A question or two about Vuescan

[Michael Bailey]

I'm using Vuescan with my shiny new Epson V750 scanner. I really like how it allows me to quickly scan black and white negatives into DNG format for later editing in Adobe Camera Raw. But, of course, I have questions.

1.- Aside from resolution, is there any setting that will affect the resulting file? If not, it's great to know that I can ignore so many check boxes!

2.- Is there any loss of quality when using this method compared to others?

3.- Is Vuescan somehow better than ACR at editing its DNG files?

Thanks,

MB

[Bart_van_der_Wolf]

I'm using Vuescan with my shiny new Epson V750 scanner. I really like how it allows me to quickly scan black and white negatives into DNG format for later editing in Adobe Camera Raw. But, of course, I have questions.

1.- Aside from resolution, is there any setting that will affect the resulting file? If not, it's great to know that I can ignore so many check boxes!

Hi Michael,

VueScan's strength is its capability to communicate with the scanner while bypassing the manufacturer's driver. It communicates directly with the scanner's firmware/hardware. This allows it to optimize the per channel exposure time. By optimizing the exposure time, you will be able to "expose-to-the-right", IOW maximize the Signal to Noise performance, for each individual color channel (48 bits per pixel RGB mode). This even helps to improve Black and White scans. It is beneficial to scan a Black and White film as an RGB film. You will produce 3 very similar grayscale images, but the noise will be averaged.

To achieve the optimal per-channel exposure levels, you can follow the suggestions in the "advanced workflow" instructions for color negatives as explained in the manual.

VueScan also allows direct control over the scanning resolution, higher resolution will reduce grain-aliasing issues. Always try and use the native resolution of the scanner for the best results, and then downsample if needed. It will also maximize resolution.

2.- Is there any loss of quality when using this method compared to others?

The only thing I could think of is that due to longer scanning time, film flatness becomes even more important. When the film heats up it will expand, but a good filmholder can prevent most issues. Depending on atmospheric and storage conditions, it may help to scan twice and use the second scan where the film was already a bit warmer/dryer.

3.- Is Vuescan somehow better than ACR at editing its DNG files?

Once the physical scan is done, VueScan is just a postprocessing application like Photoshop. the only benefit is that it has several filmcurves built-in, but in theory you could also do that with ACR. I usually save a TIFF as output from VueScan, which already has its inverted filmcurve and gamma applied, it's just more convenient than figuring out filmcurves in ACR. TIFFs can also be opened as ACR camera Raw files.

Cheers,
Bart

[dmerger]

Bart already provided sound advice. As Bart mentioned, it doesn’t make much difference whether you save your scans in DNG or Tiff since both can be edited in ACR. It really depends on which gives you the best end result after your ACR adjustments.  You may want to do some tests to see what you like best. 

I’m very skeptical, however, that Vuescan or your scanner is capable of physically adjusting the scan time for each individual color CCD channel. Instead, I believe that the scanning software makes the adjustment to the scan data.   Even so, such adjustments may be useful and give you good results. 

 A couple of other things to add to what Bart already mentioned is that you may want to save your scans in 16 bit and use either your scanner’s native color space or maybe ProPhoto or aRGB (I’m not sure what’s best for B&W).  Also, if you save your scans in Tiff format, be sure that your color space profile is embedded in your file before you open it in ACR, otherwise ACR will assume it is sRGB.

I’d also recommend that you do some tests with your Epson scanning software.  It may turn out that you like the results, but you won’t know until you do some tests. I use a Minolta scanner and after some tests I found that the Minolta software provided much better results with color negatives than did Vuescan.  YMMV

[jaapb]

Hi Michael,

Vuescan is very capable scanning software with lots of settings te be considered but challenging to master.
I usually scan B&W negatives with a profiled scanner, saved out as inverted high bit full res greyscale TIFF's with gray profile assigned. I always use 10% border to compensate for black or transparent edges of the negative interfering with Vuescan exposure calculations if you don't get the cropping right or using "maximum" in the crop dialogue. TIFF file type is fully capable of capturing all the scanner data and allows for proper editing in your favorite program.

