EPSON Expression XL & XL Pro

A major part of our feature on reproducing art has been concerned with the performance of the Epson Expression XL. Here Lukman Sinclair and Mike McNamee show the results of their investigations.

General (LS)

A3-sized scanners are quite a rarity but important for many art reproduction tasks. The most common size scanned by GMB is A2 but the A3-size market is quite substantial, many artists finding that A4 is a tad small to work on. The Epson Expression 10000XL comes in two guises, the XL Pro has an additional transparency hood along with its own lamp, optimised for transparency scanning. It is not a new product, originally appearing around 2004 if the footnote on the data sheet is anything to go by. The key parameters are shown in the table below:

· New MatrixCCD optimised for A3 scanning
· High bright Xenon lamp for zero standby time
· Advanced auto-focus system with manual focusing
· Dedicated ASIC for hardware image processing
· Optical density of 3.8 Dmax
· Built-in USB 2.0 Hi-Speed and IEEE1394 (FireWire)
· Optional 10BaseT/100BaseTX network interface
· Optional A3 Automatic Document Feeder
· Features SilverFast® Ai 6 complete with IT8 calibration targets

Optional A3 transparency unit:
· Features two separate lamps for optimised scanning
· A3 scanning area takes up to 48x35mm frames
· Robust 35mm, medium format and 5"x4" film holders

This is not a small scanner (they never are!), weighing 14.5kg and requiring a desk area of 656mm width by 458mm depth. The height is 158mm but requires space to open the lid. The lid is removable for scanning books or very large originals. Connectivity is via Firewire or USB 2. The scan interface is the typical Epson version, which is very easy to use in both normal and professional modes. Set-up was trivially simple and the scanner is also accessible via TWAIN from Photoshop.

The document sizes are:
Flat Bed: 12.2x17.2 inches (310x437mm)
Auto Document Feeder (optional): 11.7x17.0 inches (297x432mm)
Transparency Unit (XL Pro only): 12.2x16.5 inches (309x420mm)

Scan resolution is 2,400dpi optical. The transparency scanner is supplied with holders up to 5x4 inch, but scanning is available over the 309x420mm area. Scanning up to 4,800dpi is available via a sub-scanning which moves the carriage.

The scanner is quiet and fast in operation (a full A3 scan takes 33 seconds at 300dpi and just 14 seconds at 150dpi) and, once the bulb is warmed up, there is minimal fall-off of illumination around the edges. The lid is heavy because it incorporates the transparency scanning unit, but the hinges are solidly engineered and clever damping ensures that the lid stays up when lifted with no danger of dropping it back down accidentally – it actually needs gently pushing back into place. The only disadvantage of the lid is that it cannot be removed, so that scanning larger artwork in sections and stitching together is limited to about A1. This is unlikely to be a problem for me as I currently print up to A2 on an Epson 4000 with plans for A1 printing in the near future with an Epson 7880.

The Expression 10000XL offers a substantial improvement over my current A3 scanner which has proved clunky in use and has been poor at differentiating pale shades of grey, tending to bleach out the finer details. The Epson has no difficulty in handling these subtle shades, including the pale greys on a white background found in the white petals on Margaret Stevens’ painting of the lily (see the detailed discussion elsewhere in this feature – Ed).

This is a reassuringly solid and reliable device which is stable in use and offers high-quality A3 scanning, happily importing into Photoshop the finest details and delicate shades of colour that I have struggled unsuccessfully to manage so far. I want one! (He's now got one – Ed.!)

Colour Performance (MMcN)

Such was the fanatical quest to optimise the scans from both the XL and the V750 that the review turned into an analysis of the merits of using either the Macbeth SG target or the more common IT8 target. McNamee has long harboured a suspicion that the IT8 target was not quite up to the standard of the Macbeth Colour Checker SG, despite earlier remarks about the differences between the SG and the 24-swatch Colour Checker (CC24). We ended up taking the actual CC24 we were using and building that data into a bespoke spread sheet to eliminate errors in the accuracy of the chart itself. In summary, the profile built using the IT8 target was inferior in all departments apart from the density of the darker colours which it computed more accurately (in the lightness component only). Crucially, we found on two separate tests that the IT8 was creating non-neutral neutral tones (!), that is the profiled greys were up to 5 Lab points green. IT8 targets are made from photographic paper prints and often shift pink in tungsten light. We are suspicious that the pink-biased metamerism is detected by the fluorescent tube of the scanner and then over-corrected towards the complementary colour, that is, towards green.

When correctly profiled, the XL produced very accurate colour across the whole area of the platen. We could just detect a variation from one side to the other but only on one or two swatches, it was essentially inside the measurement error of the Photoshop colour picker – that is the value should have been 50% Lab L and sometimes flicked to 51%, in other words trivial!The average error across the whole chart was 2.00 ΔE2000. By coincidence, one of the largest errors became the foliage green (6.8 ΔE Lab/3.73 ΔE2000). The largest was the dark skin swatch which lost out with a high error in saturation. Bear in mind though we are talking about small errors and looking at fine differences. Overall this was a top-rate performance from a scanner across the expanse of an A3 original. It is, however, unrealistic to expect to achieve the level of precision we were seeking for our reproduction without hand tweaking. To some extent this is a classical example of where the craft scores over the science – the science got us as close as it could, but still required manual intervention to proof out a final that was beyond reproach (and, it has to be said, of a quality that would go unacknowledged by many viewers). Perhaps an incident that occurred during the trials tells the full story. We spent some time carefully measuring a print and making copious notes only to discover, near the end of the process, that we were not working on the original, we had picked up a proof!

CAPTIONS: The data says it all. While both the IT8 profile build and the Macbeth SG build improve over a raw scan, the SG profile out-performs the IT8 in almost every quarter including the important neutrals. Comparing the two Macbeth Charts gives a visual comparison of the scan accuracy. The only real loser from the SG Chart is the dark skin patch. Overall though these are very good data, about as good as you can expect from a scanner of any quality.

Summary

If you need an A3 scanner look no further! With the difference in price between the XL and the XL Pro it is a moot point whether the additional cost is worthwhile; for the same money as the XL Pro you could have the XL and the Epson V750, which really gives you the best of all worlds. To some extent it depends on the amount of desk space you are prepared to devote to your scanning (or perhaps a need to scan up to 48 slides at a time). Either way, you will not be let down on flatbed performance by the XL; it gives some ground to the more modern V750 on 35mm scans.

Photo Quote: It is my intention to present - through the medium of photography - intuitive observations of the natural world which may have meaning to the spectators. - Ansel Adams