Scanning strategy?

<p >I have quite a large archive of 35mm black and white negatives and colour slides. I still use the odd role of HP5 and have an Ektapro for projection but in other respects rely on digital. Seems to me there are two possible approaches to scanning; a) using Vuescan, do once only scan to raw, which become primary archive (PS what format does Vuescan use for raw scans – is it a proprietary code or DNG?); b) scan when needed, maintaining film as primary archive.</p>

<p > </p>

<p >Obviously I can’t rely on my Nikon Coolscan V lasting for ever, so strategy b) depends on still being able to buy an equivalent quality scanner and software if/when it breaks so that’s a risk. On the other hand, strategy (a) depends on Vuescan raw scans being a “universal” format.</p>

<p > </p>

<p >Any thoughts?</p>

<p>Vuescan will scan to DNG. That's the "official" DNG, which was developed by Adobe. There's a box you can click.</p>

<p>I consider my film as my primary storage, and my scans my temporary working area. Film can be thought of as optical storage media and the scanner as the drive. Just like the disk drives of old, scanners have become faster and higher density. Luckily our film was written to by higher density writers than our readers are. In the last 10 years I have scanned film that I scanned long ago and found many advantages.</p>

<p>For example this image is from the first ever film scan I did on a Nikon LS-1000. I was pleased with it then, but since that time both my skills and the equipment have moved on</p>

<p><img src="http://farm4.static.flickr.com/3101/3227691775_b05f6ff4c5.jpg" alt="" /></p>

<p>Even this has been adjusted somewhat from the murky thing which I still keep on my hard disk. If equipment never moves on perhaps your skills may.</p>

<p>"I consider my film as my primary storage, and my scans my temporary working area."</p>

<p>I don't think of color materials that way because they degrade over time.</p>

<p>As for the origianal quaetion, I make linear positive scans with Nikon Scan to wring as much info as possible from my scans. I store as TIFF files and make all edits in a copied version.</p>

<p>yes, a good point about stability of colour. I do have E6 and C-41 from 1980 which still looks good, but it is a gamble because I have had c-41 film become bad in a few years. In this instance I no longer rely on my film and hope my digital archival strategy is sufficient. Luck for me the Kodachrome from 1954 is still looking as bright as it ever did. I doubt tri-x or other black and white films will be a problem</p>

<p>I have 4x5 E-6 which I am not certain that there is another good storage alternative however</p>

<p>Long-term digital storage does present challenges. I'm now looking into network storage and RAID arrays. That's in addition to local back-ups to an external HD and on-line backup. Maybe I should just shoot B&W film . . .</p>

<p>"Vuescan will scan to DNG. That's the "official" DNG, which was developed by Adobe."</p>

<p>It does? Very cool! What version does that? I guess I need to upgrade since my current version uses a TIFF file as the RAW file and it's very dark unless you look at it and process it in Vuescan.</p>

<p>There's really no benefit to DNG in this workflow, the files are from a trilinear, true RGB output referred capture device. </p>

<p>"There's really no benefit to DNG in this workflow, the files are from a trilinear, true RGB output referred capture device."</p>

<p>Not sure what you mean. Can you explain? I assume Michael is saying that Vuescan will create DNG files that I can manipulate in Lr and Ps just like the DNGs that I import from my camera. In that case, I see a tremendous advantage over the dark, unusable-in-anything-but-Vuescan TIFFs that Vuescan previously used as RAWs.</p>

<p>Its not Raw data. Yes, you could in theory take a DNG from a scan and a TIFF and apply metadata instructions in LR or ACR to either. But that's already rendered pixel data. The DNG doesn't make it non demosaiced Raw data and in a way, that's good (its not interpolated color, its real RGB color). <br>

The place to do the big tone and color work is in the scanner driver. That's your Raw processor. The container, DNG or TIFF is immaterial. There's nothing a scan in a DNG brings to the party you can't do with a TIFF. <br>

