Silverfast users?

[Bormental]

Methinks there may be disappointment in your scanning future as for one not only is the Opticfilm already slower then the Coolscan, ICE is not as good and maybe hundreds of color negative scans later before we really know just how good it is.

There are two reasons I purchased it. First, I am hoping it to surpass my DSLR scanning in both the quality and speed, especially for medium format which is 100% of my color film usage.

Second, I am ready for a new hobby project, and building a custom scanning rig sounds about right. It's a perfect intersection of my professional work (hardware/software engineering) and my hobby (photography). I've done DAQ and DSP work before and I'm pretty good with embedded/systems programming. Having a brand-new "state of the art" reference point is cheaper than using my local lab.

Have you heard about the Nikon D850's built-in color negative conversion? That maybe a solution to your color negative DSLR scanning problem although I have yet to try it myself. BTW, what DSLR do you use?

AFAIK the Nikon kit doesn't do medium format and also it is JPEG-only and doesn't produce 12-bit output (RAW or TIFF). I have a couple of Canon 5D bodies, including the latest revision, with Canon 100mm macro and I also have a Fuji X-T3 with a macro lens. I have a copy stand with a Negative Supply 35mm film holder and my B&W workflow is already a "dream setup". But, as I said, for medium format color scanning... ughh.

As far as speed and quality on a scanner, back many many years ago, they used to hold a camera convention in Las Vegas and I got to try a Durst scanner system which gave similar output as a Coolscan at about 5 seconds per frame. I suppose it was an optimized motorized feed DSLR scanning system.

Right now I am very intrigued by two approaches: one is high-DR, high-resolution pixel-shifting camera sensor. Another is using a dedicated monochrome sensor but doing 3 exposures each with an R/G/B filter and merging them. I want to be in 100-150MP range. Linear CCDs used in scanners are flinstone tech, in my opinion. Not only they're slow and rely on physical motors to move them, they also do not have on-chip signal amplification like modern CMOS designs, so they can't compete on dynamic range as well. This is, of course, an outside observation. I haven't had access to dev kits and design specs.

I say it provided Coolscan results because at that show I took a strip of color negatives with me and I went around the different booths to have them scan a frame for me and the Durst and Coolscan results looked identical in resolution, color/contrast but of course no ICE. I remember it was about $35K then so . . .

For my purposes, I consider ICE completely useless. I don't have any problems with dust. My film handling pipeline is clean, and I just do not see much dust in my DSLR scans. Ok, maybe 2-3 specks per roll? Dust is less than 1% of my total scanning time. I suspect ICE to be a bigger concern for scanners as they tend to accumulate dust internally, i.e. a self-induced problem. I do not have it.

Also, I understand that Noritsu scanners are now getting to be more available at reasonable prices, I believe depending on the resolution setting, it can pump out scans at a very high rate.

This is true for 35mm but less true for medium format, IIRC that model is HS1800? They're over 10K on eBay.

 

[Les Sarile]

There are two reasons I purchased it. First, I am hoping it to surpass my DSLR scanning in both the quality and speed, especially for medium format which is 100% of my color film usage.
. . .
I have a couple of Canon 5D bodies, including the latest revision, with Canon 100mm macro and I also have a Fuji X-T3 with a macro lens. I have a copy stand with a Negative Supply 35mm film holder and my B&W workflow is already a "dream setup". But, as I said, for medium format color scanning... ughh.

I first got my Coolscan 5000 when it was released early 2000 and the first thing I did was to compare the scans to optical prints made of the same films. I wanted to see how it matches color/contrast and detail achieved so I sent out various films to Fuji Pro lab to make 20" X 30" optical prints on glossy paper. The Coolscan+Nikonscan automatic scans matched color/contrast very well but exceeded details captured then would show up even on that size glossy stock which of course can show more detail then matte and other more textured paper. I also sent in some slide film but unfortunately by then prints from slides came via slide copy negatives which loss a lot more detail and color/contrast changed too.

