The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Scanning + Intrisinc Character of a Lens/Exposed Film

t0tor0

New member
I have been trying out some lenses recently. I do my own developing (usually HP5+ and TriX), and then scan using a Nikon 8000ED with SilverFast. Not an expert but have been doing for the last 3 - 4 years.

Sometimes I wonder what is the intrinsic character of a lens or film, and how do I get them. There are so many controls in the software. There is contrast, dynamic range, different film profiles (NegaFix), etc... And alot of time, changing a different profile gives a completely different look. eg. NegaFix

When it is all neutral or "zero" (which i am not sure it actually is), the film comes out rather flat, which doesn't seem to look like what it should be when printed either (comparing to wet printing here).

Just a thought here and perhaps we can discuss and share.
 

Valentin

New member
I am no scanning expert but I'm looking into it since I will be starting to develop my own film.

My opinion is that is desirable to get a "flat" scan since you will then adjust the image per your liking in LR/PS or whatever software you use. Those will give you more options compared to a scanning software.

I guess they include it for people that don't want to fiddle more with it after the film is scanned. I prefer to use LR/PS for post production since they were build with that in mind.
 

Cindy Flood

Super Moderator
I use Vuescan and a Nikon 9000ED. I go for the flat scan. I disable as much as possible in the scanner software (I don't sharpen or use filters). I do all of the adjusting in Photoshop. The intrinsic character of the lens or film will stay intact with this method (unless I choose to alter them in Photoshop).
 

pfigen

Member
I've been scanning black and white for many years now, and I've come to the same conclusion as the rest. Scan the full range of the neg and then manipulate tonally in Photoshop. In the early days, it was more efficient and much faster to make a scan that was as close to final output as possible, as the computing power for Ps was so limited. We did as much as possible from setting exact endpoints and contrast to sharpening, in the scanning software. Now, I do a much more simplified full range, non-sharpened 16 bit capture that is then finalized in Ps.

Compared to analog prints from the same neg, the scans I make today always have more recorded detail - that is, there is less generational loss in the scanner optics than in the enlarging lens - and a more complete and usable tonal range to begin working with.

Anytime you put another lens in the equation, it has the potential to add its own signature to the mix, but once you scan and tonally correct and conservatively sharpen, you should be looking at a pretty fair representation of your film/lens combination.

I can remember back in the late 1970's spending what seemed like an inordinate amount of time trying to find the best enlarging lenses and aligning the old Beseler, Omega and Durst enlargers and trying to find the exact sweet spot of the lens - the point at which the lens was both sharpest and was sharp into the extreme corners of the neg. Glass carriers and four more surfaces to clean. Newton's Rings. Kodak Film Cleaner.

Todays scanner optics largely get around all those issues. The hybrid technology of shooting black and white film for digital scanning and printing yields results that we could never achieve prior to the technology of the last twenty years.

The black and white prints I'm making today have a level of detail and luminosity that even after decades of making great prints in the darkroom, was almost never at the level of what I'm seeing with the modern tools.

What you will find out is just how good films like Tri-X, HP-5, Plus-X and especially, T-Max100 are, and just how good your lenses really are.
 

t0tor0

New member
Thanks for all your replies.

First of all, I am a bit surprised. I usually try to do right on Silverfast. I assume that since the negative is in the scanner, it contains all the RAW data and that means, will have the most room for manipulation. (eg, higher intensity light for shadows and lighter intensity light for highlights). Once digitized, the file is much more limited.

Nevertheless, I will try saving an initial minimally touched-up image as 16 bit JPG. Perhaps 16 bit does have more than enough room to play about. The only problem with 16 bit is that it cannot go directly into Aperture (well, at least when I tried). If I do all the manipulation on Silverfast and save it as 8 bit JPG, it goes into Aperture straight away. Now with 16 bit JPG, PS have to come in between.

As for the intrinsic character of a negative/lens, I guess it will always be this way. Just talked to a friend over this and we concluded that even in wet print, we are still exposed to other factors like kind of paper, enlarging lenses, bulbs, etc... too many factors so guess we just have to take the whole package as it is!
 

pfigen

Member
"we are still exposed to other factors..."

Not to mention burning, dodging, developer concentration and timing, rubbing hot water on portions of the developing print to "encourage" that area.

