Achieving high quality results with Epson V750 Pro

[paul_bohman]

<p>For those who are constantly asking which film scanner to buy (I was one of them), I have a couple of posts about my recent experiences using Kodak Portra 160 and Portra 400 with a Mamiya RZ67 Pro II and an Epson V750 Pro.<br>

http://www.paulbohman.com/blog/?p=596<br>

http://www.paulbohman.com/blog/?p=631<br>

The summary is this:<br>

After spending a lot of time getting to know Silverfast 8 and creating a custom Negafix profile (along with some "global color correction" tweaks to warm up the mid tones and highlights), and after adjusting the height of the film tray (very, very important!), and after using Photoshop's unsharp mask, I was able to get scans with the big 6x7 negatives that are equivalent to digital files in the 27 to 35 megapixel range.<br>

Careful scans like this, after learning how to take maximum advantage of the hardware and software combination, can reveal details down to the film grain, especially if over-sharpened.<br>

You can get results like this on Imacon scanners without having to use the unsharp mask in PhotoShop, but you also have to pay several thousand dollars for an Imacon. Now that I know I can get comparable results, I don't feel the need to rush out and get an Imacon, though of course I'd use it if I had one. Imacon is slightly better because it requires no sharpening, but the difference isn't worth the price difference, at least for me right now.<br>

But to get results like this, you absolutely must invest the time getting to know the hardware and software, and you must experiment with the height of the film trays. The scanning solution by ScanScience is good too to improve sharpness and shadow/highlight detail, but less important than the other factors.</p>

<div></div>

[paul_bohman]

<p>"...especially if over-sharpened."<br>

In saying this, I don't mean that you *should* oversharpen. I just mean that if you take it too far, the image will look really crispy *and* the film grain will show prominently. I prefer to sharpen up to a certain point, but not so much that it looks artificial.</p>

 

[zoltan_arva_toth]

<p>First. congratulations on the portraits - they are very nice aesthetically. You certainly know your craft!</p>

<p>On the technical side, I like the skin tones and other colours a lot - wish you could share your profiles for the new Portra 160 and 400! The scans look very nice in general, and the 50% crops on the second page you linked exhibit very good detail and sharpness. I'm not sure how you have arrived at the conclusion that you can eek out 35 megapixels' worth of detail with the Epson scanner, though. Mathematically, if the original scan is 71 megapixels as stated, the 50% crops are the equivalents of 100% crops from a roughly 18-megapixel image (=seventy-one divided by two squared). For that, they are really nice and crisp, arguably more so than a 100% crop from a 16-20-megapixel digital camera image, so it's not unreasonable to argue that your scans contain about as much detail as a 24-27-megapixel digital photo, but 35 seems a little overstated to me. Of course this doesn't matter much - I'm sure you can make really big and nice prints from your scans, and that's what really matters. Also, film has a distinctive look that I like a lot, like yourself.</p>

<p>As to seeing grain in flatbed scans of modern, slow-ish colour negatives, I don't think that's possible. Sure, a high amount of crude sharpening will make the scans look rough at 100%, but that roughness is hardly grain itself - it's more like the complex interplay of grain, grain aliasing and scanner noise. The folks over at apug.org who make wet/optical prints from the likes of Ektar and Portra 160 say they can't use the time-honoured method of grain-focusing with these emulsions, because they can't see the grain!</p>

[paul_bohman]