Thought you might be curious what 4000dpi looks like compared to DSLR scanning using K20D 14.6MP and D800 36.3MP.
Film was 35mm Kodak Techpan shot at ISO25 and developed in Kodak Technidol.
Bottom left crop shows the whole target 4 X 4 arrangement of 12233 charts. Center in red is shown as 100% crops above.
Above it the 14.6MP Pentax scan.
Above it the Coolscan 4000dpi scan.
Above it is the 36.3MP Nikon scan. Even though the D800 applies more pixels then 4000dpi, they are pretty much equivalent,
Big crop on the right is that red center portion optical magnification at about 4.5X. Clearly there is much detail unresolved by any of the above methods.

Full res version -> http://www.fototime.com/8372250EA44CB06/orig.jpg

 

[Bormental]

@Les Sarile thank you. Yes, I was curious To satisfy my curiosity I used to order high-res "pro" scans from a few labs, the fancier one was about $60 per roll. Those files where insane and they've shown that beyond a certain point you're looking at diminishing returns as storage requirements begin to explode.

 

[138S]

... and to make it possible one needs to construct a film scanner with a specialized light source (similar to a C41 enlarger?) which makes the color dyes react similarly? I'm quite certain we don't need a special "C41 sensor" in our hypothetical perfect scanner, but the negative+light source combination is puzzling.

With the illumination alone you cannot adjust the spectral sensitivity of the scanner to have a perfect RA-4 match, there are concerns like the channel crosstalk.

This is crystal:



but there are other papers... and historicaly some films worked better or worse with some different paper in the darkroom.

At the end you have a complexity that color mapping solves amazingly well since the digital minilab era, what you can solve perfectly with a digital calibration don't solve it by investing in hardware, becasue you are tied to what hardware does while software is flexible.

The color software conditioning (3D LUT like) works perfectly because can do all you need and it's configurable to your taste. The real information loss was produced in the taking by the spectral sensitivity of the film, this determines what hues are separated or confused. From that point you only have transformations from one color space to another one, and color mapping does all you need.

 

[Alan Edward Klein]

Alan, it makes sense. Imagine shot a very high contrast scene with a DSLR. You have highlights and deep shadows but the DSLR cannot take all well: if you expose to have good detail in the shadows then highlights are blowed, if you use a shorter exposure to conserve highlight texture then deep shadows are lost... but the HDR function of the DSLR takes two or three shots and combines both images in a single one. That image has texture in the highlights and detail in the shadows. The scanner does exactly the same.

The benefit does not comes from the scanner running slower, it comes from the second scan made with longer exposures for each row.




the 4.0 DMax rating of the Epson is inflated as in many other scanners. First measuring DMax in scanner is complex, imagine you scan a single 35mm strip, it is not the same if you have the other receptacles dor strip covered with an opaque foil or if those windows are openned and generating flare from illumination.

Take a really dense Velvia and try Multi-Exposure, you will notice an impressive difference.





It is not about running slower or faster, this is about having two photographs for each row, each made with different exposure times.




Personally I use more Epson Scan than SIlverfast, I only use Silverfast for color negative film because Negafix allows to selecs the film type, and for the conversion an optimized color map is used for each film type, and als I use SF when I need to recover ultras deep shadows in slides.

For BW I don't much the advantage of SF, and I prefer Epson Scan for that.

I'll have to try it to see for myself. But just a few comments.

The scanner only has to recover 5 stops for chrome film and about 7 stops for negative film as film has already condensed the 20+ stops of original light into those 5-7 stops. A DSLR is shooting the original ambient 20+ stops of light which has too many stops for its sensor. That's why there's a need for graduated ND filters or multi-shot HDR or better performing sensors. But even the best only have around 13 stops or so. That's not the case for the sensor in a scanner. If you look at a histogram of a scan where the white and black input points are set at 0 and 255, you usually don't get results that extend much more than half the full possible width. That's because the scanner is easily handling the 5 stops of light. So if you can raise the dMax of the sensor by slowing down the scan time, you won't necessarily go over the 5 shot range of chrome film. It does seem though that you would increase the light through the film as you suggest. So why didn't Epson do that and increase the dMax specifications for their scanner, a great selling point? A good question.

