Vuescan Workflow for B&W

[shaaktiman]

I'm doing some low-level programmed post-processing myself, but if you're not in to that sort of stuff, the best approach would be to load the raw file into photoshop

My understanding was that the RAW produced by Vuescan is not really a digital negative that you can translate using PS software. (Just usable in vuescan if you scan from disk.)

You can open the file, I realize, but it opens just as a simple PS file without the digital negative advantages. (For me at least.) I was unable to get Adobe Bridge to recognize the file as RAW in order to really use it as a RAW file. Is there something I am doing wrong?

 

[vicmortelmans]

You can open the file, I realize, but it opens just as a simple PS file without the digital negative advantages. (For me at least.) I was unable to get Adobe Bridge to recognize the file as RAW in order to really use it as a RAW file. Is there something I am doing wrong?

What are the "digital negative advantages"? (I don't use Adobe PS, so maybe I'm missing something?)

Would other RAW formats contain extra information that is used by PS to do postprocessing? If you call something 'RAW', I sounds like a bare collection of pixel values and nothing more. Just like the VueScan 'RAW' tiff's are, as far as I know.

Another issue may be the value representation (i.e. linear to the light intensity transmitted by the negative, or logarithmic, or anything else), but that would depend on the scanner, probably. Most detectors (if not all) are linear, so that's the kind of RAW output I'd expect.

Groeten,

Vic

 

[michael.panoff]

Would other RAW formats contain extra information that is used by PS to do postprocessing? If you call something 'RAW', I sounds like a bare collection of pixel values and nothing more. Just like the VueScan 'RAW' tiff's are, as far as I know.

Well, "real" RAW files, out of a digital camera, contain no additional processing by the camera image engine. They are as close as you can get to what the sensor sees. So you'd get extra bit-depth, no sharpening, no compression, etc.

It's another story with vuescan, it uses it's own "RAW" format for internal use. Pretty much useles to load those info PS. Just look at the histogram of a RAW vuescan tiff, it totally wacked.

At least that is my understanding.

 

[Oh Two]

I've given up on Viewscan, on both my Monolta Scan Multi II and Epson 2450 it can't match the factory software. It's RAW scans are only for Viewscan and there is no DNG correlation for it. Either with the flatbed or film scanner when multi pass is selected it doesn't multi pass at all (or from my tests I can see no difference and I can clearly see a difference with the factory software). It's bit depth doesn't seem to come close to factory settings. 2400 dpi looks like 1200 dpi to me, and the factory interpolation is far better than Viewscan's. It's film settings seem worse than none at all in the factory software.

If you have a bunch of scanners lying around (like I do) Viewscan is kind of a Swiss army knife, but my old Umax Power Look III makes better images with the original sofware than the far newer Epson 2450 and Viewscan.

Yes, I bought the pro version too!

 

[kaiyen]

Quite simple. Since you scan a negative, the highlights as perceived by the scanner are actually the shadows in your final image. By increasing the scanner exposure, detail in the image schadow area's will be lost.

Groeten,

Vic

Huh. That's all kinds of backwards terminology, but I get what you mean 🙂.

One usually uses the same terms whether we're talking negatives or the prints, even though the negative is...the negative of the positive 🙂. So highlights in the print are still referred to as highlight (density) on the negative, even though it's the dark part of the negative.

But that's just semantics, and neither here nor there.

allan

 

[Nick R.]

I open the RAW or TIFF files from vuescan in PS all the time and the histograms look fine. That's my normal workflow: scan as TIFF; process in PS. I have two scanners, an epson flatbed and Minolta scan elite. Each came with their own software and I also have Silverfast. In the end, I always come back to Vuescan. In advanced mode, it gives me the most control over the output.

Well, "real" RAW files, out of a digital camera, contain no additional processing by the camera image engine. They are as close as you can get to what the sensor sees. So you'd get extra bit-depth, no sharpening, no compression, etc.

It's another story with vuescan, it uses it's own "RAW" format for internal use. Pretty much useles to load those info PS. Just look at the histogram of a RAW vuescan tiff, it totally wacked.

At least that is my understanding.

 

[kaiyen]

Nick,
Big question here but...how are you scanning such that the RAW tiff out of Vuescan is giving you a usable histogram directly in PS?

allan

 

[planetjoe]

Just wanted to chime in on this excellent thread.

About the Vuescan raw file: it's just a TIFF, as Nick pointed out. (Note also that I don't write it as "RAW", to avoid confusion with camera RAW formats. There's actually nothing special about it, format-wise.)

You will find, however, that it's a pain to edit directly, as it's usually much darker than is useful. Folks have said they needed to temporarily boost the gamma when editing it (for scratches, hair, etc.) and then drop the gamma before re-saving. When it's read through Vuescan again, in "scan from raw" mode, the result is another TIFF that has the more familiar contrast you'd expect from a scanned negative.