<p>Ok, you're right that sizing the image to 50% doesn't drop the resolution in half, it drops it to a quarter of the original size. I should have thought through the math better. I overstated the resolution, and that's an important clarification.<br /> That being said, when I resized it to 5400x6568 (35,467,200 pixels) it still held up well, with a bit more softness than the 50% crop, but still sharp enough for a very good print at that size (about 18x22 at 300dpi, or about 30x36 at 180dpi). So, now that I've been corrected on my math, I'm going to estimate the true resolution of 6x7 negatives when compared to digital files to be between 16 and 27 megapixels, depending on scanning technique and photographic technique.<br /> I'll eventually post more about how I use Silverfast on my blog. Basically, what I did was go into the expert mode in the Negafix dialog (click on the little graduation cap icon) and straighten out the main curve (so that all colors move together) so that it's almost a straight 45 degree line. See http://www.paulbohman.com/assets/images/photo/silverfast/negafix.jpg Then for the colors, I went into the expert mode in the Global Color Correction and warmed up the values a bit as shown in these screen shots (note that I adjusted the values separately for 25%, 50% and 75%):<br /> http://www.paulbohman.com/assets/images/photo/silverfast/globalcc-25.jpg<br /> http://www.paulbohman.com/assets/images/photo/silverfast/globalcc-50.jpg<br /> http://www.paulbohman.com/assets/images/photo/silverfast/globalcc-75.jpg<br /> That's my baseline. If I need to do any other color adjustments, I can do that in Aperture or Photoshop.(I ended up warming up the picture a little bit more in Aperture, but it still looked good with the settings mentioned above).<br /> I should clarify that these are the settings that work for this particular lighting setup, using AB800s. I may end up needing to create different color settings for different light temperatures.<br /> And yes, I still like the look of Portra's skin tones more than most digital files, and I can still get high resolution files, even if I did accidentally overstate the resolution originally.</p>

[kparratt]

<p>Thanks for the thread Paul.<br>

Regarding adjusting height of the film holders, (I have the V700) .. I have never sat down to address this one yet, because I'm a B&W wet darkroom printer, and the colour images have for the time being been for reference only. But the day will come when I need to produce higher quality colour scans. Is there a source you know of, a guide on how to do this? .. or is it simply hit and miss ..</p>

[paul_bohman]

<p>From what I understand, every scanner is slightly different, so you have to spend the time to find the optimal height for your scanner. Getting the right height is way more important than using unsharp mask in Photoshop. When I did my first scans before adjusting the height, I had to really sharpen the files to get a clear image, and then it looked really over-processed. When I found the right tray height, the sharpening was much easier and more natural looking. And if you scan at 2400dpi or below, you can get away with no sharpening in Photoshop at all. At least that was my experience with my scanner.</p>

[paul_bohman]

<p>In answer to Zoltan's assertion that the film grain can't be seen on Epson scans, I'm pretty sure that the texture I see in my scan is the film grain. Compare the closeup of my file (look at the out of focus areas to see the grain):<br>

http://www.paulbohman.com/assets/images/photo/portraits/ania/film/ania-straighton-crop-mouth.jpg<br>

... to the closeup of a scan Stephen Schaub did on an Imacon at the level of the film grain (look at the shot of the eyebrow):<br>

http://figitalrevolution.com/2011/02/21/new-kodak-professional-portra-160-film-new-negative-c41-scan-hybri/<br>

His closeup is zoomed in more than mine, so the grain looks larger in his photo, but the shape and texture looks the same as in my photo. So yes, I'm fairly confident that what I'm seeing is the grain and not just digital noise due to sharpening. After all, the unsharp mask doesn't sharpen smooth areas or change them at all. It finds differences in contrast and sharpens those differences, so the texture was already there. The unsharp mask revelead it and accentuated it.</p>

[john_shriver]

<p>Very nice results. I can really resolve film grain in 1950's B&W film with my V750 and a Better Scanning LF film holder.<br>

I had to scratch my head a bit about your choice of 3600 dpi scanning resolution. So I went back to the V750 specs, and realized that the native resolution is 6400 dpi when not using the full area, and only falls to 4800 dpi when using the second lens with the "full area" (8x10). So you did scan at an even sub-multiple of the native resolution, so there isn't any fancy interpolation. I'd been preferring resolutions like 1200 (which is an even multiple) and 2400 (which isn't).<br>

Yes, NegaFix isn't perfect. The profiles definitely vary in quality. Thankfully the recent Ektar 100 one is good, it's a tricky film to scan.</p>

[paul_bohman]

<p>"I had to scratch my head a bit about your choice of 3600 dpi scanning resolution"</p>

<p>Um, yeah, I meant 3200, which is one of the choices that automatically pops up when you shift the slider in Silverfast. Sorry about the confusion. I fixed the blog entry.</p>

[mauro_franic]

<p>Epson scanners (2,000 dpi true resolution) do not resolve T-MAX grain for sure since it is grainless at 4,000 dpi with the Coolscan 9000 (3,850 dpi true resolution).</p><p>Other films may vary.</p>

 

Epson scanners add a significant level of digital noise that may be mistaken for grain after sharpening.

