EPSON Scan settings: which Gamma setting is "normal" ?

[Ken Lee]

The EPSON scan software's "Histogram Adjustment" allows you to change the Gamma as well as the endpoints.

Would 1.0 be a normal Gamma ? What setting will result in a "linear" interpretation of the negative ?

How does this relate to the option of using either 1.8 or 2.2 as the Gamma, in the general settings ?

http://www.kenleegallery.com/images/forum/AdjustedHistogram.jpg

[Henry Ambrose]

Hard to say for sure about the Epson setting as we don't know for sure what encoding is set behind the Epson curtain. My guess is that "1" in the box above would be no gamma adjustment. A standard setting is 1.8 to 2.2. But again we don't know for sure if there is some magic being done that Epson doesn't tell us about. Gamma is used to adjust the luminance values of the machine image to human vision.

In settings such as above, obviously the film densities are going to have a large influence on what is captured, so I don't think there is a set answer that fits for all negatives, rather an individual answer for each. Just like adjusting the negative to fit your darkroom paper, its a good idea to adjust the negative to your scanner through a personal testing process. On the other hand once you get all the information into the computer you do have a good bit of lee-way in distributing the values.

I personally think that you're almost always better off getting a full luminance range on input. In the above example that would look like a full histogram window with values from left to right edges and not a clump in the middle as shown. Then gamma is applied to adjust the values of the image to vision. And your intention plays a big part in all this. You may have a completely different way of seeing the picture in your mind from what someone else might choose -- your correction might be very different from mine.

To your specific questions:

"1" should be no gamma and linear, 18 or 2.2 would be "normal".
General settings of what? If in Epson scan, play with the settings and look at the changes in the "Tone Curve Viewer" in the screen you supply above. That shows any changes in the adjustment being made.

[Greg Miller]

The Gamma in this case is a relative number (1.0) used to express the midpoint between the black point and the white point (no relationship to gamma discussion of 1.8 or 2.2).

When using the default black and white points (0, 255), a gamma of 1.0 would translate to a midpoint value of 128. But as you change the black and white points, the actual value would also change. The 1.0 is simply a relative way to express the midpoint value, without having to see it change every time you move a black or white point - 1.0 will always represent the midpoint. You will also notice that it is logarithmic, with a range of .1 to 9.99 where 1.0 is the midpoint.

[Ken Lee]

Perhaps the best approach is to scan a step-wedge and see what happens as different adjustments are made.

Have people done this already ?

[Greg Miller]

What is it that you are trying to accomplish? Your step wedge results will be different every time you change the black point, white point or gamma.

The gamma simply represents the midpoint between the black point and white point. It is a relative number, not an absolute number. The black point and white points are absolute numbers. Changing the gamma simply makes the midpoint brighter or darker (it has no effect on the black point or white point), but it is directly related to the black point value and white point value.

[Ken Lee]

Thanks for your help.

I'm trying to scan a negative in the most linear fashion, so I can compare the results of film speed and development.

I'm trying to view the negative without the scanner imposing its own curve on the result.

[D. Bryant]

Thanks for your help.

I'm trying to scan a negative in the most linear fashion, so I can compare the results of film speed and development.

I'm trying to view the negative without the scanner imposing its own curve on the result.

Leave the gamma setting alone. This goes double when you are scanning printed paper targets for calibration of digital negatives. See Mark Nelson's manual about scan settings since you are a PDN user.

Don

[Ben Syverson]

1.0 is unity—no change. The 1.8 / 2.2 setting relates to how images are displayed on your monitor, not how they're scanned and saved to disk.

[mdm]

On vuescan you have the option to lock settings, so scan both negatives with the same settings and compare the files in photoshop.

[Henry Ambrose]

Thanks for your help.

I'm trying to scan a negative in the most linear fashion, so I can compare the results of film speed and development.

I'm trying to view the negative without the scanner imposing its own curve on the result.

