Epson 4990 Second Impression

For what its worth, first impressions sometimes are wrong. Last week I praised the 4990 and my initial impression was that it was superior to the Microtek 1800f. Upon further analysis I find that I was considerably mistaken. The 4990 produces very sharp scans, perhaps a tad sharper than the Microtek, but it is significantly inferior in dmax. Part of the problem in my first impression is that I was using the native software of the Microtek, for whatever reason, could be error operator, the native software is not as good as Lasersoft Ai. This weekend I installed Lasersoft, which comes standard witht he 1800f and the scan quality doubled in quality and comparing the 4990 and the 1800f but with Lasersoft, the 1800f produces much more shadow detail, richer and deeper colors and better color fidelity. Another problem in my use of the 1800f is that I had the scanner on top of a file cabinet that evidently was not sufficiently stable. So the vibration produced by 100mb scans introduced enough movement to make the scans soft. Placing the scanner on the floor considerably improved the scan sharpness. The moral of the story: new technology is hardly plug and play and there are so many variables that novices like myself can easily overlook. Many of the varialbes are never discussed on the internet and you just have to slug alone with trial and error. Bottom line: because I shot mostly slide film the Microtek stays and the Epson goes back to the store.

You don't say what difference, if any, using Silverfast with the Epson 4990 might have made. I have found limitations with the Epson software, and prefer using Vuescan.

In the earlier "First Impressions" thread, I posted a link to my own tests between a 4990 and a drum scanner here

http://www.photo-i.co.uk/BB/viewtopic.php?t=49

My view is still that the Epson does an excellent job. Software clearly plays a big part in the performance of any hardware. You might want to try using Vuescan with the 4990. I assume you have already tried Siverfast as that comes bundled with the 4990.

Quentin

Quentin, thanks for the link to that very informative thread. the atached scans really makes the 4990/vuescan/focal blade look like a viable substitue for all but the most important scans.

Quentin

I used Lasersoft with the Epson. Did you compare the Vuescan with Lasersoft? I think the Epson does a very good job for slides that are not too dense or contrasty, however, the difference in the dmax seems fairly clear when you use the Epson to try and pull out detail from these dense slides.
However, I don't profess any great expertise in scanning or post processing in photoshop, which both might greatly improve the final result of the Epson scans.

IMHO your two messages just exemplify the hazards of unscientific testing of equipment whose performance is affected by variables such as the software used with it and the knowledge and talent of the person operating it. It's good of you to post your experiences with these two scanners and I don't mean to criticize that, but since you didn't know how to use the Microtek to its best advantage why should we assume you're using the Epson to its best advantage?

I had the 4870 and returned it for the latest Canon flatbed (9950F) and I really prefer it. I am a newbie, but I just work better with the Canon. I cannot give any technical info other than I prefer the interface and the film holders (plus i can do 30 35mm's at a time)

Morey,

I quite liked Silverfast with the Epson, but I saw more noise in the shadows with it than with either Vuescan or Epson software, and that affected the ability to sharpen. I reached the stage when I felt I had gone far enough in testing; I'l still use the drum scanner (of course) for critical transparencies, but would happily use the Epson for a lot of work. Incidentally, Vuescan software does have an option to scan using more than one pass, whch in theory at least increases the dynamic range. I think Nikjon first did this with the LS2000 scanner.

Be aware that there is a buffer limitation with the 4990 that prevents it being used at full resolution for large format. Essentialy 2,400ppi ought to be enough to get 99% of data from a LF transparency in most cases, and my tests showed little would be gained with the 4990 scanning at 4,800 over 2,400.

Quentin

I'm using the Microtek i900 scanner, and have owned it for about 9 months now. I use it to scan 4 x 5 black and white negatives but I also shoot and have in my collection 8 x 10 negatives , when the scanner came on the market last year it was one of the few scanners or maybe the only scanner that could scan 8 x 10 negs for under a $1000 (cdn) which fit my budget, I think it does a good job, but then I am only printing out to about 16 x 20 inch size prints on my Epson 4000.

Gary
http://www.garynylander.com

Regarding VueScan with the 4990 the problem is no Digital ICE. This alone is enough to toss it in favor of Silverfast or even Epson Scan. The ICE works so well at removing all dust it's amazing.

I have spent a few weeks testing my 4990 now. I actually went to the trouble of having drum scans done of a couple of 4x5 b&w negatives and a couple of 4x5 transparencies for comparison. I even split them up and had two different companies do the drum scans. My impression is this:

If you are a very careful worker, that is if you take the time to get your film flat and really learn what Silverfast AI (or maybe Vuescan) can do (like with multiple passes to lesson shadow noise) and combine that with a few techniques like blending exposures (which is useful even with the drum scans for holding shadow detail); do very careful and workflow appropriate sharpening (which is also necessary with the drum scans) and don't print above 20x24 ( I can't actually print a 20x24 print that size on my 4000 but I can do a cropped piece of it); you can come close enough with the 4990 to a drum scan that they are virtually indistinguishable at this and smaller sizes. At 16x20 with a tiny touch more sharpening on the scan from the 4990 the prints are totally indistinguishable. By indistinguishable I am not talking about this ridiculous inkjet spin of "normal viewing distance". I look at inkjet prints the same way I look at silver prints-with my reading glasses on and my nose on the print!

My tests were done comparing Epson Ultrasmooth Fine Art paper and Hahnemuhle Ph0to Rag, both smooth mat surfaces (which I prefer for inkjet) with Epson Ultrachrome inks, 360 PPI file printed at 8 pass 1440 DPI. I suspect that because of the higher acutance (is that the right term?) of glossy and luster papers that the difference might become more apparent above a 16x20 print.

Epson has raised the D-Max of their flatbed scanners to 4.0 by interpolation which creates additional noise in the shadows. To realize this 4.0 a 16 pass scan with Silverfast AI (or Viewscan?) is required which all but eliminates the noise. I did not find a reduction in sharpness with 16 passes as reported by some. A plugin like Noise Ninja can help with that too.

At the price of drum scans the Epsons are very attractive technology. At $600.00 for the pro version mine will be paid for in less than one month!

It seems that many people reviewing scanners are comparing scans on their monitors.

The key is not what the scans look like on your monitor side by side at the beginning of the workflow, but what the images look like on paper side by side at the end of the workflow.