How should the beginner scan? (Nikon Coolscan)

Hi,

I have made a series of pictures that a friend of mine, who also is a professional photographer, thinks is good enough to show to a local gallery. This situation is new to me, and I am of course flattered. Now, I should get going at once and make a portfolio. But how? The technicalities make me hesitate.

This question should be about scanning, so lets forget everything else for the moment. (I have an adequate computer and a pretty good computer screen.) The negatives are from Tmax 3200, Neopan 1600 and Ilford 3200, all ordinary 135 film. I have a Nikon Coolscan 4000, which I so far have used only for scanning for the web and I really dont know much about it. Software: Nikons own.

So I checked the internet for help. I would need a simple and robust method, but the internet is full of opinions, experiments and self-appointed experts. As long as I dont know the basics, this does not really help.

Maybe someone of your could point me towards basic guidance. The goal, for now, is only to make credible scans that can be used to print the portfolio and the exhibition, should I be so lucky. No alternative methods, no sorcery. Just something that a beginner can do. (Yes, I checked the sticky thread im this forum but I'm afraid I didn't find it helpful enough.)

Thanks in ahead, everyone.
/Erik Petersson

By the way, here is a link to some of the pictures if you should be curious:
(there was a url link here which no longer exists)

Your scanner will work fine. I used to have a similar Nikon scanner (V or 5000; I forgot which). I would just start practicing. You'll get it pretty quickly.

Insert strip of film, select b&w, set auto expose, auto focus on with all other controls off or neutral and scan. For critical work, preview and adjust gain to maintain highlights as appropriate. If the scans come out as you like, you might want to save the settings to be able to recall and apply on other scans. TIFF is best and max quality JPEG is usually good enough.

After you make the scans using auto exposure, look at them in Photoshop and make sure the shadows or highlights are not clipped. If they are you will want to turn off the Auto exposure system in the scanning software and set the white and black points manually. I've never used the Nikon software, so I don't know if will be needed. I do know on the other packages I have used scanning on Auto results in clipping.

Also, make sure you use 16 bit scanning, turn off all sharpening in the scanner, and output as a TIFF (not JPG). That should give you a good raw scan with which to start your adjustments in Photoshop. Feel free to email me a small version of a file if you want someone to look it over and see if there's anything obviously amis.

Edit: Forgot to mention, turn off the Digital ICE, since it's not compatible with black and white film. Cleaning your film well before scanning will save you hours of spotting later on.

Nikon Scan should not normally clip on auto. The ICE will only work on dye based films. In general this means colour slide, colour neg and c-41 process black and white. Have fun!

Thanks a lot, this is the kind of advice that I need.

One more question: should I always scan with the maximum amount of pixels per inch? (I have read discussions that the pixels will be rendered differently, many pixels per inch may result in harsher grain. But my pictures have large grain anyway)

You can either scan at the res you need at the time and save time and disk space or you can scan at high res for the sake of not having to scan again. Both are valid philosophies. I don't think you will see a huge difference in grain character or whatever. I don't see much difference on my Coolscan 9000 between 1000dpi, 2000dpi and 4000dpi.

Sam

- Scanning Basics 101 - All about digital images
- I expect Google will turn up lots of useful info

I find that scanning at 4000 dpi helps reduce the appearance of grain, particularly with some of the grainier B&W films. At least using Vuescan with my Coolscan V. For portfolio type work, I think I'd want my scans at the highest resolution. I use lower resolutions for quick scans for the web.

This is essentially the process I use (with Vuescan):
Scanning B&W negatives with Vuescan (The SOPA thing will go away at the end of the day)

Thanks again. I will check the links and read what's there. I'm sure I will meet lots of problems still. Scanning starts in the weekend.

So why do the people on APUG say that DPUG is empty? All my questions so far - five or six over a few years - have been met with good advice within a few hours.

So why do the people on APUG say that DPUG is empty?

you would have to ask that question on APUG

Erik Petersson said:

Thanks a lot, this is the kind of advice that I need.

One more question: should I always scan with the maximum amount of pixels per inch? (I have read discussions that the pixels will be rendered differently, many pixels per inch may result in harsher grain. But my pictures have large grain anyway)

Click to expand...

IMO, it's extremely wasteful and inefficient use of time and space to scan at a much higher resolution than you need. Why tax your computer and fill up your hard drives much more than is necessary? The way i look at it, one is best served by scanning close to the pixel dimensions he/she will need to make the print or Web images. That way you will only need to do a minimum of resampling, and won't be storing vast amounts of unneeded info.

As an example, say you want an uncropped print that measures 8x12 in the end. Your lab can make a fiber lightjet print from a file that is either 200 ppi or 400 ppi. For sake of example, let's just say you want maximum resolution, so you decide you want to submit a 400 ppi file to your lab. Multiply the size of the desired print (8 inches) by the final resolution you want to use in the file that you submit for printing (400 pixels/inch). In the math, the inches cancel each other out, so you are left with pixels. 8x400 = 3200 pixels total across the short side of the image. Since the short side of a 35mm frame is about 1 inch, it is nice and easy to figure out what resolution you should use to scan: 3200 ppi. I'd probably use 4000 ppi to scan, just to give m a nice flat number, and a little wiggle room.