No one of any experience ever makes a straight print. Hell, the image of Willie Dixon on my website used to take four or five hours to make a good print and maybe all day to make a great print, but after drum scanning, digital manipulation, output to new neg via LVT, a new great darkroom print can be had with minimal hardship and time.

BTW, jpeg does not support 16 bit per channel - only tiff, psd, psb and maybe one or two others. Photoshop CS5 will let you save a 16 bit tiff as a jpeg, but it's really just saving a copy and converting it to 8 bit for you automatically. What are you using Aperture for, and can it not see a 16 bit per channel tiff? It can surely produce them from digital raw files.
 

t0tor0

New member
Thanks for the input. There is an option in Silverfast to save as 16 bit JPG. Maybe this is why Aperture is not recognizing.

I will try 16 bit tiff next time. Hopefully that will do the trick. Or perhaps I will realize that I need a more powerful computer by then :D

Btw, for Black and White, do you recommend to scan on grayscale or color and convert to grayscale in PS? I read somewhere that scanning a BW negative in color then coverting to grayscale will yield even more tonal range. What are others' experiences here?
 

t0tor0

New member
@pfigen:

i am using Aperture to archive all my digital images. i guess lightroom works the same but i started off using Aperture since the begining so am too lazy to learn lightroom. Moreover, to re-catalogue all my stuff will be some work!!
 

Tim Gray

Member
Scan 'flat' and do your adjustments elsewhere. Unless your scanning software gives you the kind of control and feedback that Photoshop does. Which it probably doesn't unless you are using a drum scanner or a Flextight or something similar.

That being said, in my experience, 'flat' has different meanings depending on what your are scanning and how you are post processing. For B&W, just try to capture all the detail, bring it into PS, adjust contrast with curves and/or levels (using layers and layer masks if it is appropriate), and sharpen at the end. With my scanner, I've seen no benefit to scanning B&W in color and converting to B&W in PS. I have noticed that my blue channel is sharper than the R or G channels, so in Vuescan, I set "make gray from" to blue. Your scanner and software may be different though. You can ignore the following if you are just doing B&W.

Color is a bit trickier because in normal operation, adjusting contrast also adjusts saturation. This is all good and fine for normal adjustments, but trying to go from a really flat scan to normal contrast can jack up saturation too much, in my opinion. Adjusting contrast with the blending mode set to lightness can leave photos a bit too desaturation; something in the middle is usually called for.

Negative film profiles aren't really what's needed. Not that they hurt, but what really needs to be done correctly is the inversion process. In my case, on a Coolscan, Nikonscan does this ok, Vuescan is pretty poor, and Silverfast is somewhere in between. I get the best color with Vuescan doing a linear scan and then doing the inversion in Photoshop using Colorperfect. In my tests, I CANNOT achieve the same color from a Vuescan negative scan post processed in PS as I do from a Vuescan linear scan that has been processed with Colorperfect.

You should be able to do this in Photoshop without Colorperfect, but Photoshop does the inversion incorrectly AND has problems doing large gamma changes. You can also do the inversion with free programs like ImageMagick, which is what I'm working on... Maybe this stuff isn't relevant to scans done a Flextight or other expensive scanner, but it is if you are using Vuescan and to a lesser extent Silverfast (in my opinion).

Lastly, linear scans aren't really RAW like DSLR shots are since most scanners don't have Bayer arrays, so no RAW 'developing' is needed.
 

pfigen

Member
My discussions of scanning in this thread have been limited to scanning black and white negs, as per the original poster. Tim makes valid points regarding color negs, which do need to be treated differently.
 

Tim Gray

Member
Oh and to the original question: I wouldn't worry about having your lens/film characteristics show through after scanning. I don't buy into that stuff that once you scan a film shot you lose all the film attributes. Sure, if you convert Velvia to B&W, you might lose the Velvia feel, etc., but if you stay somewhat true to the original frame during post processing, most of the character (lens or film) comes across in the scan.
 

t0tor0

New member
Thanks for so much information. Really enlightening.

Guess we are going astray of the man topic now... hope it is ok...