[lubos_soltes]

<p>Thanks for the post Paul. I am going through the same but with the low end V500. Have yet to experiment with the film holder height (ordering variable height holders in a few weeks). Good to know that the Epsons do produce decent results.</p>

<p>Looks like I could get into the 10-12 MPx range with 645 film size.</p>

[paul_bohman]

<p>I think 10-12mp is probably a good estimate, but I haven't spent enough time with 645 format negatives to say for sure. I also haven't tried that model of scanner, so your results may vary. Be prepared to spend a lot of time getting to know the scanner's capabilities, including the software. I almost always lighten the shadows a little, and lower the contrast a bit. If the shadows are too dark in the original scan, it would be better to re-scan the negative than to use Photoshop to try to pull the values up beyond where they want to go. I can always adjust the colors and contrast later if I have to.</p>

[Tim_Lookingbill]

<p>I was expecting a lot more resolve from a scan of large format film especially of a subject that close to the lens. Mind you I don't use a scanner any more nor do I have the luxury of shooting with a large format system digital or film, but I've gotten far more detail and resolve from shooting my own eyebrow hair and skin with my 6MP Pentax K100D than what I see in those close crops.</p>

<p>Something isn't quite right about this and I suspect it has something to do with the Epson's diffused light source which has been the limiting factor IMO with all low end scanners along with the lens.</p>

<p>I saw a scan of Kodachrome 64 off a Howtek scanner which used a halogen light source and was amazed at the amount of color differentiation sensitivity captured in skin tone. You know skin to look real has to have a broad range of tertiary colors going from warm pastel pinks and beiges to reddish browns to caramels in order to not have the skin look like it's made of plastic or have a gelatinous patina. In those close crops of skin detail and eyebrow hair even on the Schaub blog I don't see those subtle differentiations.</p>

<p>Maybe I'm expecting too much from large format, but I've seen close crops of 8x10 negatives shot of Yosemite and Grand Canyon landscapes online that resolved tiny trees and shrubs off in the distant horizon. And here we're talkin' a close crop of a model's skin and hair only feet away from the lens. Something's not adding up here.</p>

<p>Do a harsh downward curve on on the original scan to accentuate micro contrast of the pores of the skin to see how many different hues of skin color there are. Maybe it's the subject that's the issue unless the model's skin is made of one color of translucent gelatin. I mean I was expecting to see fine upper dermal vein capillaries with that close up of a shot.</p>

[StuartMoxham]

<p>It's 6x7 medium format Tim so it probably is not as good as large format would be.</p>

[paul_bohman]

<p>Tim, if you're comparing the resolution of 8x10 inch negatives to negatives that are about 2.2 x 2.7 inches and expecting them to be similar, you've set your expectations way too high for medium format.</p>

<p>As far as the skin tones go, though, I took photos during this same session with a Nikon D7000, and though the Nikon performed very, very well, Portra rendered subtleties in the skin tones that the D7000 did not, even after playing with white balance and other color settings. I've worked with a Nikon D3x and a Hasselblad H3D quite extensively (though not during this photo session), and though both are excellent cameras, I still like the Portra color palate for skin tones better than the digital files. The digital files may be more accurate, in a clinical sense. Maybe. I don't know. But the Portra has a bit more life to it. Maybe you can't see what I see in the skin tones of the Portra photos, but they have more variation in the colors than the digital files do.</p>

<p>Also, keep in mind that the depth of field in my photos was extremely shallow, about as shallow, or maybe even shallower than the 85mm 1.4 lens that I was using wide open on my D7000. The contours of the face are enough to soften some parts of the photo. And, as I said in my blog, Portra is detailed, but not crispy. The information is there, and it's clear, but it's not as hard-edged as digital files. With portraits, that can be a good thing. The texture on her skin in the Portra file is more detailed than the D7000 files, even though the D7000 files have a feeling of extra crispness that the Portra does not. Maybe you prefer digital crispness. Sometimes I do too. With a portrait like this, though, where I intentionally kept a lot of things out of focus, the detailed-but-softer look of Portra suits it very well.</p>