You really don't want to use develop in ACR/LR on untagged Vuescan DNG files. According to Andrew Rodney (I can't remember which thread) the problem with the vuescan DNG's is that they are not real raw files (not non-demosaiced, non linear) and are untagged and ACR/Lightroom then assumes sRGB. Under the hood conversions of these vuescan DNG's give unpredictable results.
I 'm sure Andrew can explain this much better.


Hope this helps.

Jaap

[Mark D Segal]

I’m very skeptical, however, that Vuescan or your scanner is capable of physically adjusting the scan time for each individual color CCD channel.

Scanners with LEDs are so enabled, but the Epson V750 doesn't use LEDs, so this is probably correct, though there is a fair bit of technical discussion about how non-LED light sources can be so controlled as well. I have no idea, however, whether the Epson v750 uses any of those systems and I'd also be inclined to doubt it.

[Bart_van_der_Wolf]

Scanners with LEDs are so enabled, but the Epson V750 doesn't use LEDs, so this is probably correct, though there is a fair bit of technical discussion about how non-LED light sources can be so controlled as well. I have no idea, however, whether the Epson v750 uses any of those systems and I'd also be inclined to doubt it.

Hi Mark,

The exposure control is done by varying the exposure time for a channel, so lightsource intensity is not important. However, I've double checked with the most recent version (9.0.21) of VueScan Pro on a V700 scanner model, and VueScan doesn't offer separate control of the channel exposure time when in Black and White negatives mode. However, it does alter the relative channel balance in the Raw scan data to something pretty well balanced in R, G, and B. VueScan probably uses/exploits a firmware trick by altering the channel balance.

I used a manual override for the file type, changing it from Auto to 48-bit RGB, and have my preferences set up to show a histogram for the Raw data with a logarithmic vertical scale. I then tweaked the overall exposure to maximize the exposure of the brightest channel, without clipping. The attached settings example histogram shows how close the per channel exposures were, and the slight tint in the highlights is simple to fix to neutral when converting from Raw to gamma and tonecurve adjusted output.

This will result in 3 almost optimally exposed channels which will result in even lower (scanner) noise scans due to averaging.

Cheers,
Bart

[Mark D Segal]

Hi Mark,

The exposure control is done by varying the exposure time for a channel, so lightsource intensity is not important. However, I've double checked with the most recent version (9.0.21) of VueScan Pro on a V700 scanner model, and VueScan doesn't offer separate control of the channel exposure time when in Black and White negatives mode. However, it does alter the relative channel balance in the Raw scan data to something pretty well balanced in R, G, and B. VueScan probably uses/exploits a firmware trick by altering the altering the channel balance.

I used a manual override for the file type, changing it from Auto to 48-bit RGB, and have my preferences set up to show a histogram for the Raw data with a logarithmic vertical scale. I then tweaked the overall exposure to maximize the exposure of the brightest channel, without clipping. The attached settings example histogram shows how close the per channel exposures were, and the slight tint in the highlights is simple to fix to neutral when converting from Raw to gamma and tonecurve adjusted output.

This will result in 3 almost optimally exposed channels which will result in even lower (scanner) noise scans due to averaging.

Cheers,
Bart

Bart, in my post I was confirming Dean's perception about whether this scanner allows controlling exposure via per channel lamp control, without specifying whether that control would be exercised by changing lamp intensity or exposure time. I agree that more often than not, when enabled, it would be exposure time, however I am reliably informed that depending on the scanner, there are cases where it could be either/or/or both. I don't believe the firmware is being tricked. The per channel adjustments are most likely the software remapping the colour balance.

[Bart_van_der_Wolf]

Bart, in my post I was confirming Dean's perception about whether this scanner allows controlling exposure via per channel lamp control, without specifying whether that control would be exercised by changing lamp intensity or exposure time. I agree that more often than not, when enabled, it would be exposure time, however I am reliably informed that depending on the scanner, there are cases where it could be either/or/or both.

Hi Mark,

Even with the scanners with RGB LEDs I know (the Nikon Coolscan range), the intensity is constant, and the exposure time per channel is varied. There may be other scanners that cut a few corners.

If one wants to determine whether it is time, or intensity (or sensor line gain), it's easy to verify. When it takes longer to scan an exposure boosted scan, it is exposure time. The V700 takes longer to scan when the RGB exposure control is boosted (Ed Hamrick knows that extra photons means higher S/N ratios). Timings are also be influenced by the data interface overhead, so there may not be an exact correspondence between the amount of boost and the extra time it takes.

