Category Archives: scans

Big Film, Big Picture

100713_4692_5 / LA props, junk and stuff


I shoot film for aesthetic and technical reasons. Aesthetics are covered by quality of films and lens-sets. The films are far more sensitive and responsive than digital, the coverage is wall-to-wall, and is far more robust than a digital interface. The optics and chemistry of the lenses are more interesting.

I’ll discuss the mechanics of scanning further along, but first, some fun stuff. Here’s a comparison of various digital sensors compared with medium-format film.

Imaging area of film and digital sensors

Simplifying matters even further, compare the imaging differences between 35mm and 6x9cm:

35mm imaging areas compared to 6x6 and 6x9cm film

Already you can see that with a larger field, grain is reduced (if that’s a concern), but detail can be recorded, and extracted at will. Absent from this discussion are medium-format digital systems, whose costs are stratospheric compared to readily available film interfaces. But hey! If you have $30,000 burning a hole in your pocket, or can bill the client your rental day rate, go for it.

Film, and why I love it

  • every image is high-res
  • film has a greater range than digital
  • shooting film is shooting “cold”, while digital is always a form of “hot”.
  • Film is visually inspectable, digital is not
  • My film has way more info than your MP files.
  • There is no interpolation, data loss, cutting corners.

Ken Rockwell discusses all this and more in his hilarious takedown of digital, here.

Cameras: ways of seeing and recording

In 2000 I read an interview on a retired Washington Post photo editor. The journo noticed his office was lined with thirty years worth of cameras. When asked about them, the editor said [and I paraphrase] that “95% of the news images we see are made with Canon and Nikon glass. These lenses are very good, but not the only way to see the world…”

That pole-axed me. It made perfect sense. And it changed the way I looked at cameras, and later, film.

In 2008 I went to a Walker Evans retrospective at Stephen Cohen Gallery in LA. There were images I’d grown up with, but now new and fresh, literally 48″ wide. There was detail I’d never seen before. Why?

The lens saw it. The film recorded it. But the reproductive media of the time couldn’t convey it. Photo paper, printing technology, paper chemistry all were not capable of reproducing these images. Digital scanning brought all this forward to a new century.

Walker’s lenses were probably 40 years old at the time. Brilliant optics. Going back further, to the dawn of photography, those lenses killed. In 2005 I saw a Southworth & Hawes [c 1845] daguerrotype  show in New York. The image quality was utterly brilliant.

I began to shoot with demi-vintage cameras, and getting very familiar with the eternal Sunny 16 rule, to sharpen my mind. Also at that time, I was working at the Workbook as a pre-press tech, and saw hundreds of pro-shooters’ work a year. It showed me what I could do, and more importantly, what I didn’t have to do.

The Royal Scan

The luxury of shooting a large juicy negative never pales. The fun starts in translating those negatives into editable digital files. Either you do it, or someone else does.

In 2008 I bought a Nikon LS9000 ED film-scanner to access 35mm negs I’d shot back in 1979, and current 120 negs. At $2199 these machines are not cheap. Rationalizing that the unit cost per scan of that project was $10 ea, if I scanned 250 images it would pay for itself. It did.

The next thing I realized was that the supplied 120 film carrier was inadequate. What you really need is the optical-glass neg carrier which sandwiches the neg flat. I ordered one from B&H for $250. An occasional Newton Ring occurs, but its all in focus.

Nikon then announced that their NikonScan software would not be updated to the Mac OS 10.5. I started using the Hamrick VueScan.  More details here.

Hamrick VueScan is not easy, but not thoroughly impossible either. You conform to it, not the other way around. Then you get on with it.

In time I discovered that some scans I’d gotten from labs were horrifyingly bad. Like magenta shifts in color files, crushed shadows and blown-out highlights in black and white files.

Horrifying lab scan on left, my un-curved scan on right.

Details below:

Note crushed tones and blown highlights on left. Much more detail on right, which can be shaped.

