Epson V600 - Epson Scan good enough?

blockend · Dec 8, 2017

Alan Edward Klein said:
But the key is to have flat negatives.

Agreed. I put keepers through Photoshop Elements, mainly to clone out spots and stray hairs, and sometimes to tweak contrast. Grain is perceivable in flat negatives, in less flat ones it's mushy. From a 35mm image grain is my barometer of sharpness, but it shouldn't be confused with digital noise which occurs at lower resolutions, being perceivable at 2400dpi but not 3200 dpi. As you say, a flat negative is a baseline for all other technical judgements.

My work goes into printed books and the quality stands up to all but the best duotone printing (Steidl), and presumably drum scans. One problem scanning shows is increased noise in less exposed areas. An optical print renders such areas as a solid black, whereas scanning - like digital camera files - tries to find detail in those areas which manifests as ugly noise.

Mainecoonmaniac · Dec 8, 2017

Take a look at this article.

https://petapixel.com/2017/05/01/16000-photo-scanner-vs-500-scanner/

Ted Baker · Dec 8, 2017

tedr1 said:
Thank you. If I understand correctly one of the issues is linearity of conversion between linear (sensor output) and "non-linear" (display), perhaps because such conversions always include errors?

Not really, you can back and forth as many times as like, because your dealing with integers nothing is lost. Switching from 16bit or 14bit to 8bit introduces errors but that is something different.

shutterfinger · Dec 8, 2017

Take 3rd party scanner reviews with a moderate grain of salt. They use a 1951 USAF film resolution test target to base their resolution on. There are several versions of this target that range in price from $270 to $1450 at https://www.edmundoptics.com/test-targets/
other companies have different offerings. The 3rd party testers should be using ISO 12233 chart http://www.graphics.cornell.edu/~westin/misc/ISO_12233-reschart.pdf
The link in post 27 is a good real world comparison. Most state the resolution to be what they can see with their eyes.
For general looking to see if the image is OK as in my camera repair testing I use 1200 to 2400 dpi scan resolution, for my master file of keepers I use the manufacturer's stated optical resolution which for the V500, V550, V600, V700/V750, V800/V850 is 6400 dpi, above that is software interpolation. The higher resolution scan can be downsized without loss of detail or sharpness while a lower resolution scan cannot be upsized without loss of detail/sharpness.
Vuescan tries to be everything to all scanners buts fails. Its good for scanners that will not run on newer operating systems but not all scanners are supported. My Plustek 7200 will not run on Windows 10 and Vuescan does not have a driver for it. Check Vuescan's table of supported scanners before buying it.

Clicking on the Configuration box at the bottom of the EpsonScan Professional page gives the box on the right. Clicking on the Color tab and selecting No Color Correction gives a basic scan with no corrections. Continuous Auto Exposure resets any settings you have made every time you press Preview with Color Control selected. With Color Control selected the histogram is always clipped at each end, but with Continuous Auto Exposure deselected you can set the 5 adjustments on the settings page for the best scan for a group of negatives without them changing when you press preview or advance to the next frame.

Take a known good negative or chrome and play with the settings one at a time then in a combination of 2 settings then 3 or more at a time. Small changes are all that are needed especially with multiple adjustments in use. A good preview may not result in a good scan.
Take the time to learn the software and limits of each adjustment and problem negatives won't be such a a challenge and different software will be easier to learn.

blockend · Dec 9, 2017

shutterfinger said:
Take the time to learn the software and limits of each adjustment and problem negatives won't be such a a challenge and different software will be easier to learn.

I have a question Shutterfinger, you may be able to help. If you recall I was having problems opening the marquee lines, until I was told to untick that tiny thumbnail box. That fixed things. The other day I was scanning and ticked the thumbnail box again. Now scans are showing the entire 35mm negative plus black rebate edge up to sprocket holes. I'm happy with this, I always aim to shoot without cropping and the black edge suitably cropped adds something. That's mostly how I optically print. However I'm not sure how I arrived at this look as all I did was tick the thumbnail box which previously showed a tightly cropped image - no black rebate. Something changed in the interim and I'm not sure what it was!

Thanks for you practical contribution to scanning threads, they make a lot of sense.

Ted Baker · Dec 9, 2017

shutterfinger said:
...

Everything described here, can be done for better or for worse with other software, and at a later stage. By all means choose the software solution that suits you best.

PerTulip · Dec 9, 2017

Ted Baker said:
Everything described here, can be done for better or for worse with other software, and at a later stage. By all means choose the software solution that suits you best.

That's why I want to get the best data possible when scanning. I don't need a software package that delivers a final product (Silverfast?), since, as you correctly state, everything can be done afterwards. In fact, my only concern is if I am getting better data ("resulution", dynamic range) using something other than the included Epson Scan.

Alan Edward Klein · Dec 9, 2017

Ted Baker said:
Or you can apply a simple mathematical function to a "non-linear" scan and get back to exactly the same "linear" scan you would have gotten in the first place... The only exception is when clipping has occurred.