I scan my B&W negatives as RGB positives, using the "Image" mode in VS, so the VS-raw files in this case are even harder to edit directly. I've found some methods that work from emulsion to image, so to speak. Some details:

1. I tweak and save a unique Vuescan ".ini" file for each roll of film scanned. This way, I can go back and see what settings created what effects. This also helps, say, if I scan a roll of Tri-x @EI400 and then need to do one @EI1600 - I can use the original .ini file as the starting point for tweaking a new one.

2. I don't apply grain reduction at the raw scan step, but rather at the "scan from raw" step. I'm not sure if this makes a difference, yet.

3. I use the exposure control (VS curves) like adjusted developing; that is, I create "push" and "pull" versions of the master .ini file when needed. It works a treat for me.

When I get back to my home PC, I'll post some examples of these .ini files for those that are interested.


Cheers,
--joe.

 

[kaiyen]

Joe - those ini files would be awesome.

At some point, I'm just going to start blanket-stealing this info for the wiki. Unless others feel like posting directly.

allan

 

[Nick R.]

To start with, I'm using a Mac and PS CS. Cross platform comparisons might not be applicable.

 

[kaiyen]

Well, other than gamma issues, comparisons are fairly valid.

As others have pointed out, when I try to open up a raw tiff from vuescan in photoshop, there is a massive color cast and the histogram is really compressed. it is only when I output from Vuescan, either at the time of scan or later from the raw file, that I get the full tonal range. So clearly you're doing _something_ different that is giving you a fuller histogram.

allan

 

[michael.panoff]

I'm with allan.. the histogram of the vuescan tiff is shifted over to one side. Using levels I can isolate the usable part of the histogram, but this tends to be more trouble than its worth. Maybe I'm missing a step in PS.. to generate a usable histogram.

When you make a gamma adjustment, do you just bump the gray (center) slider in a PS levels adjustment layer up to the correct gamma. What gamma does vuescan use, and what is the proper gamma?

 

[vicmortelmans]

If your histogram is limited at the left half of the bit range (= there are no 'raw' values larger than 1/2*MAXVAL), then your exposure setting is too low. Increase the exposure in Vuescan by factor 2. This may be due to a dense film base or just becaus your images don't have deep shadows. This will help you to fetch more detail information from the negative.

If your histogram is limited at the right half of the bit range, first of all check if the scanner is not overexposing (typically, the histogram would 'bump' into the right range limit, rather than fade out somewhat before the limit). If so, decrease exposure by half.

If you have a full histogram, but it's still cramped in the right half of the bit range, your negative image must be very flat. The dynamic widht of the right half bit range is 1 (one!). Either you've done some bad development, or your subject is contrastless. In se this is no problem, and still you'll get the most information possible out of your negative! There's no use in trying to decrease exposure further, because you'll only loose information.

Groeten,
Vic

 

[kaiyen]

Victor,
I think you are coming to unjustified conclusions about the negative from the _raw_ tiff file from VS. Yes, I could get the histogram moved over if I increased exposure. But that would also increase exposure for the version that VS outputs as well. Which would result in blown out everything. It's not the raw file that I want to get to look right, anyway - VS uses it to reinterpret the file and give diffferent versions, that's all.

I don't think that using the raw data of the scanner sensor is a valid way of judgeting exposure and development. The raw data is meant to be interpreted, just like a negative is. It's not the final result.

allan

 

[vicmortelmans]

For sure you shouldn't evaluate the raw scan data as you would a evaluate an image. Just like you say, you should also not evaluate the looks of a negative slide as you'd evaluate the looks of a print (which I learnt only recently: an overdeveloped negative *looks* better to the eye, but it's a pain to scan)

But still the raw scan data and mainly the histogram can tell you something about

1) quality of the negative (e.g. overdevelopment/overexposure is something you'll see by a histogram stretching or peaking far to the left; underexposure/underdevelopment is something you'll see by having a very narrow histogram)
2) quality of the scanning process (e.g. exposure setting of the scanner)

As a main input, the raw data histogram will tell you exactly *how much information* you can get from the negative.

And I stay with my point that you should set the exposure such that the histogram is as much to the right *as possible*. "As possible" meaning that the histogram should not pass the right boundary of the data range!

When I get bad results after postprocessing an image, the first thing I'd do is look at the raw scan data, because most probably I'll find the reason for bad quality there.