I don't believe the firmware is being tricked. The per channel adjustments are most likely the software remapping the colour balance.

I haven't done rigorous per channel measurements (e.g. on the slide film settings), I usually trust Ed Hamricks choices to achieve optimal quality with each scanner model. I wouldn't be surprised if he did tweak per channel exposure times, just as the overall exposure time is adjusted, unless the firmware didn't allow to change more than the gain settings between channels. Because the channels do not match exactly, I seriously doubt it is a simple software fix in the V700 scanner. The user is not given the separate channel exposure control with this scanner (it is available with my other dedicated filmscanners), so whatever happens at the per channel level is hidden under the hood. Ed is usally pretty good in reverse engineering the command set by analysing the instructions sent over the data interface, so that's all that can be done (from a user standpoint anyway).

Cheers,
Bart

[AFairley]

I need help here from the scanner gurus. 

How can the channel exposure times be different in what I thought was a single-pass scan with the Epson 700/750?  AFAIK, there is one fluorescent lamp that travels along the platen with the CCD in its carrier.  (I can see how the right software might be able to tweak the gain for each of the CCD's color channels, if the hardware supports that.)

Also is the Vuescan DNG a true raw file?  I had read somewhere that scanner raw files (perhaps it was the NEFs from Coolscans) were TIFFs in a DNG (or NEF) wrapper and not what we think of as out-of-camera raw files.  It seems that if the software cannot make individual channel hardware-based adjustments to the scan, manually channel balancing DNG output is really no different than doing the same thing post-scan because the actual raw data captured from the scanner is the same in either case.

Thanks for clarifying.

[Bart_van_der_Wolf]

I need help here from the scanner gurus. 

How can the channel exposure times be different in what I thought was a single-pass scan with the Epson 700/750?  AFAIK, there is one fluorescent lamp that travels along the platen with the CCD in its carrier.  (I can see how the right software might be able to tweak the gain for each of the CCD's color channels, if the hardware supports that.)

Given the fact that the current version of VueScan (don't remember what happened in the past) doesn't allow separate channel control, it could be the latter. If the firmware allows, it could in theory be implemented with an electronic shutter which either stops collecting charge (which is hard to do), is reset at a slightly later time (starts recording later), or takes multiple sub-exposures which are discarded for other channels. More likely it is implemented as a lookup table with some gain presets (e.g. a different one for Color Neg), but I didn't test for that (by analysing the per channel electronic noise). We are talking about pretty small sensels (I believe in the 3 micron pitch range), arranged in 6 lines (2 staggered lines per filtered color) so they saturate pretty fast (which would make analog gain noticeably boost noise at the low end, and photon shot noise at the high end as well).

Also is the Vuescan DNG a true raw file?  I had read somewhere that scanner raw files (perhaps it was the NEFs from Coolscans) were TIFFs in a DNG (or NEF) wrapper and not what we think of as out-of-camera raw files.

Correct, the only thing Raw about the files is that the data is as captured, but in principle it is already RGB with a linear gamma (in the TIFFs anyway). Whether the DNG files are anything more than a wrapper, I don't know (I only use TIFFs).

It seems that if the software cannot make individual channel hardware-based adjustments to the scan, manually channel balancing DNG output is really no different than doing the same thing post-scan because the actual raw data captured from the scanner is the same in either case.

Correct, QED.

That's my take on the matter anyway, but I don't know everything, I'm not a Guru ... ;-)

Cheers,
Bart

[Michael Bailey]

Thank you all for your responses, both public and private.

Because this forum was designed and built exclusively for my convenience and no one else's, I am going to bypass the discussion of lamp intensity and exposure time for the moment to get back to my original question about a simplified scanning routine for black and white negatives.

Here is what I see as a consensus:

1.- There is indeed no drawback to scanning first and editing later, provided there's no clipping of lights or darks. (Though Bart's advice about adjusting Vuescan Exposure setting is worth noting.)

2.- So, aside from resolution and, possibly, Exposure and bit depth, Vuescan does appear to ignore all other settings when scanning to raw.