Could you explain what you said?

shutterfinger · Dec 9, 2017

blockend said:
I have a question Shutterfinger

I don't have a clue other than the thumbnail size had been set by a customization and deselecting then reselecting it reset it to default or a different customization.
I turned the scanner on, opened EpsonScan, and made the screen shot for another post weeks ago. I usually deselect the thumbnail when scanning.

shutterfinger · Dec 9, 2017

PerTulip said:
In fact, my only concern is if I am getting better data ("resulution", dynamic range) using something other than the included Epson Scan.

The dynamic range and scan resolution limits are the hardware (lens, mirrors, sensor) limits of the scanner. The resolution limit is the Optical resolution as stated by the manufacturer, above that is upscaling software interpolation.
The dynamic range is the scanners Dmax a stated by the manufacturer.

PerTulip · Dec 9, 2017

shutterfinger said:
The dynamic range and scan resolution limits are the hardware (lens, mirrors, sensor) limits of the scanner. The resolution limit is the Optical resolution as stated by the manufacturer, above that is upscaling software interpolation.
The dynamic range is the scanners Dmax a stated by the manufacturer.

I understand that. But the hardware is controlled by firmware+software and it woudln't be the first application where improvement is possible. If Epson Scan is already getting results limited by hardware only, fine with me.

Ted Baker · Dec 9, 2017

Alan Edward Klein said:
Could you explain what you said?

Will try, but it may just may just result in more questions... If the data is saved non-linear, which is a common default. (jpeg, tiff etc) then a simplified version of the formula to save the data is Y=X^(1/2.2), where X is the original value and Y is the value to save in the file. e.g. 18% becomes 45%. You can then apply the reverse which is Y=X^2.2 and you are right back were you started from.

If that is just gobbledygook, that's fine, the point being pretty much all imaging software is just math, and many operations are in fact reversible, but not all.

shutterfinger · Dec 9, 2017

PerTulip said:
I understand that. But the hardware is controlled by firmware+software and it woudln't be the first application where improvement is possible

Firmware only tells the hardware how to operate and other software how to interpret it. I have never encountered a firmware update for a scanner.
Scan software improves on the adjustment of the data from the scanner or the interpolation above manufacturers optical limit.

JOENAME · Dec 9, 2017

I've ran the gamut with scanners. I had a super nice Nikon LS-8000, but had to sell it for financial reasons. I sure wish I still had it. Since then, I've had a couple of Epson flatbeds, a 4990 and a V750. They worked great for what I needed at the time, but I was doing my enlargements in a darkroom and just scanning the negatives for viewing / choosing. Since then, I've been shooting mostly 35mm and just bought a Plustec 8100 - works perfectly for me. But I scan only black and white and have only used Vuescan. I didn't even load the software for either Epson flatbed or the Plustec. Vuescan has a feature, "multiple scan" where you can scan the image twice at different exposures and the software put the image together with the best overall exposure. You get a very flat bright image, but all the data is there in post for editing. Now my darkroom is gone (moving) and may not set up a new one because of the quality I can get with scanning and printing. I still choose to print with labs that print silver gelatin. Maybe being color blind, I'm more sensitive to the color cast. They look like they reflect green light to me. Go figure.

Alan Edward Klein · Dec 10, 2017

Ted Baker said:
Will try, but it may just may just result in more questions... If the data is saved non-linear, which is a common default. (jpeg, tiff etc) then a simplified version of the formula to save the data is Y=X^(1/2.2), where X is the original value and Y is the value to save in the file. e.g. 18% becomes 45%. You can then apply the reverse which is Y=X^2.2 and you are right back were you started from.

If that is just gobbledygook, that's fine, the point being pretty much all imaging software is just math, and many operations are in fact reversible, but not all.

I don't understand how you applied percentages to the scan machine or post scan processing software. Please explain.

Alan Edward Klein · Dec 10, 2017

Some have claimed that adjusting the white and black points in Epsonscan stretches out the range of the scan to provide initial data that is better than allowing the scanner to scan "flat" and doing it in post. Any thoughts on this?

Frank53 · Dec 10, 2017

Alan Edward Klein said:
Some have claimed that adjusting the white and black points in Epsonscan stretches out the range of the scan to provide initial data that is better than allowing the scanner to scan "flat" and doing it in post. Any thoughts on this?

Adjusting black and white points is the only thing you should do in the scanner software and the only way to get all information available in the negative in the scan.
In a colour negative you should do this for all three colours.
Regards
Frank

Ted Baker · Dec 10, 2017

Alan Edward Klein said:
I don't understand how you applied percentages to the scan machine or post scan processing software. Please explain.

The percentages are just examples, there must a shade that matches 18%, some of the time, but when it does exist it will be saved as 45%. Jpeg is 8bit 45% or 256 is so that comes out to be 115. Anyway these scanners have a 14bit A/D AFAIK so what you get is a integer between 0 and 16383 per channel, everything from that point on is math wrapped in software. What the software does is obviously important but it can be done differently from that point on, and is in fact done differently, which is the point...

Alan Edward Klein said:
Some have claimed that adjusting the white and black points in Epsonscan stretches out the range of the scan to provide initial data that is better than allowing the scanner to scan "flat" and doing it in post. Any thoughts on this?