What about scan resolution. I always scan at max resolution that my scanner can allow me (their real native max and not interpolated). 4000dpi for my Nikon 8000ED. This takes some time really. I think about 4 - 5 minutes each frame. Sometimes I wonder if it is really worth the wait. A 2000dpi is significantly much faster. Any advise for this?

Is there a "DPI Recommendation Table" recommending optimal scanning DPI base on type (or ISO) of film?
 

Tim Gray

Member
I scan frames that I really like at 4000. All the others that are just ok but I still want scanned I scan at 1333 dpi. In Vuescan, it is significantly faster to do this. It does give a slight increase in grain, especially on a film like T-Max 3200, but it really cuts my scanning time down.
 

Valentin

New member
...A 2000dpi is significantly much faster. Any advise for this?

Is there a "DPI Recommendation Table" recommending optimal scanning DPI base on type (or ISO) of film?
I remember reading (here and other places) about the fact that above 2000dpi there is no more gain in detail. I can't find the exact thread, but maybe a search will help you find it.
 

pfigen

Member
"What about scan resolution. I always scan at max resolution that my scanner can allow me (their real native max and not interpolated). 4000dpi for my Nikon 8000ED. This takes some time really. I think about 4 - 5 minutes each frame. Sometimes I wonder if it is really worth the wait. A 2000dpi is significantly much faster. Any advise for this?"

I wouldn't worry about 4-5 minutes for a 4000 ppi scan from 35mm. The Howtek 8000 is a tad over seven minutes and a Howtek 4500 about twenty-one.

It's easy enough to test to see where your scanner maxes out on resolution. The tests I've seen on those suggest that a 4000 ppi scan has about 2700 ppi of real data, so, depending on the image, you may or may not see a big difference. What I do see on the drum scanner is a real difference in how the grain is rendered. So, do two scans - one at 4000 and one at 2000, then rez the smaller one up and compare side by side. That will tell you what you need to know.
 

pfigen

Member
"I remember reading (here and other places) about the fact that above 2000dpi there is no more gain in detail. I can't find the exact thread, but maybe a search will help you find it."

You know what they say about things you read... Here's the deal: It all depends. It depends on the film, the scanner, the subject, etc.

In actual comparative drum scans of a really sharp Tri-X 35mm neg, I could see a minute improvement in detail between a 4000 ppi and an 8000 ppi scan, but because there are no resolution increments between 4K and 8K, it's hard to say where the limit is. That's Tri-X. T-Max100 is quite a bit sharper and much finer grained. The grain structure of that really does support 6-8K scans.

Velvia 50 is extremely sharp. In one test I did, I scanned a really sharp Velvia 35mm at 8000 ppi and then put the film in my Beseler 4x5 enlarger, racked it all the way up, and looked at the aerial image with the expensive Omega grain focuser. There was more detail on the film viewed that way than I could see even on an 8000 ppi drum scan.

Color negs, on the other hand, and I haven't tested the new Ektar yet, really are mostly in the 1600-2000 range for optimal scan resolution. Scan color neg at too high a rez and you just get huge grain.

The one thing that drum scanners have that CCD scanners don't is the ability to set the file size and resolution separately at the time of the scan. We do this by de-coupling the scanning aperture from the hardware resolution settings. Normally, when scanning at 4000 ppi, the scanning automatically sets the aperture through which the image is passed to the sensors to approximately 6.35 microns. That 6.35 micron hole is what determines the level of detail that is sent to the photomultiplier tubes. The speed of drum rotation and how fast it slides past the lens determines the number of megabytes. Multiply that 6.35 micron aperture by 4000 and as if by magic, you come up with 25.4 mm or one inch. That's your 4000 ppi.

When scanning grainier films, you can set that aperture independently. You might need to have 4000 ppi worth of file size, but need something less grainy. By manually setting the aperture to something that more closely matches the grain structure of the film - 7.9, 9.4, 12.7, 16, 19, etc. you scan for the file size you need but only record down to the actual film grain. Through trial and many errors, I've found that most color negative films scan best between 13 and 19 microns, or between 1333 and 2000 ppi.

I hope this description wasn't too complicated, but I think it helps people visualize what's going on in the scan.

So, don't believe everything you read, not even this. It's really easy to test any of these theories for yourself and see what works best for your needs with your film and your scanner.
 

