High Quality Scans
As a little exercise before you read on, write down the letter of the image you prefer.
Image A
Image B
Please read on!
The Engineering Definition of High-Quality Scans
1. Low-noise images: This is hard to define, but easy to see in an image of a clear blue sky made into a poster-sized image. The noise shows up as random image artifacts similar to jpeg compression artifacts. The noise is generated by the scanner in all kinds of ways. Vibration, movement of any kind, quality of the CCD and quality of the imaging electronics all play a significant role in random noise.
2. High MTF: Modulation Transfer Function (MTF) is a test that engineers use to establish the scanner's ability to resolve image data. This data is rarely if ever published because it is not easy to translate into simple scores. High MTF scores mean the scanner really does resolve detail (this is not dpi!) and improve color capture.
3. Low Delta-E: Delta-E is a way of measuring color accuracy. The highest accuracy in flatbed scanners are XY scanners. The only reviewer I've seen mentioning Delta-E as a way to evaluate scanners is Popular Photography. Way to go Popular Photography!
What About Image A or B?
Image A is from a Polaroid Sprintscan 45. A really good film scanner with lousy film holders. The second image is from a Microtek X12USL. The X12USL retails for about $300 at the time of writing and the Polaroid scanner was about 10X that. The Microtek images are very noisy compared to the Polaroid. So they won't reproduce as well as the Sprintscan 45. See the articles below about resolution for more imaging information.
In conclusion, it's not very likely a potential end-user will ever see objective scanner performance data like MTF or Delta-E information from manufacturers.
There are a few good flatbed scanners out there. The Scitex products are great.
Mis-conceptions regarding High-Quality Scans
1. High Quality Scans = High Resolution Scans. Most resolution claims have little to do with the performance of the scanner. For a thorough discussion, see below.
2. A higher bit depth is some kind of reflection of performance. This is not at all true. The higher bit depths matter most at the extreme end of scanner pricing.(US$3000+)
3. Drum scanners are always better than flatbeds. Drum scans are largely money down the drain for 80% of drum scanner customers. Most print production is done on low-resolution presses. Ink-jet resolution is a paltry 250 Pixels Per Inch even on US$1500+ printers and wide-format printers, so a drum scan is wasted money. There is no better scan than a drum scan done well but there is rarely an oportunity where it makes a huge difference.
Scanner Resolution
Scanner resolution claims made by the manufacturers used to be based on "conventional" 5000 element (or less) tri-linear array CCD. The CCD's linear array is part of a calculation taking into account scan area, stepper motor and voila! resolution is determined.
A tri-linear array is three(tri)rows(linear array)of CCD elements. In flatbed scanners, the industry has moved from three to six rows of CCD elements and claimed double the resolution. The scanner really doesn't resolve double the resolution, but there is a resolution improvement over conventional tri-linear CCD's. Performance is not very near what the vendors are claiming because other parts of the scanner have been so compromised that the scanner doesn't resolve very much detail. Even high bit-depth scanners suffer from poor performance in these situations.
Compare these two details from an IT8 target image.
Question:Which one would you think is a 1200 DPI scan?
The image on your right should have less detail than the image on your left. If you were to print the two images, the image on your left should be more pleasing in a way that may be difficult to describe.
Answer:
Both images are 1200 dpi.All content copyright 2002, Michael Papet.