I nearly had an aneurysm when I saw the image on the left. Had I really forgotten everything I’d learned? And then on the right, when I realized I’d been screwed. The image on the right is before I went in and recurved it slightly to give it more depth. That lab is no longer in business.

6×6, 6×9…oh, what the hell!

In 2009 I began shooting with the gorgeous Fuji GSW690iii (shown below), which yields massive 6×9 cm negs on 120 or 220 stock. I was having problems pulling effective preview scans of 6×9 negatives. The last time I would pull a preview scan of one neg, flip the strip, preview the other one. This is time-consuming and very annoying.

Big G-One comes home to find Little Miss Olympus 35RC being romanced by hulky Mr Fuji GSW690iii. This is one of three Fuji medium format rangefinders. 690 image size: 9x6cm. The 65mm lens corresponds to a 28mm wide-angle lens in a 35mm format. This beast has no batteries, no meter, and lives in a Zip Code all its own.

Scanning medium format negs is a different mechanical process than 35mm film. The 35mm carrier is physically indexed with 2 rows of 6 frames each. The medium-format carrier is blank, in both the default version and the way-better optical glass carrier. You have to provide the correct numbers in the VueScan interface to make proper previews and scans, because the scanner does not automatically know if its going to scan 6×4.5, 6×6, 6×7, or 6×9.

I looked for answers. The VueScan manual was opaque. A salesman at Samy’s Camera in Pasadena archly told me that the scanner would not batch scan all 3 negs in the tray, so why bother. Thanks.

Finally this spring on, Dave Goldsmith provided the answer

(edit) “It’s a simple fix – use “Input | Frame spacing” – use a ruler to measure the distance between the beginning of the first frame and the beginning of the second frame.

Note that it is is Frame SPACING, not Frame OFFSET! For the 6×4.5 ratio the value for the Spacing parameter that I have used is 48 +/- 2 mm, and for 6×6, 64 +/- 2mm. With those values I get for a strip of 6×4.5 images four distinct frames and for the 6×6 ratio three distinct frames.

So, set VueScan for Batch mode (All or List), set the appropriate Frame SPACING value, click on Preview, and when the previews are done, yes, you will probably have to adjust the crop frame (not a big deal!) for each picture frame. Finally just click on Scan. Considering all of the frustration about this subject (including on my part as well) the recommended method worked perfectly.

OK! This begs the question as to why the hell didn’t Ed Hamrick say that in plain English in the User Guide? Why do User Manuals have to be written in Techno-Esperanto?

Hamrick Vue-Scan, with a black & white neg. Note numbers on left-hand side of image.

My old-school Schaedler rule was put back to work as I measured it out.

For scanning 6×6 negs:

120, 6x6 with frame offsets

under INPUT:

  • BATCH LIST: 1,2,3

That took care of the 6×6 neg issues, on to 6×9.

For scanning 6×9 negs:

120 6x9, with frame offsets

under INPUT:

  • BATCH LIST: 1,2


After scanning the image to its proper folder, I’ll preview it in Adobe Bridge. This is where I use the EXIF template features to mass-tag files. Create the template[s] you want and go for it.

Now your gorgeous film images are in the digital workflow. Color-balance, spot, edit at will. More on that, later!

LA1980: I Build A Book

book cover.

LA1980: book cover.

This post is about  “LA1980: a photo memoir”. Yes, a naked, blatant plug. Bear with me, I’m going to talk about the technical aspects of making this project happen.

Self-publishing a decent proof-quality book has come of age. For any derision about ‘vanity publishing’ I’ll say “demo”. As in musician. How good it looks and reads is up to you. It’s your baby. Treat it with respect, but work it.

Introduction to Self-Publishing

I’d stumbled on in December 2007. The idea was seductive. A closer examination revealed some serious issues.