When comparing negatives, pull the end points out and have the middle slider set at the midpoint 128. In other words, ignore the contents of the histogram and let the negative drop in where it falls. The scanner should hold this each subsequent scan if you save it as a setting.

[Ken Lee]

When comparing negatives, pull the end points out and have the middle slider set at the midpoint 128. In other words, ignore the contents of the histogram and let the negative drop in where it falls. The scanner should hold this each subsequent scan if you save it as a setting.

Can you specify a number in the range of 1-255 in the "gamma" input box of the EPSON Scan software's "Histogram Adjustment" tool ? (see screen shot in post # 1)

[Steve Wadlington]

Ken,

I have a Epson scanner and a test awhile back on Stouffer 4x5 step tablet (transparent). That density range is really pushing past the limits of the design of the scanner but I was suprised to find that 1.0G that I always scanned negs at wasn't the most linear range on the scanner. I'm away from home at the moment but best I remember it was about 1.5 to 1.6.
I graphed out on and Excel spreadsheet to include all the densities of the tablet, mind you it wasn't linear on the high or low end, but all values where on the scan.

[Ken Lee]

Thanks for that. I intend to do the same and plot the results. I think it's very helpful to know.

If we profile a scanner, does it automatically work in a linear fashion ?

if so, then how do we profile the scanner ?

[Greg Miller]

Can you specify a number in the range of 1-255 in the "gamma" input box of the EPSON Scan software's "Histogram Adjustment" tool ? (see screen shot in post # 1)

No. But if you do precisely as Henry suggests (leaving black and white points at far ends of the scale) then a gamma of 1.0 = 128.

I personally would rather move the black and white points just outside the data in the histogram to allow the tonal gradations to be mapped over a larger mathematical base, but there are 1,000 ways to scan a cat.

[Ken Lee]

No. But if you do precisely as Henry suggests (leaving black and white points at far ends of the scale) then a gamma of 1.0 = 128.

I personally would rather move the black and white points just outside the data in the histogram to allow the tonal gradations to be mapped over a larger mathematical base, but there are 1,000 ways to scan a cat.

Excellent - Thank you for the clarification.

I always adjust the bottom of the histogram to render the black film edge as 0. However, not every photograph has a 100% white value, so IMHO it's best to adjust the high end according to... taste.

[Steve Wadlington]

Ken,

"Thanks for that. I intend to do the same and plot the results. I think it's very helpful to know.

If we profile a scanner, does it automatically work in a linear fashion ?

if so, then how do we profile the scanner ?"

I was thinking of setting up a Photoshop curve to correct a scan to linear, but that may not necessarily be the best. I was trying to figure what the absolute limits of the scanner was. I have had problems rendering the very high values in clouds and such.

I shoot tmax and use pyrocat hd as I believe you do. Wish I had more time to shoot.

[Daniel Moore]

The instructions for setting up linear output from Epson Scan (http://www.colorneg.com/scanning_slides_and_negatives/scans/Epson/Perfection/Epson_Scan/) for Colorperfect may be of interest to you.

[Ken Lee]

The page to which you linked, gives instructions on using their product Color Perfect and Color Scan.

They point out that what comes out of the scanner software, is Gamma-encoded to Gamma 2.2 on Windows, and 1.8 on Mac - and that they have a product which will linearize the image. I presume they apply a complimentary curve.

As far as I could tell, their product operates "down-stream" from the EPSON driver. In other words it's doing post-processing to correct some limitations inherent in the EPSON driver.

This suggests that the EPSON Scan driver can not be configured with respect to Gamma.

When I scan on a Mac, my 16-bit grayscale images come out Gamma 1.8 encoded. I normally convert them a Gamma 2.2, since that's how my monitor is calibrated. The 1.8 standard is pretty much obsolete at this point.

[Daniel Moore]

Reading further on I see this about Vuescan: "The normal Tiff files however are images edited and gamma encoded by VueScan while those of the RAW output mode are of linear gamma and are always equivalent to the light intensities read in by the scanner."