An over dark scan should be corrected at the scan stage. </p>

<p>Thanks. I guess I'll just have to give it a try and see how it works.</p>

<p>In my version of Vuescan (no DNG capability), the scans are just fine, but the RAW file the program creates is a very dark TIFF (when viewed outside of Vuescan) that would be very hard to correct in Lr or Ps. You can re-open the TIFF in Vuescan and it looks just fine, because the scan was fine, and do your adjustments. The purpose is to only scan the negative or slide once, save the sensor output to a file, and then make adjustments as needed. If I can get the output of the sensor as a DNG, in the same way I get the output from the sensor in my camera as a DNG, I'll be a happy camper. Cheers. </p>

<blockquote>

<p>In my version of Vuescan (no DNG capability), the scans are just fine, but the RAW file the program creates is a very dark TIFF (when viewed outside of Vuescan) that would be very hard to correct in Lr or Ps.</p>

 

</blockquote>

<p>They are just providing the linear data. Its probably untagged so it looks dark in Photoshop. IF you had a profile reflecting the scanner for the linear data, it would look fine. There's little reason most users need linear data when you are essentially working with all the data possible at the scan stage. Now maybe the UI or tools suck and you'd rather use Photoshop. You'd either want gamma corrected documents (in a working space like sRGB, Adobe RGB Etc) or you could start the gamma encoding from scratch using Photoshop with that linear data. There's really no advantages otherwise. I'd do all the work at the scan stage, its way faster. Use Photoshop for selective work and pixel editing. </p>

<p>Vuescan must have a profile for my scanner to make the RAW TIFF display correctly when viewed in Vuescan, right? (It includes a brand and model profile and I can create individual profiles for my particular scanner too.) And if it applied that profile to create the DNG from the linear data from the sensor, wouldn't I have a DNG (which is not truly a RAW) with all the data possible that would look normal in Lr and Ps? After all, the sensor in my film scanner and the sensor in my camera are basically the same animal.</p>

<p>Ideally, I would like to put 50 slides in the autofeeder and have my scanner and Vuescan create 50 DNGs that I can manipulate in Lr and Ps just like images from my camera, and with the same quality as if I had manipulated them in Vuescan. That would be much faster and more convienent than working on each image in Vuescan.</p>

<p>Its not a Raw. </p>

<p>ViewScan should have a profile for the linear data one would suspect. </p>

<p>DNG is a container format for holding data. That data isn't necessarily Raw (despite the confusing name). </p>

<p>You can batch scan and "fix" the images later in Photoshop or you can fix them when you scan them. The former will provide better data and might be faster (you're working with a low rez preview, you apply corrections as the scan is made). There's no free lunch here. </p>

<p>Okay, thanks. It's still not quite clear, so I'll have to do some thinking and reading up on this stuff. But now I'm off to paint the living room. Thanks again. Cheers.</p>

<p>Vuescan's "raw file" is typically tiff format. You now have the option to output as DNG format, but you're gaining nothing, and a DNG format can be produced from the original tiff format raw file later, if you like, by doing scan-from-disk and outputting DNG.</p>

<p>Vuescan's author's intent with this raw file is to record all the data as it comes from the scanner, without gamma adjustment, and with no stretching/clipping of the histogram, to be used for Vuescan's Scan-From-Disk function, essentially: pseudo-scanning, using the raw file instead of "live" data from the scanner.</p>

<p>So, this file is a regular tiff. The visible differences between this file and a regular output file:</p>

<p>1. It appears much darker, due to no gamma adjustment.</p>

<p>2. It is quite flat, due to no histogram adjustment.</p>

<p>3. It is not inverted, in the case of negative film scans.</p>

<p>I'm going to point out that much of the original purpose of the TIFF file format was to store scanner output. So it makes perfect sense to use it as a raw for a scanner.</p>

<p>It is confusing and incorrect to use the term Raw scan when many readers will associate this with a Raw (digital capture) file. There's a huge difference. An uncorrected (default setting) on a scanner might produce a scan someone could have called "Raw" but there's little reason to do this today and naming it Raw adds a lot of confusion when non should exist. </p>

<p>Vuescan RAW is a tiff. I don't see any benefit to scanning as Vuescan Raw/DNG versus tiff format- either is fine. I output 48 bit TIFFs with IR cleaning applied.<br>