One theory is that you might add noise . Even in a DSLR, there's an optimal ISO for the cleanest capture. Let's say 100. If you raise it to 200 ISO, you'll capture the picture with less photons but with more noise. If the camera manufacturer allowed you to go to ISO 50, the sensor would be noisy on that end. So they've picked the optimal setting for the DSLR sensor in question. That seems to be similar with Epson. They picked the best speed for that sensor to get the best dMAX and optimally "clean" image with the maximum pixels. If you change the speed or the brightness of the lamp illuminating it, you're going to lose something.

 

[138S]

The scanner only has to recover 5 stops for chrome film and about 7 stops for negative film as film has already condensed the 20+ stops of original light into those 5-7 stops. A DSLR is shooting the original ambient 20+ stops of light which has too many stops for its sensor...

Alan, Velvia slides can reach 3.8D, this is close to 12 stops, if you adjust well levels and you can scan with an effective 12 bits per channel in a single pass, then the deepest shadows will have only 1 bit to be described: 0 or 1, and the shadows that are 1 stop less dense than the deepest will bave only 4 possible levels: 00, 01, 10 and 11 binary levels. No problem if those shadows are depicted near black in the monitor or in the print, but when you recover those shadows by pulling the curve (expansion) then those lack detail and sport noise.

To recover those shadows with an Epson you need to make a second exposure (Multi-Exposure) that is longer and in that image the deep shadows are descrived with a larger range of significative bits, allowing detail if they are expanded in the edition.

When you can, just try ME with very underexposed velvia, sure you notice a great difference. scan 16 bits channel, set the levels allowing some marginn at the left and save tiff, you will see that you can recover much better the deep shadows, this is specially useful for underexposed slides, some shadows are better rendered black in an image, but not always, well these are personal decisions about how the image has to look, of course.

The sample ME images shown by Silverfast are quite consistent with real behaviour https://www.silverfast.com/highlights/multi-exposure/en.html

In that particular example enhacement is a bit in the sweet point, not all images are optimized that way.

 

[Alan Edward Klein]

Alan, Velvia slides can reach 3.8D, this is close to 12 stops, if you adjust well levels and you can scan with an effective 12 bits per channel in a single pass, then the deepest shadows will have only 1 bit to be described: 0 or 1, and the shadows that are 1 stop less dense than the deepest will bave only 4 possible levels: 00, 01, 10 and 11 binary levels. No problem if those shadows are depicted near black in the monitor or in the print, but when you recover those shadows by pulling the curve (expansion) then those lack detail and sport noise.

To recover those shadows with an Epson you need to make a second exposure (Multi-Exposure) that is longer and in that image the deep shadows are descrived with a larger range of significative bits, allowing detail if they are expanded in the edition.

When you can, just try ME with very underexposed velvia, sure you notice a great difference. scan 16 bits channel, set the levels allowing some marginn at the left and save tiff, you will see that you can recover much better the deep shadows, this is specially useful for underexposed slides, some shadows are better rendered black in an image, but not always, well these are personal decisions about how the image has to look, of course.

The sample ME images shown by Silverfast are quite consistent with real behaviour https://www.silverfast.com/highlights/multi-exposure/en.html

View attachment 252220

Do you need multiple scans on properly exposed Velvia 50? With MF I always bracket my shots and don't scan photos that are underexposed.

 

[138S]

Do you need multiple scans on properly exposed Velvia 50? With MF I always bracket my shots and don't scan photos that are underexposed.

Of course, well exposed velvia requires less the ME feature... Personally I had used ME with Velvia for landscapes (sunset) in what the sky is wanted with detail and ground illumination is low and backlighted. You know, with slides you easily blow highlights and to conserve highlights some may be left underexposed.

Many shots do not exploit all velvia dynamic range or simply we want the shadows in black, but the V850 is not able to take (in a single pass) all shadow detail velvia is able to record, with ME we recover quite more.

 

[Alan Edward Klein]

Well, now that I'm shooting Velvia 50 on 4x5 for the first time, without bracketing, I may screw up more often and need all the help I can get.

