Epson Scan No Color Correction same as other's RAW?

[Tim Lookingbill]

I have an Epson 4870 and have been curious about its No Color Correction-RAW setting and how it compares to what can be achieved using Vuescan or Silverfast AI.

I have Silverfast SE and it doesn't offer a RAW setting.

I scan negatives and notice Epson Scan's No Color Correction setting only delivers data in the sRGB space with negatives, only uncorrected for in color cast and B/W points.

Does the other software mentioned give different results in previews and data?

[Ted Harris]

One of the beauties of Silverfast Ai is that it gives you virtually total control over color management during the scan. The limits are those of the scanner and its firmware.

[Tim Lookingbill]

How can you tell with Epson Scan or any other scanner software using any setting RAW or Automatic that all the data in the media scanned was captured? Can you tell by the histogram in Photoshop and what do you look for?

[Ted Harris]

I don't use Epson scan with any regularity so I am not sure wht you mean by RAW settings but I can tell you that with Epson' sscanning software (at least the most current edition) or Microtek's or Silverfast you have several different ways to look at the information on the film and set your scan to capture all of it. If you look at the offered histogram or a thumbnail of the image you can set your white and black points, either visually or using the virtual densitometer to pull out max shadow detail and to set your point for specular white. The point is that you don't want to wait till you get to Photoshop or you 1)may have captured useless info or 2)left out important info. Once in PS you can't change the base info in terms of what is there and what is not.

Tom, you might be interested in taking one of our scanning workshop for a more detailed imersion in the process. See the details at www.fourpointlanding.com BTW the dates listed need to be changed, we are going to have to reschedule both workshops due to some construction and power problems both in Ohi and NH. Best guess for the moment with announcement in the next few days is the next workshop will be at Midwest on the weekend of Friday December 8, 2006.

[Tim Lookingbill]

Thanks, Ted.

Thanks for workshop tip, but I don't think I'll need it. I'm just trying to understand the behavior of the Epson Scan because I can get so many variations in previews using all the tools and techniques scanning negatives. It just doesn't feel like I'm capturing 48bit RAW data zooming in and editing in PS.

I'm comparing it to old 48bit RAW scans off an Agfa Arcus II which are very dark and soft but seem more beefier when applying edits. The edits behave more smoothly where as the Epson scans are kind of jumpy and abrupt when editing in PS.

I've been refining my NegPos technique using ICM/Colorsync set to Epson Standard as Source and Joseph Holmes RGB as Target turning off Auto Exposure with the Reset button. This resembles a RAW scan somewhat. I seem to get a more variety of colors in PS previews within textures like in fleshtones but a bit on the saturated side with more pronounced sheen of noise over using Epson Scan's edit tools which give more muted, smoother but monochromish feel to the color with less of a noisey feel. They produce such different previews in 48bit captures from each other that it's left me wondering what RAW actually is suppose to look like.

Comparing the histograms in PS from an Epson Scan edit tool image over the NegPos version are quite different as well. The NegPos version shows quite a lot of tiny spikes along the top edges of the hills and valleys in each channel before any edits are applied where as the Epson Scan finished version is much more smooth.

I do like the color and dynamic range achieved using NegPos with ICM/Colorsync. I just need to set correct end points for each channel in PS levels to tone down the saturation. It does produce a much more vibrant and dynamic image. I viewed the two versions of the same scan, the EpsonScan edit tool next to the NegPos and the differences are quite stark in appearance.

The thing is I thought the EpsonScan edit tool version which I did first was great until I viewed it next to the NegPos version and had to ask myself what was I thinking. I wish I could figure out what Epson is doing to the data to produce such a variety of previews and histograms.

Thanks for the exchange.

[Tim Lookingbill]

I'll give you an idea about what I'm talking about by posting the two scan versions. The EpsonScan tool edit version took me a lot longer than the NegPos edit with levels and curves in PS to arrive at the final image.

I think adaptation did a number on my eyes when editing in EpsonScan because it took so long and I still ended up with such a dull looking scan. But at the time I thought it was great looking.

The Epson is such a different (better) scanner from my old Agfa Arcus II and it's taking me a while to get a feel for it and how it works. It's also the first time scanning film as well.

[Doug Fisher]

>>I have Silverfast SE and it doesn't offer a RAW setting. <<

Tim, have you experimented with 48 bit HDR option in SE?

Doug
---
www.BetterScanning.com

[Tim Lookingbill]

Doug,

Yes, I went ahead and scanned in 48bit HDR with SilverfastSE. It disables all edit tools, so I'm left with Negafix's myriad of profiles that change the preview with varying color cast and density distribution. Negafix was pretty much a waste of time (I think) because the final tiff opens in PS as a negative tagged with Silverfast's canned tranny profile.

So I went ahead and just inverted and converted to JoeRGB and used levels and curves to get the final image which wasn't bad at all, but produced a noisey histogram as Epson Scan's with the same jerky edit behavior. The final PS preview looks nothing like Silverfast's. Check the screenshot below.

Comparing the SE with EpsonScan's NegPos version, I could've gotten the same results if I'ld increased the blue highlite in levels. I have to say though, the Silverfast negative preview had its mask already neutralized and tagged so I didn't have to correct like I did with Epson's NegPos. One less extra step, two if you include the assigning of the scanner profile.