[Tim_Lookingbill]

<p>I don't think that would make much difference, Stuart. The model was still very close to the lens.</p>

<p>It's about scaling size and amount of detail distributed across an area of film that is larger than 35mm format which I've even seen just as much detail due to the subject being that close to the lens.</p>

<p>I still expect a lot more from even 6x7cm film which is nearly 3 inches on the long end.</p>

<p>The 8x10 negative of landscape detail was picking up tiny trees that looked about a mile away. It isn't the size of the area of the negative that matters but the size of the detail captured. That model's skin detail is equivalent to the size of those tiny trees is what I'm saying.</p>

[Tim_Lookingbill]

<p>And as for DOF, I was examining the sharp portions of the skin <strong>color detail </strong>which I'm referring to the amount of color variation the film is sensitive to picking up from the reflective/absorptive nature of hot lights on translucent skin.</p>

<p>These variances of what is reflected back and what is absorbed is what gives the illusion of depth and can be mimiced in Photoshop by layering color variances of multiple layers on top of one another similar to the effect of adding depth with a gloss varnish. Ever apply varnish to wood and see the added depth and richness? It's not that it just gets dark, it's the variations of brown (tertiary) tones that tells you the wood is not made of plastic.</p>

<p>This is why I asked you to apply either a downward curve to accentuate the skin detail or copy the image to a layer and set it to Multiply blend mode and look for the color variance detail down into the pores of the skin. That's how you know how sensitive film is to reflected color off of natural surfaces.</p>

<p>It's not noise or grain and if that's all that's left to see then something is off somewhere up the process chain.</p>

[paul_bohman]

<p>If you want to go through all of that with my files, go ahead. You've got access to high enough resolution files on my blog to do that. I've looked at the digital files and the Portra files, and they're similar, but I give Portra the edge in terms of pleasant color rendition of skin tones. That's my subjective opinion. You may disagree if you like.</p>

[Tim_Lookingbill]

<p>Not disagreeing with you, Paul.</p>

<p>I'm only telling you what I'm observing compared to what I'ld expect from a medium format film shot of a subject that close up. </p>

<p>My observation based suspicions are that there are subtle differences in scan captures between using an Epson flatbed scanner compared to a high end Howtek that I myself couldn't quite define at first until I examined your Epson scans. I get the impression the Epson, for its price, is an adequate device for digitizing medium format film, as its been established many times in the past in discussions of this sort except no one could quite define what those subtle differences were.</p>

<p>There were only vague arguments made similar to film vs digital claims that there was just something they couldn't put there finger on why the Howtek and similar high end scanners excluding the Imacon, gave better looking scans. That kind of explanation didn't help anyone. It didn't help me and is the main reason I no longer shoot film besides the expense.</p>

<p>I'm just saying I think your 6x7cm negative has a lot more detail than what the Epson can deliver. I know that doesn't help in this situation because who can afford a scan from one of these high end scanners with superior optics, light source and software anyway.</p>

<p>Heck, the thought crossed my mind what kind of capture you'ld get shooting the 6x7 neg with a high resolution DSLR by attaching the neg to a black mask surround and having it backlit by direct sunlight and expose for the highlights which means, since it's a negative, you could overexpose grabbing a lot of shadow detail. </p>

<p>I just want </p>

<p> </p>

[paul_bohman]

<p>I was just experimenting with shooting the negative with a DSLR and a view camera macro rig this morning. I can confirm that I get a slightly sharper image that way (not by a lot, but it is noticeable), but color management is, so far, impossible. I downloaded a trial version of SilverFast HDR, which lets users convert tiffs and jpgs into positives from photographed negatives. Maybe I'll figure out how to make that work eventually, but so far the colors are not reproducing well, because of the issues with the orange color mask. It looks like an effective solution for black and white though. Maybe for color too, but it's going to take a lot more work and experimentation to get there, if it's even possible. Part of the problem, I think, is color shift on the strobe I was using to light the negative: Alien Bees AB800. I experimented with several exposures, and the color balance is not the same on them as I change the power setting on the strobe. I may have to invest in an Einstein for color consistency, or figure out some way to make it work with what I currently have. Using sunlight would cause the same problem, because of the color shifts by time of day, season, weather, etc.</p>