[Ken Lee]

Reading further on I see this about Vuescan: "The normal Tiff files however are images edited and gamma encoded by VueScan while those of the RAW output mode are of linear gamma and are always equivalent to the light intensities read in by the scanner."

Now you're talkin' - Thank you !

[D. Bryant]

Now you're talkin' - Thank you !

Just take note of this snippet for downstream processing.

The right hand side Gamma pull down is the output encoding (Gamma C) which must be set according to the working space you prefer to use. ColorPerfect saves these values so that you will usually have to set them up but once - unless you work with images in multiple color spaces or from different scanners.

ProPhoto is 1.8 gamma - Adobe RGB is 2.2.

And this by Andrew Rodney:

"If you are printing with Epson ABW then

It wouldn’t be a bad idea but keep in mind that you could find a combo of ProPhoto’s 1.8 gamma and one in ABW that could work. The various settings (Dark, Darker, etc) are actually different gamma settings. Most stick with a 2.2 gamma and use Darker (the default) but you could probably come up with a setting other than that which could work with 1.8. If you don’t want to futz around testing, use Adobe RGB and start with the default (Darker)."

If you are creating digital negatives using PDN you need to use Adobe RGB 1998 color space which is gamma 2.2.

If you are using Epson ABW for digital negs (as Ron Reeder teaches in his workshops) then depending upon your settings listed above will mitigate your gamma setting. But if you use Reeder's method use 1.8.

If you are using QTR I would stick with gamma 2.2

The native gamma for displays for Windows and Macs is 2.2.

Confusing isn't it! :(

[Ken Lee]

It's confusing, and like a lot of this stuff requires study.

My view (for better or worse) is that less digital manipulation translates to a more analog image. That's why I try to get things corrected as far up-stream as possible. At exposure time is best, then development, then scanning, etc. Each is a step down.

The histogram may not show it, but it's a bit like someone who's had too much plastic surgery: all the parts are there, in the right place, but something looks... fake.

As soon as I can, I'm going to explore Vue Scan "raw" mode. I note that there's nothing called Gamma in VueScan. :rolleyes:

[D. Bryant]

Ken,

You may wish to get a copy of the Vuescan Bible. You can download a free Kindle version from Amazon.

[Ken Lee]

Thank you !!

[John Rodriguez]

Ken, check out this blog on testing between the linear "raw" scan as directed by ColorNeg, versus scanning as a positive - http://cjeastwd.blogspot.com/2008/10/epson-3200-more-optimal-scans.html.

[Ken Lee]

With EPSON Scan, there is an option called "Configuration". On that window, you can specify "No Color Correction". When "No Color Correction" is chosen, all the controls by which you could adjust the image, are disabled. The result is a scan that matches what VueScan gives when the curves are set to zero and generic film is selected.

I scanned a Stouffer Step wedge, and my scanner shows a range of around 16 steps, each of which is 1/2 stop. That's an 8-stop range, which in log 10 is 2.4

Epson claims a dMax of 4.0 which equates to around 13 stops. Not this machine anyhow. But then they claim a lot more dpi than any of us have measured too.

[Henry Ambrose]

You're going to get your adjustments moved "as far upstream" as you can by scanning at 16 bit with appropriate adjustments to gamma in the scanning software. I don't think you will find any advantage to bringing a linear file into PS and doing it there. For normal scenes, you want a big fat full-width histogram - then adjust from there. Whether the gamma adjustment is made in scanner software or in PS, its the same end result. I think what you don't want to do is to make repetitive and cumulative adjustments. Hitting the right look right out of the scan is the best route to good files and the least amount of work/time.

When this thread started I thought you were looking for a way to use your scanner as a densitometer for film testing and to optimize your exposure and development for the scanner. Maybe you're looking for something else/more now or I misunderstood?

