed to obviate the flash function. The transition pixel ratio reflects a camera's ability to capture sharp images. We ascertained this ratio by counting the number of grey pixels that appear where a black and a white edge meet, relative to the length of the border. The fewer transition pixels measured, the cleaner and crisper the image. Since a camera's horizontal and vertical resolving capabilities can differ, we offer both results. Resolution is a camera's ability to discern individual lines on an ISO target--again, both horizontally and vertically.
A high number of lines indicates a good resolution. Digital Benchmarks (www.digitalbenchmarks.com) performed the analyses of these cameras.
Scanners
To test the performance of these scanners, we connected them via USB and then hand-timed several types of scans, starting via the TWAIN interface and timing until the images were fully rendered in Adobe Photoshop. We scanned each target image at 300 dots per inch, and we scanned the monochrome photo as a greyscale image. The first target was a 4" x 5" colour photo, which we scanned in 24-bit colour (RGB). The second was an 8" x 10" monochrome photo, which we scanned as an 8-bit greyscale image, and the final target was an 8" x 10" colour photo, scanned at 24-bit colour (RGB). Photo quality ratings, from our expert-jury evaluation, reflect dynamic range, colour, and resolution and are measured on a 10-point scale. Graphics quality, also based on a 10-point scale, takes into account exposure, colour focus, and distortion.
Printers
The Epson and hp printers used usb, and the Canon had a parallel-port connection. We used the highest-quality settings for all the photo printers on vendor-supplied photo paper. We selected a variety of document types and hand-timed the printing of each. If the document was more than one page, we clocked the first-page and last-page completion times, from the print command until the page was resting in the output tray. In the text and graphics category, the Word document contained simple text and both vector and bitmap images. We tested many graphics files as well. The Photoshop image was an 8 x 10in, 300dpi tiff created in Adobe Photoshop 5.5. The first Excel document we timed was a 1-page Microsoft Excel spreadsheet containing graphs and charts. The Excel document was three pages and also contained graphs and charts. We compared photo quality for colour, dynamic range, and resolution, and we checked graphics for posterisation, banding, and registration.
Hard Drives
Business Disk WinMark 99 measures the performance of the disk subsystem. Disk Access Time reflects how quickly the drive can access data after receiving a request; it includes command processing, disk latency, and seek time. High-End Disk WinMark 99 measures performance for graphics professionals. The DV format has a fixed data rate of 3.6MBps, but because high-end editing often involves mixing multiple video streams, real-time cards must be able to retrieve two streams simultaneously, or about 7MBps. Most of the SCSI drives start out with an initial transfer rate of about 35MBps on the outer edge of the disk and taper off to a final transfer rate of about 25MBps on the inner edge of the disk. The IDE-based Maxtor drive does have the throughput for this type of editing--with DMA enabled. But if you forget to turn on DMA (dynamic memory allocation), you may notice markedly worse performance. The CPU utilisation test reflects the amount of stress on the processor.
1%
4%
