The Epson 2450 replaces my old three-pass scanner, which has seen better days. I was tempted to buy the Epson mainly on the grounds of specification and value for money. With a built-in 4" x 9" transparency unit (as well as slide and 35 mm holders), it has more to offer than many dedicated film scanners, but at a fraction of the cost. The technical specifications are indeed impressive for a consumer level flatbed scanner, claiming a resolution of 2400 ppi and support for USB 2 and Firewire. Whether this is a true optical resolution is not made very clear, however. A more comprehensive list of its features can be obtained from the Epson UK website.

Slide Positives
Although transparency scanning is not my primary use for the 2450, I thought it might be interesting to compare results with my Microtek 4000t.  Epson's Twain 5 software has a Configuration dialogue that can enable colour management, so first of all I chose the "Color" tab and selected the options shown below. The "Epson Standard" profile was the most obvious choice, although I have since created a custom profile for this scanner which works a lot better with reflective scans; more about that on page 2. Clearly, a generic or 'standard' profile is not likely to be accurate for all film profiles and for reflective as well as transmissive scan modes, but this seems to be the only type that Epson has supplied.

The next two images were scanned from a Kodak Elite Chrome slide (into Photoshop), first, using the Epson Twain 5 "Auto" exposure button, and then without it (for reasons which I hope will become apparent). The third scan was made with the 4000t, using Microtek's ScanWizard "Automatic" setting for dynamic range. No other adjustments were made, except to scan in 24 bit mode. Since I knew I would be resizing for the web, the Microtek scan was made at 1200 ppi - a lot quicker than using 4000 ppi. The Epson scans were made at 1600 ppi, but all three images were resampled during web optimization to standardize pixel dimensions to approximately 500 x 300 pixels.

Epson 2450: Auto Exposure

Epson 2450: Without Auto Exposure

Microtek 4000t

My initial impression was that the Epson made a reasonably accurate scan, the "Auto" exposure producing a slightly brighter image than the Microtek, but with a small shift towards red/magenta. This is particularly noticeable in the clouds, where there should be some neutral grey. Without "Auto" exposure the second, full range scan is rather flat, but the magenta cast is not as pronounced this time. 

The next set of images show a small detail area from the left bank of the river shown above. The Epson has not scanned as crisply as the 4000t (despite being scanned originally at a slightly higher resolution), and this is quite noticeable in the telegraph wires, which descend from top left to centre of the image detail. There also seems to be some evidence of red colour fringing on some of the masts in the first image. Optically, I believe this is known as chromatic aberration, or dispertion, where different wavelengths of light passing through the lens are not focused at precisely the same point. However, the effect is nowhere near as apparent in the second image, where Auto exposure was not used.

Epson Detail: With Auto

Epson Detail: No Auto

Microtek 4000t: Detail

The following shots, taken from the detail views above, show the differences more clearly. The image scanned using "Auto" exposure seems to have greatly exaggerated colour fringes, a pole in the centre revealing almost the the full spectrum of colours. Interestingly, the second image, from the full range scan, suggests that the problem is not as bad as it appears. Whatever it's doing, the AE facility in Epson's Twain software clearly cannot be trusted with this task. Fortunately, the problem can be resolved by not using this feature; but can this be said for the apparent lack of sharpness evident in the Epson scans?

2450 

2450: No Auto

4000t 

For good measure I have included two similar detail views scanned with my old Nikon LS20, one of the earlier dedicated 35mm film scanners. This machine is by no means perfect; it has always had difficulty handling the shadow areas of slides and some colour fringing is present (no doubt emphasized by the red colour imbalance), but nevertheless, it has still managed a slightly sharper scan than the Epson 2450 Photo.

Nikon LS20 Detail 1

LS20 Detail 2

Epson 2450

Microtek 4000t

In contrast to the first slide comparison,  I cannot detect any significant difference in the sharpness of these two images. The result seems wholly satisfactory to my eye, so I must confess to being at a loss when it comes to explaining the discrepancy with the first slide. I have tried scanning it again, making adjustments with 'Image control' in Epson's Twain 5 software. Better control of tonal range and exposure is possible this way, and improved resolution of detail was achieved. However, fine detail still appeared relatively 'soft' to me. It occurs to me that the thickness of the slide mount may be a factor here, and that a sharper scan may be possible if the film is placed directly on the flatbed, but I have not tried this approach. Be prepared for Newton rings (next page) if you do try it.