Say you only want to print from a file with a resolution of 200 ppi. That obviously means that you only need to scan at 1600 ppi. I'd probably use 2000 ppi.

Say you only need Web gallery images that are 6x9 on a computer screen. You'll want a 72 ppi file for many monitors. 6x72 = 432. If there is no other possible use for the scans, why scan at a much higher resolution than that? I'd probably scan at 600 ppi in that case.

The Epson v750 manual says that at the highest resolutions, it engages a higher resolution system. Does this matter, or is it marketing hype, since apparently the highest optical resolution of the device is around 2400 ppi?


---
I am here: http://maps.google.com/maps?ll=40.029256,116.510727

The Epson's resolution seems to be optics-limited, so using the higher-quality lens (reportedly) increases the effective resolution. I have my doubts about the 2400 ppi figure, as others have measured somewhat higher and, in any case, the delicate set-up of the machine makes getting a definitive figure almost impossible.

I scan at the highest effective resolution of my scanners. The only exception is on the drum scanner with 4x5 film, 5000ppi creates lines too long for the scanner buffer, so I scan those at 4000ppi. Disk space is cheap, and I never know what I want to do with the image until after I've scanned it. My time is more valuable than the storage space. If it turns out that I know I'll never print it large, then I can downsize it. But that is rare, as usually it's a focus issue that prevents that, and it still looks bad small. Those just get deleted.

Scanning is very much like making an ordinary (film) exposure. The quality and characteristics of the ultimate (printed) photograph are very much dependent on the quality and characteristics of the original scan. In fact, the process of scanning film actually begins *before* you trip the shutter on your camera!

In the traditional darkroom workflow, the tonal range of any print depends on the photographer's ability to calibrate his exposure and development process so the tonal range of a negative most closely matches the tonal scale the paper can reproduce. Scanning works very much in the same way, where the scanner is analogous to "paper", in the above example.

Since your question relates to scanning I'm going to make that the focus of my response, but I urge you to read/listen to this blog post on scanning techniques by Steven Schaub:

http://figitalrevolution.com/2009/05/15/film-scanning-technique-101/

http://figitalrevolution.com/2008/03/20/processing-black-and-white-film-for-scanning-diafine-and-tx/

Most of the tips and suggestions posted will accomplish what you were looking for in an entry-level scanning workflow. But as you gain more experience, you may find yourself quickly outgrowing this, and so I've outlined some more advanced details of the process which I hope you find useful.

Scanning happens in two steps:

(-) 'reading' the image pixel data (acquiring numerical data)

(-) rendering that data as colors and tones (rasterizing -- jpeg/tiff)

Both steps occur automatically (in most common scanning configurations), but understanding this distinction will help you immensely in planning your edits, retouches and refinements.

The first step in the scanning operation is predominantly dependent on your scanner's hardware (as directed by the software). In the second stage, your scanner rasterizes the data from step-1 and outputs the scan in whatever format you specified (jpeg, tiff, etc). While this second stage often too occurs automatically, it doesn't *have* to.

I've been experimenting with RAW scans in Hamrick's Vuescan Pro and Nikon LS-4000/9000 scanners. The idea is to save the RAW image data from the scanner, and 'process' the data into an image file using an external Raw Image Workflow application (eg. Aperture, LightRoom, ARC, etc). This method allows you to process the same scan-data multiple times; each pass is configured to extract detail from different parts of the image (lows/highs/mids/etc). The resulting (raster) images can then be selectively combined in Photoshop, yielding maximum tonal range in ever part of the image.

If this sounds a bit like HDR, that's because it is exactly the same process. The difference is that with true HDR you have actual (different) source data for each exposure, while here you're kind of improvising. Still, RAW files contain a great deal more data then ANY raster format can truly represent and while more laborious, I find this specific method gives me maximum image potential.

I hope this isn't too far over your head, and that it helps you (and possibly anyone else) in their own scanning workflow.

Good luck!

Vuescan's "RAW" files are actually TIFFs in a DNG wrapper. You can get similar results by scanning any film type as Colour Positive (giving a linear tonal response curve) and outputting as 16-bit TIFF with no or minimal adjustments at the time of scanning. Your scan will look fairly flat, but this allows you plenty of scope for adjustment later - and it will be readable in Lightroom and ACR.

artobest said:

Vuescan's "RAW" files are actually TIFFs in a DNG wrapper.

Click to expand...

Wow I didn't know that... I know that ACR can 'process' TIFF files so wrapper/no-wrapper doesn't really make any difference. I'm surprised though. It seems a bit misleading to call them "RAW" files if they don't contain the actual raw scan data (unrasterized)

artobest said:

You can get similar results by scanning any film type as Colour Positive (giving a linear tonal response curve) and outputting as 16-bit TIFF with no or minimal adjustments at the time of scanning.

Click to expand...

I've read that the best way to scan *any* film media was as an RGB positive, but I'm wondering if there is any practical value to the suggestion. I shoot exclusively black and white (negative) film. At 16-bits per channel, an RGB TIFF would be 3x the size of a 16-bit grayscale scan.