 

[vicmortelmans]

Vuescan improvement

as a spin-off of this discussion, I'd like to point out an idea of mine that could give an extra advantage to Vuescan software. It would be nice if it could be configured such that it does a dual pass of the negative, but with different exposure settings (*). Especially for dense negatives, this could be advantagous. Doing a single exposure pass, you either set the exposure high to capture the highlight (dense) area's withouth too much noise interference, or you set the exposure low to capture the shadow (transparant) area's without clipping. In both cases, you loose information on either end of the dynamic range. If you pass once with the low exposure and once with the high exposure, you can then combine both results by some rule that takes the low-value pixels from the high-exposure pass and the high-value pixels from the low-exposure pass. This would increase the effective dynamic range of your scanner dramatically!

Hallo Mr Vuescan developer, did you hear me?

Groeten,

Vic


(*) does someone know what the Vuescan "Long exposure pass" checkbox actually does? Does it perform an extra pass next to the 'normal exposure pass', and if so, how are the results brought together?

 

[Roger S.]

Vuescan improvement

as a spin-off of this discussion, I'd like to point out an idea of mine that could give an extra advantage to Vuescan software. It would be nice if it could be configured such that it does a dual pass of the negative, but with different exposure settings (*). Especially for dense negatives, this could be advantagous. Doing a single exposure pass, you either set the exposure high to capture the highlight (dense) area's withouth too much noise interference, or you set the exposure low to capture the shadow (transparant) area's without clipping. In both cases, you loose information on either end of the dynamic range. If you pass once with the low exposure and once with the high exposure, you can then combine both results by some rule that takes the low-value pixels from the high-exposure pass and the high-value pixels from the low-exposure pass. This would increase the effective dynamic range of your scanner dramatically!

Groeten,

Vic

(*) does someone know what the Vuescan "Long exposure pass" checkbox actually does? Does it perform an extra pass next to the 'normal exposure pass', and if so, how are the results brought together?

The long exposure pass sort of does what you describe, except that the two scans aren't really aligned and you might see evidence of CCD blooming or other artifacts in the combined scan. It's a crudlely implemented feature and I personally avoid it. If you need to make two exposures to increase the dynamic range of your scanner, I'd suggest scanning once at nominal exposure, and then once again 2 EV values or so higher (e.g. 1 and 4, or 2 and 5) and then combining in a program like Photomatix which will align images and smoothly blend them together. I use this for dense slides, and the free trial will blend two images and output an unwatermarked file.

With B&W negatives your scanner should be able to get all the data in one pass. If your scanner is particularly noisy and the negative dense, there might be some benefit.

My suggested B&W workflow with Vuescan (I use a Canon FS4000US, by the way) is to enable the histogram, and set it to image (or raw), do a preview scan, (for problem negatives, lock exposure and preview again and look at the histogram) set color balance to none, negative type to "generic color negative" (even for B&W), output a 16 bit TIFF and then do all adjustements in Photoshop. This should give you a flat image with no clipped data to work from.

If your highlights are noisy or you see the histogram is clipped, try upping the exposure.
Also, Allan, if you overexpose the scan, you'll block up the shadows, *not* blow out the highlights. Remember negatives are the opposite of slides.

 

[kaiyen]

When I get bad results after postprocessing an image, the first thing I'd do is look at the raw scan data, because most probably I'll find the reason for bad quality there.

Victor,
Interesting. While this has not been my experience or, rather, my _methods_, I have not done enough testing in any case. Good info to know.

allan

 

[michael.panoff]

Well.. to add a little more fuel to the fire. I decided to go back to using Nikon Scan. I did some side-by-side comparison.. Nikon scan gets awfully close to clipping the histogram, although I'm not entirely sure it does. It has much less overall control.. but it gets just a little closer to what I'm looking for.. with much less PSing (as compared to vuescan).

I'm probably missing something with my vuescan workflow, but the photoshop tweaks are always much more dramatic (with vuescan scans). In the long run it ends up saving me time.

Another nice feature of the Nikon software is the little thumbnails it produces when inserting a 6 frame strip... I believe it uses this information to attain frame boundries, it pretty much nails it everytime.

Anyone else have similar experiences?

 

[michael.panoff]

and now I'm back to using vuescan.. clipping, Nikon Scan loves to clip both shadows and hilights. There is much talk all over this forum and others in regards to this... it really too bad, because I like the preview feature.

Anyway, it took months to figure out all the finer details of Vuescan, but it was worth the effort. Nikon scan by default also seems to boost the contrast too much, at least too much for my liking.. I'd rather not bring contrast down in Photoshop, but start with a less contrasty scan and bring it up if needed.

Also my Coolscan V seems to have a mesed up CCD, I get one streak all the way across the frame.. seems to happen most often on really dense tri-x negatives. But for some strange reason the streak only appears in the red and green channels. I vuescan my negatives as slides, then use the channel mixer in PS to isolate the blue channel.