Observations on scanning software

[mrladewig]

These are some observations from my experience with just a couple components of scanning software. I haven't tried Vuescan and I know it has a good reputation for many scanners. My experience is limited to color material in a Nikon LS-4000 and Epson 4990 with NikonScan, EpsonScan and Silverfast Ai 6.6.

About a year ago, I noticed there were some weird artifacts in my color negative images when scanning with Silverfast Ai on my Epson 4990. Basically all my color negative came out of the software looking like a watercolor filter had been applied. The colors were blocked and there was little if any gradation. I'd picked up a Nikon LS-4000 and scoffed at Silverfast's outrageous pricing on that model and had been scanning with NikonScan and noticed I wasn't seeing this in the Nikon scans.

So then I decided to take a look at EpsonScan more seriously and see if I could get better color negative results. I will say up front that scanning color negative in EpsonScan professional mode requires much more user input, an understanding or RGB and CMYK color corrections and patience to work with curves in individual color channels. However, on my first scan, I immediately noticed a difference in the quality of color gradation with color negative and was overall quite happy with the controls offered. I tried scanning a slide film in the software, like the result and never looked back. My paid license of silverfast Ai sat on the shelf.

The only downside to EpsonScan was that the on-screen previews weren't entirely accurate, so some fine color correction is required in photoshop and that it doesn't seem to correctly scan into the aRGB(1998) color space even when the settings are configured for this. I was seeing some subtle shifts in certain colors and over-exuberant greens. This seems to affect color negative scans more acutely than positive scans, but both can be impacted. The software seems to be creating an sRGB scan, then tagging it aRGB and calling it a day. Easy fix- assign color profile sRGB then convert to aRGB in Photoshop ACE.

More recently I'd noticed that my current scans seemed sharper than my old 4X5 scans done in silverfast, and I decided to take another look at this. I needed to get some old film out for a project and decided to invest a bit of time in this.

These are all 100% crops of 2400 ppi scans of slide film. All are straight out of the scan software as far as sharpness is concerned. I've done some color correction to the middle one as it acquired a bit of a red cast in the scanner.

One of my first scans done by a friend on his 4870 just before I bought my 4990.






Now in this case, both were scanned on my 4990, both are the same frame of film. This was actually a case where the silverfast watercolor effect was showing up in a slide scan and I wanted smoother detail in the grasses. The yellow leaves really stood out to me in this example.

[mrladewig]

Continued...
This scan, the same piece of film was in the holder for both scans. I scanned, and then switched software and scanned again (with time between so heat didn't warp the film). Silverfast indicated that I had plenty of highlight room left when I set levels. Obviously not, I should have checked this more carefully in the densitomiter. I tried silverfast's USM turned up a bit. I've always turned Epson's USM down from its default. To my eye, the Epson scans seems to keep better color separation than the Silverfast scan.



significant photoshop work done here to improve on the flat appearance of Astia including luminosity masked curve layers, and saturation adjustments.




So then I started to wonder what EpsonScan on my 4990 could produce against a drum scan I've got on hand. I'll post that comparison later. I was surprised by the result. I have a Betterscan wet/dry mount holder on the way (finally-should have bought this a LONG time ago) and I've very excited by the prospects of wet mounting on my 4990.

[David Luttmann]

No offense, but posting one sample with sharpening enabled, and the other with sharpening disabled, hardly gives an honest impression of differences between the two.

If you run each without sharpening, then apply USM, you will not see any difference in detail.

[SW Rick]

Thanks for posting this! EpsonScan isn't a bad program, except for the interface. Do you generally scan @ 2400? I've been using higher resolution lately, thinking I could get more info out of the film- have read pros and cons here and on APUG. Sandy King's recommendation to scan higher moved me in that direction. Certainly no problems with what I see here @ 2400. I have Vuescan (as well as the SF which came with it, and the EpsonScan), and sometimes it's the best, other times not. Haven't been able to figure out the "why".
You're right about the Betterscanning holder- I have the dry holder. You'd think Epson would have a decent holder after all the years and previous models, since this is so important to getting a good scan.

[mrladewig]

On the third image, I set USM in EpsonScan to medium. In Silverfast, it was set to USM - 50%. I don't know if they're equal of not. What really bothers me in silverfast is this.



In Epsonscan, I just get smooth continuous tone. I see this same thing happening in the transparenct files, just to a lesser degree. Its not a matter of USM, its a difference in how the software interprets the scanner output. This image was the one that made me take notice of how silverfast was handling color. This is a particularly bad section, but it looked like this throughout the image.

This was Fuji NPS by the way.

[Chris Strobel]

I have a Betterscan wet/dry mount holder on the way (finally-should have bought this a LONG time ago) and I've very excited by the prospects of wet mounting on my 4990.

First off, VERY nice pictures!Number one and three look outstanding.Now as far as the prospects of wet mounting on your 4990, I too was very excited by the prospects of wet mounting on my 4990, so I bought a whole wetmount turnkey kit for both 8x10 and 4x5 from Scanscience.I would have gone Betterscanning, but they didn't offer 8x10 then.Anyway after loads of test and pixel peeping, neither my wife nor I could see any advantage to the wet mounted 4x5 scans versus dry on our 4990, nor could we make any rational decisions on the focus going from a neg placed directly on the platen up to the full stack of shims.They just all look the same at 100% to 400% zoom in CS3.So I have gone back to using the stock epson 4x5 holder with dry 4x5 negs, and only use the wetmount on 8x10 to avoid newton rings which it does quite well.It will be interesting to see what you come up with with your color work.We only do b&w at this point.Please keep us posted, and thanks for sharing.

[mrladewig]

BTW, this is a sample from EpsonScan with another sheet of NPS. I wish I had the same image above from both EpsonScan and Silverfast. The difference was striking.

[Jack Dahlgren]

I always turn the epson scan in-scanner sharpening off. It looks a bit harsh on some of your scans.

[Don Hutton]

Unless you want to spend a bunch of time learning about sharpening, I'd recommend scanning everything with sharpening turned off and use a utility like NIK sharpener, which has nice starting points for capture and output sharpening. IMO comparing scans objectively is tough enough without the added confusion of capture sharpening.

[gnuyork]

Another good utility sharpener is PhotoKit Sharpener. It also has nice starting points for capture and output sharpening.

I have done similar comparisons (with default sharpening off) with Epson Scan and Silverfast on the the v750 and find very little or no difference in the scans. From the time I did those comparisons I have been just using the Epson Scan driver because I felt it was a lot easier to use than Silverfast.

That comparison was done with 35mm B&W and slides.