Not sure when to invert and convert to working space but I think it came out alright, still much better than Epson Scan's Negative film setting and editing tools.

Is this SilverfastSE Negafix "positive preview scanning as negative" suppose to work this way or have I missed some buried setting within the software? I guess I'm going to have to download the latest update that says it fixes color managed previews.

[Henry Ambrose]

Epsons won't give a "raw" scan like what you could get on your Arcus. That it returns an sRGB file is a good clue that something is very wrong if you thought you'd get an untagged "raw" file out of the machine untouched by the Epson Scan driver. If its really in sRGB you'll never get anything good out of it no matter how long you work in PS.

I suggest you set Epson Scan in "Professional Mode" and go to "Configuration> Color" panel. Click "ColorSync", select "Epson Standard" as "Source (Scanner)", or if you have profiled your scanner then use that profile name in the "Source (Scanner)" box and your favorite working space as "Target".

Now you can draw and drag the selection around various parts of the frame to get various renditions of color and tone. Move the selection to an area and click the Auto button. Keep doing this until it looks right. When you get one you like, "Save" it at the top of the Epson Scan window in the "Settings" area. This will let you go back and use that setting on subsequent scans to maintain continuity for other film shot under same conditions. (like five sheets shot at same lighting conditions and you want them to look like they were indeed shot together - i.e. the wall paint color matches or the cabinets are the same color in each picture.)

What you'll get are good color reproductions at 48 bit into Photoshop in your choice of working space to fine tune. If you practice with this a bit you'll be getting good files of average scenes right from Epson Scan and spending only a short time in PS to make them great. Great meaning that you'll have excellent global color.

The only thing wrong with all this is that for 4X5 film Epson Scan craps out on you at 2400 ppi. Anything bigger than this and it breaks. For big scans from big film you have to use Silverfast. (roll film scanning seems to not have this bug). As far as I know Epson is not fixing this admitted bug. Ain't that nice?

[Ted Harris]

Why would you want to scan at over 2400 when you are already excceding the real resolution of the 4870 or 4990?

[Keith S. Walklet]

Henry, my results are slightly different. My 4870 won't scan 4x5 above 3200 spi (in transparency mode).

Haven't tried it with a negative. Is there a lower threshold?

Ted, I note marked incremental gains in the quality of the resulting files up to 4800 spi. After one moves beyond the claimed optical max spi into interpolated territory, they go south in a hurry.

Are you possibly confusing your measurements of the optical resolution with what Epson's marketing folks are calling it? Regardless of what the "true" resolution of the scanner is, the resulting files get better with each bump upward to their claimed optical limit of 4800 (something-or-others per inch) in my experience.

FWIW, I scan into Joe's Ektaspace, no sharpening, no ICE, 48 bit. Files are full gig at the 3200 spi resolution.

My workflow is slightly different than yours, too, Henry. Your suggestion for using the Auto feature seems a little less trouble than my approach, which involves using the virtual densitometer that Ted mentioned, plus multiple sample scans to make sure I'm getting full black and white ends of the data. It seems as though you're looking at a broader range of data than the densitometer is and could possibly zero in on the shadow and highlight limits more quickly. I'll have to give that a try.

The histogram in the version of EpsonScan I use is at best a generalization and I find I have to do the work to get the values set properly myself. I wish the software had a means to find those high and low values on its own. I know the third-party stuff does.

[Ted Harris]

Keith makes a good point about accurately finding the best white and black point. Silverfast Ai, for example, will show you were the points are after your prescan and then where they are after you have tinkered. You get both a histogram and a visual looksee.

[Henry Ambrose]

The workflow or method I wrote is what I'd call the simple intuitive approach. It requires no additional expense and its straightforward. You can look at the histogram after you click the "auto" button to see what its giving you. If you need to move the black or white point at that time you can, or apply curves if you want. I usually set the black and white points out a tiny bit so I don't clip off either end. For negatives that are "good" this method will give you a fine scan and its fast and easy.

I bought the full version of SilverFast so I could scan at higher resolutions. If I remember correctly, my 4990 goes buggy with Epson Scan at 2400 ppi on 4x5 film. I wanted that much (and more) resolution for big prints. Whatever the actual resolving ability of the scanner is measured or calculated, the prints look better when the scanning resolution is higher. In SilverFast I go back to using more of the driver's tools and I'm getting really nice scans. My clients were impressed and I was happy.

[Keith S. Walklet]

>The workflow or method I wrote is what I'd call the simple intuitive approach.

It does seem intuitive. I'll have to give it a try, though I don't trust the Epson histogram. I always open the file in PS to look at the histogram there. But at least EpsonScan has one. I know some very expensive scanners with software that don't.

>If I remember correctly, my 4990 goes buggy with Epson Scan at 2400 ppi on 4x5 film.

Again, I'm talking about transparencies, but was pretty sure they bumped the buffer size up so that it could handle 4x5 at 4800 spi on the 4990. But I turn off all the extra stuff. For instance, I can't even select the "dust removal" at the 3200 spi resolution. That larger buffer was the primary reason I contemplated moving up to the 4990 or V700 from my 4870. My thinking was that even if I didn't end up printing the image super large, with the larger scan file, I get the benefit of less noise by down-sampling.

>Whatever the actual resolving ability of the scanner is measured or calculated, the prints look better when the scanning resolution is higher.

I agree, up to 4800 spi. ;-)