3.- Because DNG and Tiff contain the same information, there might be a slight disadvantage to scanning to DNG.
     Tiff allows for color management while DNG doesn't carry a profile.
     Tiff can be opened "out of the box" into Photoshop, should I choose.

4.- Scanning to Tiff means I have the option of basic editing in the scanning step, if I want it. A running start in Photoshop or ACR never hurts.

5.- If I'm going to scan to 16 bit Tiff anyway, I could just as well stay with Silverfast Ai, which comes with the scanner, and I won't have to buy one of their upgrades.
     Okay, this is a trick conclusion, because my original posting didn't mention that in asking about Vuescan I was leaving unstated my 10 year-long frustration with Silverfast and it's cryptic tools, instruction and web site navigation.
     Note to the folks at Silverfast: Please find a native English speaker with some photography knowledge to translate your instructions and narrate your videos. -Eure English ist besser als mein Deutsch aber, wie sie hier lesen, das meint nicht viel.-
     On that note, thanks to Mark D for his posts, and to Ian Lyons for his tutorials on computer-darkroom.com. Without your supplementary information I think Silverfast would be completely unusable, to me at least.

So this all leaves me with another question or two.

1.- I can understand how scanning a bw negative as RGB would refine my bit depth and minimize noise a bit. But I'm guessing that once the most basic tonal editing is finished I can safely convert the file to 16 bit Grayscale. Surely there can't be any significant loss in quality at this point. (It's not that I'm cheap about storage space. I just don't know how much I want to deal with gigabyte sized files. Hell, my pictures aren't that good.)

2.- In scanning and in Photoshop I have so far been working in Gray Gamma 2.2. My files in Photoshop look like the previews in Vuescan, so I conclude I'm doing the right thing in this regard. Am I missing something?

3.- In the Vuescan Output tab, I'm still confused by the selections for Tiff, Tiff DNG, Raw file, Raw DNG format. Any advice there?

Thank you all, again.  MB

[Mark D Segal]

I need help here from the scanner gurus. 

How can the channel exposure times be different in what I thought was a single-pass scan with the Epson 700/750?  AFAIK, there is one fluorescent lamp that travels along the platen with the CCD in its carrier.  (I can see how the right software might be able to tweak the gain for each of the CCD's color channels, if the hardware supports that.)

Also is the Vuescan DNG a true raw file?  I had read somewhere that scanner raw files (perhaps it was the NEFs from Coolscans) were TIFFs in a DNG (or NEF) wrapper and not what we think of as out-of-camera raw files.  It seems that if the software cannot make individual channel hardware-based adjustments to the scan, manually channel balancing DNG output is really no different than doing the same thing post-scan because the actual raw data captured from the scanner is the same in either case.

Thanks for clarifying.

I think you are correct about all of this. Gain is not "user-tweakable" per channel in an Epson V750 no matter what software you use with it. The scanner itself does not support this.

The DNG file Vuescan produces is not the same thing as a raw file from a digital camera. You can see this for yourself comparing the file size of the same image as a "raw DNG" and then as a ProPhoto RGB image as converted in ACR. There is no  difference. Both of them would appear to be three channel pixel data, one with a DNG shell and the other a standard ProPhoto. The DNG designation has the convenience of opening the image directly into the raw converter for those preferring to edit there. But one can also edit TIFFs, JPEGs and PSDs in ACR these days.

I too would like to hear more from anyone who actually knows the engineering of the hardware and the programming of the firmware and software what the real meaning of this DNG designation truly is in a scan. Of course there is nothing "raw" in the original media. It is far, far removed from being scene-referred data.

[AFairley]

Bart and Mark, thanks for your replies. 

I realise that I don't actually know how a scanner CCD captures the item being scanned.  It it a series of discrete exposures as the CCD moves along the platen in discrete steps?  Can anyone point me to a resource for learning more about the mechanics of the whole process?

Thanks again.

[jaapb]

Michael,

In my view the Vuescan raw workflow enables you to do batchscanning without individual adjustments to the files first. Once you scanned your raw batch you can come back in adjusting each file individually later in Vuescan. Forget about DNG in Vuescan.

Jaap

[Bart_van_der_Wolf]

Bart and Mark, thanks for your replies. 

I realise that I don't actually know how a scanner CCD captures the item being scanned.  It it a series of discrete exposures as the CCD moves along the platen in discrete steps?