[Tim_Lookingbill]

<p>I don't see anything wrong with the overall color of your scans. That's not my point.</p>

<p>I'll show you what I'm talking about. This is all based on ten years of observing the effects of light on colored surfaces. Really it started back in the early '80's studying print reproduction as a graphic artist/illustrator when I was testing light's effect on the absorption and reflectance quality of high grade $15 an oz. Windsor Newton pigmented watercolors against Dr. Martin dyes lit by window, tungsten vs fluorescent light on quality rag watercolor paper. Lights make a big difference on how these looked.</p>

<p>Then when digital came I started seeing this effect in scans of what looked to be early 1960's 5x7 inch contact prints on plastic like cibachrome in my first restoration job (one shown in my PN bio pic) using an Agfa Arcus II flatbed that claimed 1 billion color sensitivity capability.</p>

<p>Didn't know what that meant and thought it was just marketing hype until I started editing those 5x7 scans in Photoshop 4 and was amazed at all the color detail captured <em>off a print </em>which is very similar to what's shown below. This has nothing to do with resolution but color sensitivity and I'm seeing a lack of this kind of detail now on my newer 4870 Epson scanner that's now kept in the closet.</p>

<p>Below is a close up shot taken with my 6MP Pentax K100D DSLR about a 1 1/2 feet from my hand holding a Pink Lady apple lit by direct sunlight. I applied a darkening curve that gives a rendering similar to the type of linear scan I got on my Agfa Arcus II scanner. But this darkening curve is really to bring out and amplify the amount of color detail sensitivity that I expect from a 6x7cm medium format film scan and which I'm not seeing in yours and other's scans using Epson flatbeds.</p>

<p>Take note what I wrote about how color detail is determined by light and the lack of detail from a lack of light in the shadow detail. This is what I've observed about the effect of the nature and quality of light has on delivering reflected detail to a sensor or film base. I'm sure film bigger than 35mm can deliver this kind of detail in spades, but I'm not sure what the cause is and if it has to do with Epson or light source or maybe software.</p><div></div>

[rjm photo]

<p>I also have final found the "right combination" of mount and glass and now with my Epson V700 I get results comparable to the labs 15MP scans for which I was paying $25 for a roll (no developing). With either Rolleiflex or the Hasselblad, the prints are extremely sharp. I have not done a lot of color work with the scanner as I have the lab scan the Ektar when they process it. But the black and white film which I develop at home and I used to take to the lab to scan, I get the same results or better with the V700. I apply very little unsharp mask when I process the images in Nikon Capture NX2, less than I use for images taken with my 35mm Summilux ASPH which again is very little (19 or less).</p>

[paul_bohman]

<p>Here is a follow-up post about the same photo session, after having learned a few more things about scanning:<br>

http://www.paulbohman.com/blog/?p=689<br>

I'm now getting cleaner scans at higher resolution, somewhere in the 25-35 megapixel range, I'd say, with Portra 160 in a Mamiya RZ67 camera with my Epson V750.<br>

A high resolution crop is included, so you can draw your own conclusions.</p>

[michael_schomer]

<p>I purchased a Hasselblad 500 C/M in December and have been looking for a scanner to scan MF film. I currently use a Plustek OpticFilm 7600i Ai to scan my 35mm film and am very happy with it. I had been holding out for Plustek's new OpticFilm 120, but after seeing what you're doing with an Epson v750, I'm not so sure.<br>

The Plustek will probably scan at a higher resolution, and do a better job with 35mm, but the Epson is available now and is probably quite a bit cheaper than the Plustek will be.<br>

Decisions...</p>

[michael_schomer]

<p>By the way Paul, I've enjoyed checking out your blog.</p>