Adjusting the analog gain, is the only thing that I can think of that would make any difference, with very dense images, so if your a negative user you can probably ignore that.

Frank53 said:
Adjusting black and white points is the only thing you should do in the scanner software and the only way to get all information available

That is not the only way IMHO, but it is one way.

shutterfinger · Dec 10, 2017

Alan Edward Klein said:
Some have claimed that adjusting the white and black points in Epsonscan stretches out the range of the scan to provide initial data that is better than allowing the scanner to scan "flat" and doing it in post. Any thoughts on this?

With Auto Exposure turned on EpsonScan sets the White and Black points inside the range of usable detail a small amount to increase the contrast in the main part of the image.
With auto exposure turned off you can set the points as you wish and with color control turned off (as in post 29) the white is set to maximum (255) and black is set to minimum (0) with the mid point the operating system gamma..
This gives the full range of tones in the negative, not the best image that can be made from it.
This example was scanned on a V500 @ 6400 dpi 48 bit color (16 bit RGB per channel), with color correction off.
The master file is 1.66gb, it has been reduced to 150 dpi in PS Bicubic Sharper with its histogram in Levels captured with Advanced System Care's built in screen shot saved as a jpeg. No other editing.

Vigilant six-20, 105mm f6.3 Kodak Anastigmat, #1 Diomatic shutter, Arista EDU Ultra 100 in HC110 for 6.5 minutes at 68°F.
Smoke from the Northern California wild fires 75 miles north of the photo's location. It got worse two days later.
The example is both over exposed and over developed.
The rise above the base line is usable information, correctly exposed negatives start the information rise between 0 and 10, this one starts at 0. The fall to 0 beyond 255 is over developing. Normal development with whites correctly exposed drop to 0 between 225 and 255. Over exposure is 1/4 to 1/3 stop, over development is around 10%.

Alan Edward Klein · Dec 11, 2017

Frank53 said:
Adjusting black and white points is the only thing you should do in the scanner software and the only way to get all information available in the negative in the scan.
In a colour negative you should do this for all three colours.
Regards
Frank

What IO believe I see in my scans is that with Color Correction of, the scan will cover the full range. The resultant picture will have a range on the lower end of the histogram - from let's sy 10 to 180. Then in LR or Elements, I expand the range to 0-255 using the points or Levels.. The result gets me about 95% of the corrections and I tweak from there.

My questions is do I really get more data by pre-adjusting the white and black points (levels) to 0-255 for the scan. Or am I already getting the same bits of data anyway since the range falls into it with as per my example of 10-180? (Note that the scan file size is the same in both cases).

Frank53 · Dec 11, 2017

Alan Edward Klein said:
What IO believe I see in my scans is that with Color Correction of, the scan will cover the full range. The resultant picture will have a range on the lower end of the histogram - from let's sy 10 to 180. Then in LR or Elements, I expand the range to 0-255 using the points or Levels.. The result gets me about 95% of the corrections and I tweak from there.

My questions is do I really get more data by pre-adjusting the white and black points (levels) to 0-255 for the scan. Or am I already getting the same bits of data anyway since the range falls into it with as per my example of 10-180? (Note that the scan file size is the same in both cases).

I set black and white points before scanning, because that’s the way I learned it years ago. The argument that it’s no use wasting data on levels that do not even exist in the image, made sense to me.
But of course you’re free to do it differently.
Regards,
Frank

farmersteve · Dec 11, 2017

I've had the v600 for a few years and gave up on it for scanning anything smaller than 120. I moved to a DSLR setup with a macro lens and the speed and resolution are much better. It mostly collects dust now...

shutterfinger · Dec 11, 2017

Alan Edward Klein said:
My questions is do I really get more data by pre-adjusting the white and black points (levels) to 0-255 for the scan. Or am I already getting the same bits of data anyway since the range falls into it with as per my example of 10-180? (Note that the scan file size is the same in both cases).

Short answer is NO. If the image scans with black at 10 and white at 180 with no color correction then turning color correction On, and manually setting the black to 10 and the white to 180 or what ever the points that information on the negative starts and ends results in the same information from the negative.
By having color control off I can tell if I'm getting correct exposure and if my development is correct.
Here is another example: Vigilant six-20 100mm f8.8 Diomatic shutter that runs slow after CLA. Sunny 16 exposure on Arista EDU Ultra 100 in HC110 @ 68°F for 6.5 minutes.

Blacks start at 70.

Ted Baker · Dec 11, 2017

Alan Edward Klein said:
My questions is do I really get more data by pre-adjusting the white and black points (levels) to 0-255 for the scan.

If you stay in 16bit AND don't clip, then no difference what so ever. If you save the file as 8 bit then yes it does make a difference.

Alan Edward Klein · Dec 11, 2017

Ted Baker said:
If you stay in 16bit AND don't clip, then no difference what so ever. If you save the file as 8 bit then yes it does make a difference.

Why is there a difference between 8 and 16 bit?

Epson V600 - Epson Scan good enough?

Privacy & Transparency

Privacy & Transparency