 

[George Collier]

I owe Bormental an answer on something. He asked if the second (denser areas) exposure with SF and an 850 series scanner combined the two passes before or after digitization. I asked a friend who is an expert in all areas of digital imaging. His first statement was, "First of all, the only thing in a scanning process that is analog is the original." Negative in my case. He said that most likely, the SF software holds both scans, first per the curve settings made by the operator, the second to help expand the denser ranges in the original, and evaluates what parts of the image each one occupies, then combines them (in digital form) to benefit the denser areas with the second scan data. This is more or less what I do manually (I mentioned this earlier in the thread). I make a second scan after the first one (with curve changes, but ensuring no change in crop, resolution, etc), combine them and mask the top one to use the best of both. And I can see that my second scan is more beneficial to the denser areas. I don't do this with every image, just when I can't deliver the separation I want in both ranges, without having a flat place in the curve in between.

 

[Alan Edward Klein]

I owe Bormental an answer on something. He asked if the second (denser areas) exposure with SF and an 850 series scanner combined the two passes before or after digitization. I asked a friend who is an expert in all areas of digital imaging. His first statement was, "First of all, the only thing in a scanning process that is analog is the original." Negative in my case. He said that most likely, the SF software holds both scans, first per the curve settings made by the operator, the second to help expand the denser ranges in the original, and evaluates what parts of the image each one occupies, then combines them (in digital form) to benefit the denser areas with the second scan data. This is more or less what I do manually (I mentioned this earlier in the thread). I make a second scan after the first one (with curve changes, but ensuring no change in crop, resolution, etc), combine them and mask the top one to use the best of both. And I can see that my second scan is more beneficial to the denser areas. I don't do this with every image, just when I can't deliver the separation I want in both ranges, without having a flat place in the curve in between.

How do you do the second scan that changes the dMax of the V850?

 

[George Collier]

I'm not changing the Dmax, I'm changing the curve settings on the second scan to benefit the denser areas of the image. The lighter areas are blown out in this scan, but those areas are masked out when the two scans are combined in PShop (each is a layer). Because I'm making no change in crop, resolution, etc, the two scans register pixel for pixel. I just select all in the second file, copy, then paste into the first scan file, then make the mask. I tweak the mask until the merge is unnoticeable.

 

[Alan Edward Klein]

I'm not changing the Dmax, I'm changing the curve settings on the second scan to benefit the denser areas of the image. The lighter areas are blown out in this scan, but those areas are masked out when the two scans are combined in PShop (each is a layer). Because I'm making no change in crop, resolution, etc, the two scans register pixel for pixel. I just select all in the second file, copy, then paste into the first scan file, then make the mask. I tweak the mask until the merge is unnoticeable.

The scanner cannot clip blacks or white in a scan. If you're blowing them out, your settings are doing it.

If you're not changing dMax, assuming that even possible, then the scanner does not need two scans. Just adjust your curves appropriately in post. Or if you insist, duplicate in post processing a single flat scan file. Apply different curves to each duplicate file. Then combine file results.

 

[MattKing]

I expect George Collier is changing the gain settings between the two scans.
One scan, at a lower gain setting, images the thin areas of the negative well, but doesn't image the dense areas of the negative well.
The other scan, at a higher gain setting, images the dense areas of the negative well, but doesn't image the dense areas of the negative well.
When you combine the two, you end up with a file that retains more information.

 

[George Collier]

Yes, but I am not familiar with the term "gain" in scanning (more in audio). But I just adjust the settings in the curve pallet as needed. The preview tells me when I'm right.

 

[MattKing]

Gain adjustments in scanners are the same as ISO adjustments in digital cameras.

 

[George Collier]

OK, I get it. But I might also change the slope of the curve also, not just a global up or down.

 

[Alan Edward Klein]

I expect George Collier is changing the gain settings between the two scans.
One scan, at a lower gain setting, images the thin areas of the negative well, but doesn't image the dense areas of the negative well.
The other scan, at a higher gain setting, images the dense areas of the negative well, but doesn't image the dense areas of the negative well.
When you combine the two, you end up with a file that retains more information.

Gain implies changes to the amplifiers collecting the photon information. That's what a DSLR does when changing ISO's. The higher the ISO, the higher the gain or amplification. Of course amplification creates more noise. I don't believe you can change an Epson's amplifier gain directly. Apparently you can slow down the speed. But he didn't say he did that. I think what you're' referring too is applying curves. Of course that's a software applied edit to the resultant hardware scanned data. You're not changing the scan or its resultant data. Software changes can be done with a post scan editing program or by the scanner software.