I disagree with Andrew on doing edits in the scanner driver- at least Vuescan's and Canon's tools are far inferior to Lightroom/Photoshop, and the biggest breakthrough I've seen in photo editing is the ability to do reversible edits in Lightroom (or 16 bit layers in Photoshop). Scan without clipping any data and do edits in post (and create presets to save time).</p>

<p>IF you can get the full bit depth and color gamut out of the scanner, no reason not to use Photoshop. Its probably slower and, if your scanner software sucks, well that's too bad (there have been some great scanning products over the years). </p>

<p>Metadata editing in LR on rendered images always takes place on a duplicate that's in linear ProPhoto RGB. Its not non destructive like editing true Raw data (that's pixel creation). The edits are applied in best order on the converted high bit ProPhoto space. Making LR the scanner on high bit ProPhoto out of the scanner seems quite reasonable which again makes it so important that we don't miss understand or use the term Raw (and Raw processing) incorrectly. A "Raw" scan in Lightroom isn't anything like a Raw DSLR document in Lightroom in terms of the data or the processing. </p>

<p>A raw scan file is indeed very different from a digital camera's raw file. A raw scan file can loosely be described as one that only uses a scanner's hw with all the scanner's sw correction functions disabled. The intent is to capture as much info as possible on a raw scan and then make corrections in PS. A film scanner's native sw typically supports raw scan, making a third party sw unnecessary. A digital camera's raw file has far more exact definition. If I recall correctly, the term raw scan predated ACR.<br>

The early ACR supports only digital camera raw files, but the current version supports other files such as psd, tiff and jpeg, etc. I've yet to find anything on using ACR on non digital camera raw files to be good, bad or indifferent.<br>

Here are some info on raw scans I posted in another dated thread:<br>

=======<br>

I only use a scanner's hw capability to capture a scan and do all the editing in PS. This raw scanning technique is seldom, if ever, covered by the numerous scanning books and tutorials. These authors fail to separate between what is being done by a scanner's hardware and software, leading the readers to believe that they can do everything better with a scanner's sw. In reality, PS' editing capability is far superior. Once you get over that, you can understand what a raw scan is, and why some prefer it.<br>

Here are some references on the topic:<br>

Real World Scanning and Halftones (3rd Edition)<br />by David Blatner (Author), Glenn Fleishman (Author), Steve Roth (Author), Conrad Chavez (Author) <br />Paperback: 352 pages ; Dimensions (in inches): 0.66 x 9.16 x 7.36 <br />Publisher: Peachpit Press; 3rd edition (April 8, 2004) <br />ISBN: 0321241320<br>

I found that this is the only book that separates a scanner's hw and sw in some details.<br>

<a href="http://www.naturephotographers.net/mh0202-1.pdf">http://www.naturephotographers.net/mh0202-1.pdf</a> (See "Input Method 2: Manage Color After the Scan")<br>

This author explained why he preferred raw scans, and cautioned that practioners should be able to stand ugly looking raw scans, which can be profiled and edited to look good in PS.<br>

<a href="http://www.lumika.org/gear_nikon_scan_vs_photoshop.htm">http://www.lumika.org/gear_nikon_scan_vs_photoshop.htm</a><br>

This author ran a simple test. He concluded that tutorials like Scantips to be incorrect and even recorded Ed Hamrick in agreement.</p>

<p>"I've yet to find anything on using ACR on non digital camera raw files to be good, bad or indifferent."</p>

<p>The ACR version that I have has limited functionality for white balance on non-digital camera raw files.</p>

<blockquote>

<p>

<p >"I've yet to find anything on using ACR on non digital camera raw files to be good, bad or indifferent."</p>

</p>

</blockquote>

There are a few significant advantages:

<br />

All edit history (at least in LR) is accessible after yo quit (not so with Photoshop).

All edits are applied in one convolution to the pixels, in high bit, high gamut and in the best processing order.

All edits are fast because you're not burning pixel edits UNTIL you export (or print). You're just making metadata instructions.

You can take some or all edits applied to one image, paste onto others (well you can kind of drag and drop adjustment layers, not quite as robust).

<br />

There's no free lunch however, a raw converter like ACR or LR are far more powerful on Raw data, not baked existing pixels. But there are reasons why you'd want to apply edits in a parametric editor instead of a pixel editor.