Yes, but with variations.

There are basically 3 types of CCDs being used, so it depends on the specific scanner model.

• 1. Uses an area CCD, and makes an image of the entire film-area at once. This is e.g. used in the former Kodak Rapid Film Scanners. Color is acquired by e.g. making 3 exposures through a filter wheel, but one could also use a Bayer CFA area sensor array.
• 2a. Uses a tri-linear array sensor which is moved 1/3rd between exposures so that each RGB filtered line scans the same position in sequence. It requires accurate mechanical stepping to step 1 line/color at a time, and is relatively fast because 3 colors are aquired at the same time (although from offset/displaced positions on the original at any one moment). Each line is assembled to RGB pixels after each of the 3 colors have scanned the same positon after the 3rd step.
• 2b. Epson was the first to introduce a sensor in their consumer scanners with 2 scanlines per color, so 6 in total, each second line offset/staggered by half a sensel, so they overlap the gap between the sensels of the previous line. This effectively doubled the number of sample positions and improves the MTF. Resolution is also improved but not by a factor of 2 (although that may be due to the lens and focus limitations). Stepping accuracy needs to be very good.
• 3. Uses a single linear array, which requires even better stepping accuracy, but allows the easiest control over the per channel exposure time. The lightsource in e.g. the Nikon Coolscan series are RGB LEDs which are used in sequence before the stepper motor advances a single line position. This is a relatively slow process, but produces high quality if the LED colors are matched well with the film dyes.

Can anyone point me to a resource for learning more about the mechanics of the whole process?

It's becoming harder all the time to get good info. Fewer and fewer people know about scanners for film. I don't know about recent sites, unless you go to the dedicated scanner support forums, e.g. on Yahoo. It requires a lot of reading between the lines (pun intended) to get info about the mechanics. When Ed Hamrick was still posting on Usenet, he disclosed a lot of info about his findings.

Cheers,
Bart

[Mark D Segal]

Thank you all for your responses, both public and private.

Because this forum was designed and built exclusively for my convenience and no one else's, I am going to bypass the discussion of lamp intensity and exposure time for the moment to get back to my original question about a simplified scanning routine for black and white negatives.

Here is what I see as a consensus:

1.- There is indeed no drawback to scanning first and editing later, provided there's no clipping of lights or darks. (Though Bart's advice about adjusting Vuescan Exposure setting is worth noting.)

2.- So, aside from resolution and, possibly, Exposure and bit depth, Vuescan does appear to ignore all other settings when scanning to raw.

3.- Because DNG and Tiff contain the same information, there might be a slight disadvantage to scanning to DNG.
     Tiff allows for color management while DNG doesn't carry a profile.
     Tiff can be opened "out of the box" into Photoshop, should I choose.

4.- Scanning to Tiff means I have the option of basic editing in the scanning step, if I want it. A running start in Photoshop or ACR never hurts.

5.- If I'm going to scan to 16 bit Tiff anyway, I could just as well stay with Silverfast Ai, which comes with the scanner, and I won't have to buy one of their upgrades.
     Okay, this is a trick conclusion, because my original posting didn't mention that in asking about Vuescan I was leaving unstated my 10 year-long frustration with Silverfast and it's cryptic tools, instruction and web site navigation.
     Note to the folks at Silverfast: Please find a native English speaker with some photography knowledge to translate your instructions and narrate your videos. -Eure English ist besser als mein Deutsch aber, wie sie hier lesen, das meint nicht viel.-
     On that note, thanks to Mark D for his posts, and to Ian Lyons for his tutorials on computer-darkroom.com. Without your supplementary information I think Silverfast would be completely unusable, to me at least.

So this all leaves me with another question or two.

1.- I can understand how scanning a bw negative as RGB would refine my bit depth and minimize noise a bit. But I'm guessing that once the most basic tonal editing is finished I can safely convert the file to 16 bit Grayscale. Surely there can't be any significant loss in quality at this point. (It's not that I'm cheap about storage space. I just don't know how much I want to deal with gigabyte sized files. Hell, my pictures aren't that good.)

2.- In scanning and in Photoshop I have so far been working in Gray Gamma 2.2. My files in Photoshop look like the previews in Vuescan, so I conclude I'm doing the right thing in this regard. Am I missing something?