Cindy Flood

Super Moderator
This is very good, interesting information. Thanks to everyone who is participating. I hope you keep adding information to this thread.
 

pfigen

Member
Cindy - None of this information was available when I started drum scanning about 12 or 13 years ago. I wish it had been. It would have saved me a lot of time. Most of this is common sense, but it takes making a lot of test scans and carefully examining them to understand what's going on. The more you understand the process, the easier it is to make good decisions regarding scanning.
 

t0tor0

New member
Thanks for all the information! Learnt a lot and will try out some tips I gathered here.

Would like to know more about color scanning. What about color profiling? What is a profiler?

I have purchased IT8.7/1 Scanner Calibration Targets. SilverFast has a calibration feature to calibrate the colors. So what I do is that I choose the target (some are made for Velvia, some for Kodachrome, some for Provia, etc..) depending on which film type I want to scan. I can understand why the light source needs to be calibrated. However, I am not sure why there are different target for different film type.

I was told on top of VueScan or SilverFast, etc..., a separate profiler software can be used to yield more accurate colors. How does this work? Hope someone can explain and enlighten!
 

pfigen

Member
When scanning color transparencies or prints, the scanner records the colors as RGB values, but the scanner may not record the range in a linear fashion. It may emphasize some or scan with an overall cast.

Using a scanner profiling target can help you to get more accurate results. The idea is that you scan the target in a special profiling mode which disables any current scanner profile, while adjusting the scanner controls to get a full range on the target with a little room to spare - that is - you don't want the whitest patch to end up being 255 or the blackest patch to be 0,0,0 - you want to be slightly conservative so you don't clip colors or tonality. It may take several test scans to arrive at the proper scanner settings. Once you do, you have to save those software settings so your subsequent scans are made with the same settings.

Once you have the scan, bring it in to Ps and make sure that it's straight and it doesn't have any spots or hairs going across the color patches. If it does, carefully clone them out so all the patches are clean. Then import the scan of the IT-8 into your profiling module. It's usually pretty much a push button process to generate the profile.

The next time you scan, make sure you load your new custom scanner profile into the Input profile section of your scanner software and then load the same settings you used to make your profile and scan.

The profile will take into account how your scanner "sees" the film and correct the scan to be very close to the original, all without tweaking the scanner controls. You may, however want to make changes to the scanner control when scanning specific pieces of film - film that has a color cast or that is over or under exposed. That's okay.

The scanning software should give you a choice on whether to embed the input (scanner) profile in the final tiff file or convert to a chosen Photoshop working space. Use whatever choice you're most comfortable with, unless you happen to have a Howtek drum scanner using Trident, which has a bug in the convert to profile function. I do those conversion after opening the files in Ps.

Sometimes you'll find that the scanner profile makes scans that are too dark, and there is a workaround for that. Take a copy of the scan of the profile target, open it in Ps, and using Curves, place a single point in the middle of the master RGB curve and pull it down, darkening the image slightly. Save that and use that modified target to make a new profile. The profiling software will be tricked into making a scanner profile that provides a lighter image. You can do alternate versions similarly if you feel you want more or less saturation or even want to emphasize or de-emphasize a certain color on a consistent basis. Just make sure you modify the target scan in the opposite direction of where you want it to go in the final use.

Using an IT-8 or Hutchcolor profiling target on your scanner is just like making a custom profile for your printer - it "fingerprints" the scanner and provides a known numerical state of how the scanner records color.

There are scanner targets for reflective prints, for Velvia, Provia, Ektachrome and Kodachrome, among others. In theory, you would want a custom profile for each type of film, but in practice, a single profile made from an E-6 target will cover all of your E-6 films, while a profile from a Kodak reflective target will work well for most prints.

Kodachrome is always the odd man out. It generally does not scan how you expect it to, usually resulting in bluish shadows using E-6 targets. Since I've never gone to the trouble of procuring a Kodachrome target, I just override my scanner settings, and take the blue out of the black and that seems to put everything just where it wants to be.

A typical Ektachrome rebate edge will read 8R, 6G, 5B. A Velvia rebate edge will be more like 5,4,3, but Kodachrome comes in with the blue channel up around 20-25. The dyes are seen differently. You can compensate with a Kodachrome profile or you can do it manually.

Peter
 
Top