The files become incredibly large (storage-wise) and far more resource intensive, easily 500mb in Photoshop for a 35mm frame (16bit RGB @ 2800dpi).

I suppose this is a 2 part question then...

(1) Considering a dedicated film scanner (eg. Nikon LS-4000 or 9000) what's the practical significance of scanning in "positive" mode

(2) Why go with RGB over 16-bit grayscale for a b/w negative?

stormbytes said:

It seems a bit misleading to call them "RAW" files if they don't contain the actual raw scan data (unrasterized)

Click to expand...

Quite.

The reason for scanning as colour positive is that the output has a linear TRC - no software correction is made for, say, the flattening effect of the orange mask. This helps prevent clipping or blocking of shadows and highlights and gives a flatter, but much more "post-processing friendly", scan. (It helps if you manually set black and white points for each channel before scanning). This is why I always scan b&w as RGB, even if two channels are discarded in the final image (I simply use PS's Channel Mixer).

Sometimes it can be quite handy to scan as Colour Negative using Silverfast's Negafix film profiles, but you do need to tweak them before scanning to avoid clipping.

File size isn't really an issue. I'm not sure where you get the 500mb figure - a 16-bit RGB scan of a 35mm frame @3200 ppi is around 80mb.

Vuescan has a multiple exposure option. It will scan the film more than once (two or three times, not configurable as far as I can tell) with different exposure levels. It drills down into the shadows to capture extra detail, and also tries to fine the extra gradations in the highlights.

I don't know exactly how it combines this information into the resulting scan, the documentation is lacking on this. Is anyone else using this feature? Do you have any idea how to control it, and what exactly it is doing?

I normally do a raw scan, with this feature enabled, and scan negs as positives. I use ColorPerfect in Photoshop Elements to convert the neg to positive. I've found this plugin's accuracy in removing orange masks and generating positives, based on it's extensive database of films, to be quite good.




---
I am here: http://maps.google.com/maps?ll=25.041020,121.535978

Glad that the discussion continues. This is a bit more than than I will be able to get into before starting, but I'll hope that I'm still a fast learner...

artobest said:

The reason for scanning as colour positive is that the output has a linear TRC - no software correction is made for, say, the flattening effect of the orange mask. This helps prevent clipping or blocking of shadows and highlights and gives a flatter, but much more "post-processing friendly", scan.

Click to expand...

TRC = Tonal Response Curve?

Orange mask? Are you referring to color negs?

artobest said:

It helps if you manually set black and white points for each channel before scanning.

Click to expand...

Can this be done in Vuescan? I find having to wang the little triangles on the graph very awkward, not to mention imprecise. I'd love if there was a color picker that could be used to set levels.

artobest said:

This is why I always scan b&w as RGB, even if two channels are discarded in the final image (I simply use PS's Channel Mixer).

Click to expand...

Channel mixer? My impression is that the channel mixer was effectively replaced by the black-and-white adjustment layer in the current version of photoshop. Am I missing something?

artobest said:

File size isn't really an issue. I'm not sure where you get the 500mb figure - a 16-bit RGB scan of a 35mm frame @3200 ppi is around 80mb.

Click to expand...

I may have to double check my math. I've also experimented with mf negs so it is entirely possible i'm confusing things. Regardless, 80mb does seem a bit on the thin side of things for 16-bit, RGB @ 2200 dpi (much less 3200)

L Gebhardt said:

I scan at the highest effective resolution of my scanners.

Click to expand...

I think *effective* is key here as beyond a certain point all you get is magnified grain -

I can't imagine scanning a 35mm negative at anything beyond 2800ppi (optical / dedicated film scanner) would yield any useful image data.

artobest said:

Vuescan's "RAW" files are actually TIFFs in a DNG wrapper.

Click to expand...

Quoted directly from Hamrick's user manual:

"If you're scanning several images in one go, scan each image once and save the raw CCD data file. "

URL: Using Raw Scan Files

But then afterwards he explains that "raw" is in fact a TIFF

URL: Output tab

Seems the DNG file wrapper is altogether optional....

stormbytes said:

TRC = Tonal Response Curve?

Orange mask? Are you referring to color negs?

Click to expand...

Yes, to both.

stormbytes said:

Can this be done in Vuescan? I find having to wang the little triangles on the graph very awkward, not to mention imprecise. I'd love if there was a color picker that could be used to set levels.

Click to expand...

I don't use Vuescan but I'd be surprised if there weren't a per-channel levels function.

stormbytes said:

Channel mixer? My impression is that the channel mixer was effectively replaced by the black-and-white adjustment layer in the current version of photoshop. Am I missing something?

Click to expand...

The black and white adjustment layer is for transforming colour images to b&w. It effectively replicates the use of colour filters in traditional b&w photography, with greater subtlety and variation. The Channel Mixer is better for the purposes I mentioned because it can eliminate the noisiest channel altogether (usually the blue channel).

stormbytes said:

I may have to double check my math. I've also experimented with mf negs so it is entirely possible i'm confusing things.

Click to expand...

That's the most likely explanation.