3.- In the Vuescan Output tab, I'm still confused by the selections for Tiff, Tiff DNG, Raw file, Raw DNG format. Any advice there?

Thank you all, again.  MB

Michael,

Re your "consensus"

Point #1. There can well be drawbacks to not making certain adjustments at the scan stage. Exposure and focusing (where possible) are the two key ones of which to be mindful. This is confirmed by your point 4 which I agree with. There's no point closing oneself off from options which give one a "running start" as you say.

Point #3. If you use the "DNG" approach, you can assign the scanner profile to the outcome then converting to your working space, giving you an image rendition which should be similar to a colour-managed TIFF. If you are using a 16-bit wide gsmut working space, the scan should be a 16-bit per channel scan, or "48-bit" in the language of scanner software image options.

Point #5. The SilverFast manual edition as of November 2006 is actually not too bad. The GUI ain't the greatest, but once one wraps one's mind around it, one leaves these hang-ups behind and it's pretty smooth sailing thenceforth. You don't need to buy any upgrades from what they provide with the V750, so you don't need recourse to their website. Dot versions are quite easily up-graded from a dialogue within the application and the process is fairly seamless. And thanks for your appreciation of my contributions on this website - I'm glad it's been helpful.

Turning to your other questions:

Q 1: Leaving the file as an RGB-based file even for B&W isn't essential, but it gives you options for future work. You may want to try alternative toning options, do some split toning, do some colour/B&W mixtures which can be very effective photographically speaking, vary zonal luminosity by altering the brightness of the individual RGB channels underlying the B&W conversion, so again, good to scan a file that preserves your flexibility going forward, especially as the B&W conversion options in LR, ACR, Photoshop are so good, not to speak of what you can do with a plugin like NIK Silver EFEX Pro.

Q 2. No you're not missing anything. Gamma 2.2 is usually a safe place to be. A "purist raw scan" at Gamma 1.0 (linear) tends to produce very dense outcomes that are hard to deal with.

Q. 3 To avoid all confusion just scan to TIFF. It's very straightforward and doesn't deprive you of anything.

Hope this helps.

[Jonathan Ratzlaff]

In response to tiff and or dng files.  The most important is maximum bit depth.  Depending on the Dmax of your negative, a raw scan should look muddy; the bit depth is required because the fairly narrow range of the scan has to be expanded to provide the proper white and dark balance.
I noticed that you were sampling at 16x in your input.  If you were scanning transparencies there would be a reason for this; given the lack of density in a black and white negative compared to a tranparency, this is not as much of an issue and even single pass or at max 4x sampling should be sufficient for a negative.  The shadow detail is in the thin part of the negative and the scanner has no problem with these areas.

All I worry about with the raw scan is getting all the detail from the negative or transparency into the file. Editing software is much better for adjusting levels than Vuescan. 

Scanning in RGB mode is useful.  The more data the better; can always be reduced later

[Alan Klein]

Can anyone point to photo posting links comparing Epson program results against Vuescan and Silverfast in the "raw" scan with no PP that would distort the results?

[Mark D Segal]

No, but what's the point? What are you trying to find out with this, what are the relevant comparators and what would be the best kind of test images to use for the purposes you have in mind? None of these applications are going to give you an optimal result "out of the box" without processing in either the scan software or a post-scan external editor.

[Alan Klein]

To find out which program works the best and under what circumstances.  If that requires as you say to include PP, so be it.  Did anyone actually go through the process comparing the results at the end to see what's best and can they show the results in pictures that can be posted so we can make our own judgments?  Otherwise how do you know they are improving the scanner-supplied software?  How do you know if these other programs are just 99% PP and 1% changes to the actual scanned data, if that?  I've read a lot of what these programs are suppose to do.  Some people swear by them others swear at them and say they provide minimal improvement over the supplied Epson programs at best and just cause you to spend a lot of extra time processing with little improvement over what you can get from the Epson program with PP from Photoshop.

I’d like to see comparisons with well-exposed pictures, underexposed and over-exposed shots.  Comparing Epson, Vuescan and Silverfast or at least a coupe of them if the person hasn’t done all three.  If comparisons like this can be done with cameras, with lenses, etc. it should be able to be done with scan software.