My suggestion: Find a normal development for a normal scene and a normal scanner setting that gives the big fat histogram. Save that as (you guessed it) "Normal". All your normal film gets scanned with this saved setting. Then build same for N+1, then N-1. Go as far as you wish with this but that's about as far as you'll need for most situations. This is as close as you can get to "profiling" or "calibrating" your scanning system for B&W film. "Normal" saved setting is your "profile" for normal film and so on for the other saved settings and the exposure/development scheme associated with them.

[Ken Lee]

Thanks. I'm sorry if my statements were contradictory or confused.

I asked because I want to be able to see what a negative "looks like" with minimal distortion, the equivalent of a proof image, a contact print on Number 2 paper. I've "scanned" my negatives, but never really proofed them. Going forward, I want to be able to assess an image with no "correction".

Rather than using my scanner as a densitometer, I have purchased a (used) reflection densitometer, and a (used) transmission densitometer. Densitometers allow us to make direct, rather than relative readings.

That aside, I would rather correct the tonal scale of an image once rather than 2 or more times - although it's not always possible. Ideally, it's with exposure and development, but next comes the scanning phase. A subsequent stage is in the image editor. Even later is in the printer driver I guess.

[peter ramm]

Thanks. I'm sorry if my statements were contradictory or confused.


Rather than using my scanner as a densitometer, I have purchased a (used) reflection densitometer, and a (used) transmission densitometer. Densitometers allow us to make direct, rather than relative readings.

That aside, I would rather correct the tonal scale of an image once rather than 2 or more times - although it's not always possible. Ideally, it's with exposure and development, but next comes the scanning phase. A subsequent stage is in the image editor. Even later is in the printer driver I guess.

Ken, getting deep into film densitometry is admirable, but sort of like tasting from the tree of life. It may drag you from photography eden, where your eye is the judge of what you do, into a bleak world of instrumentation and standards where precision and reproducibility become the key requirements. Really, do you want settings on the scanner to alter the visibility of shadows on an MRI? Ideally, one does not "correct" an image at a primary stage. One acquires it with high fidelity.

What scanning densitometers (and drum scanners) give you is a precise and reproducible rendering of the film. I think that with 14-16 bits over a 4D range, you have the ability to render tonal response across a fairly broad range of exposures, without introducing discontinuities into the histogram and looking "fake" to the eye. However, skilled operators do tweak the scan to help poor exposures. That is like a white lie. It is theoretically evil but may be necessary. The worse the scanner or the exposure, the more likely one needs to do this.

If you have only about a 2D linear range to work with (Epson), it becomes necessary to optimize its response for the specific exposure latitude that each film gives you. I think this is what you are getting at. One way is to adjust boundaries inwards in a process called windowing, accepting that everything outside of that range will clip. Sadly, area scanners (like CCD cameras or the Epson) suffer from flare that makes this process less linear than it should be.

The alternative is to alter input gamma, assigning more precision to the parts you want but without clipping. Both windowing and input gamma will result in compression of some parts of the data range. At that point the digital image ceases to be archival, and becomes an interpretation of what you think the original looked like.

Pretty simple, really. You can have any two of low cost, precise tonality and broad response range. My impression is that you want all three.

[Ken Lee]

Thanks for the warning :)

Sorry if any confusion persists. I have 2 requirements, loosely coupled, but not identical:

One is to see what's on a negative, or compare several in a way that is as objective as possible - akin to a proof sheet. For that, I want the scanner to be as unobtrusive as it can be.

The other requirement is to make a nice photo. I personally like to perform as few corrections as possible, so I do the first one (the "white lie" as you called it) while scanning. That "white lie" scan is not the same scan as a low-res proof (perhaps that's where confusion has arisen).

Because I possess the negative, and name each file accordingly, the two scans don't get confused. Hopefully, in learning more about the underlying Science, the Art won't get obscured. While it's good to avoid extremes, it's also true that for many of us, the Art and Science are complementary, and their interaction is fascinating. If not, we probably wouldn't be making photographs, but working with simpler materials.