 

[MattKing]

If the Epson scanners don't allow you to adjust the gain, then they must adjust the speed of the scan in order to increase the amount of exposure that the sensor receives in any multi-pass scanning procedure.
Vuescan has the ability to control both approaches, if the scanner has the physical ability to respond to those controls.
I doubt that noise is a big issue, given that the sensor is essentially just a single line being advanced by a stepper motor.

 

[Alan Edward Klein]

If the Epson scanners don't allow you to adjust the gain, then they must adjust the speed of the scan in order to increase the amount of exposure that the sensor receives in any multi-pass scanning procedure.
Vuescan has the ability to control both approaches, if the scanner has the physical ability to respond to those controls.
I doubt that noise is a big issue, given that the sensor is essentially just a single line being advanced by a stepper motor.

I once read that someone said the two scan approach is a problem for Epson scanner because the scans don't line up exactly. I don't know if that's true. Of course, with ICE, you scan twice. The second time with infrared to eliminate wrinkles and dust. So I don't know if that other guy was right or not.

 

[grat]

Silverfast says they're slowing down the second scan, rather than changing the gain (which, by the way, is only *part* of how DSLR's handle ISO, but that's a whole 'nother thread).

Changing the curves can change what analog density the digital values 0 to 65535 (or 255) will map to (which is what the densiometer window allows you to verify), but it's not going to change how much light the sensor collects as it passes over any given spot of the negative. Changing the speed, however, can increase (or decrease) the amount of light collected.

In camera terms, the ISO and aperture are fixed, but the shutter time varies.

 

[138S]

Well, now that I'm shooting Velvia 50 on 4x5 for the first time, without bracketing, I may screw up more often and need all the help I can get.

Alan, you know: spot meter !

toasting a velvia sheet is so painful that one remembers it for years

With velvia sheets one may spot meter very well all the scene to know exactly what over-under exposure we have in each place.

Al least for me, velvia sheets are too much expensive, so personally I have two solutions, one is accurate metering and only shooting those scenes that are really worth, the other one is using a roll film back to shot 6x12cm. 120 rolls are half the price per surface...

 

[Alan Edward Klein]

Alan, you know: spot meter !

toasting a velvia sheet is so painful that one remembers it for years

With velvia sheets one may spot meter very well all the scene to know exactly what over-under exposure we have in each place.

Al least for me, velvia sheets are too much expensive, so personally I have two solutions, one is accurate metering and only shooting those scenes that are really worth, the other one is using a roll film back to shot 6x12cm. 120 rolls are half the price per surface...

I do have a MF camera with Velvia. But pulling out two cameras for the same shot is too much. AT my age, I have my hands full with just one of these cameras.

I started using my P&S digital camera as an exposure meter. I can look at the screen and the histogram and pick my exposures. I know I'm taking a chance especially with Velvia. But I'm experimenting. The camera lets me frame the picture first and where I want to put the tripod. Once framed, I then know which of the four lenses I need to use. Saves some where and tear. When I shoot BW film, I set the camera for BW so I see the scene in BW. Finally I can take a video for a digital record of the scene and record the LF camera settings at the same time. Then I transcribe the settings when I get home. When I scan the film, I can store the video clips wih the scans of the picture ion a folder in my computer for record purposes.

My meter is for fallback but it's only a 10 degree spot.

 

[138S]

But pulling out two cameras for the same shot is too much.

A Nikon F65 or F55... with the 50mm E series.

You can load some 35mm velvia and make a bracketing with it, to learn it to braket +/-1 to have a feedback to learn if your single exposures are optimal.

 

[Alan Edward Klein]

[QUOTE="Alan Edward Klein, post: 2309614, member: 85761" But pulling out two cameras for the same shot is too much.

A Nikon F65 or F55... with the 50mm E series.

You can load some 35mm velvia and make a bracketing with it, to learn it to braket +/-1 to have a feedback to learn if your single exposures are optimal.[/QUOTE]
I ought to develop the 6-4x5 shots I've taken to date before ruining the rest of the 20 I bought.