Blurb uses Booksmart, a proprietary software, as a gating/formatting choke-point. Booksmart  is not easy in the same way a complex program like InDesign is easy. Its a bucket with pre-fab templates you can drag photos into. You have little control over type kerning, formatting, stylesheets, etc. Which are all the tools I need to work with. More study was required. All told,I studied the whole Booksmart/IDCS stuff for about 6 months, read all the posts, FAQs, whining. I got to post some of my own later.

In LA, There’s Always A Backstory

“LA1980” surfaced during an interminable studio-traffic meeting last summer. I’d gotten wise to the ways of the massive organization I worked for, and used the dwell-time to sketch ideas in a notebook. I wanted to do a photo book using images I’d shot between 1979-1982.

I’d shot 100+ sleeved rolls of Kodak 5297 cinema neg stock; which was cheap in those days, and I was broke. The neg would be contact-exposed to the pos stock, and slides happened.

Periodically I would look at the slides, and go “Yipes!” because the color had gone seriously magenta. The prints I made back then were on a particularly putrid Kodak stock—soft, more magenta, muddy. The images went back in the boxes, and slept.


First sign of new life was 2003, when I got a used Nikon LS2000 film scanner. The scans from the slides were awful. The negs offered more hope. It was a toss-up between OK and awful. But it wasn’t good enough yet.

In 2008 I bit the bullet and bought the Nikon LS9000 scanner in order to scan my medium format negs. The Nikon scanning software worked fine with the Mac OS X 10.4. All well and good, until my elderly G4 died, and I had to get into a MacPro.

Now I discovered that I had two scanning software choices: Hamrick VueScan or SilverFastAI. The difference was about $600. Since the Lotto Fairy hadn’t swung by recently, I went with the Hamrick VueScan. HVS has a blunt, unfriendly interface. I also looked at SFAI, and its interface was blunt, and ugly  as well. I spent several weeks steaming in circles getting the hang of HVS. Finally it began to make sense, and I was up and rolling on that.

Building the Beast: One Image At A Time

The only coherent way to find out what I had besides what I remembered, was to literally start at the beginning, and scan every roll. I’d put it off long enough, and it was time to man-up.

  • Using a cast-off lightbox, I’d loupe the roll.
  • Pull an FPO scan of the roll, typically 1400dpi at 4×6″ for starters.
  • Implementing a workable naming convention. Now that I was scanning in bulk, and going back to pull high-rez images, I needed to find them again.
  • color profiles were set to sRGB, the default Booksmart colorspace.

When You Name It, You Can Find It

I’m done naming images, its alphanumeric for me. Names, descriptions, tags etc can all be handled in Adobe Bridge using Command-Shift-I, which brings up the dialog box for naming, tagging, copyrights, etc.

Here’s a peek:

Image browsing in Bridge

Image browsing in Bridge

Image 790700_08_06 is frame 06, from roll 08, from July 1979. Variations are indicated as -1, -2, etc. This will make my life easier every step of the way down the line, especially when I’m preflighting the InDesign doc, and swapping out missed lo-rez images.

All images start as jpegs. After the curves are applied, the psd is saved, jpeg is tossed.


Here is a typical image, in the before and after mode:

The raw scan and the recurved edit.

The raw scan and the recurved edit.

I scanned close to 1000 images, and had to work fast, smart, and non-destructive. Sometimes I’d recurve an image 4-5 times over the life of the project. I’d see something I’d overlooked the first time.

The Design/Production Workflow

The book was designed using InDesign. This gives me dynamic updates, unique page formatting, typographic specificity, PDF exports; everything lacking in the Booksmart interface.

Pay very close attention to the Blurb specs. They aren’t joking. The following is contingent on your layout being the exact right size, with standard 1/8″ bleed 4 sides.

  • layout all hi-rez images in IDCS
  • page export pages as singles, w/ bleeds, to PDF-x1a
  • open up PDFs as Photoshop PSD (300dpi)
  • save PSDs as Hi rez PNG (300dpi at 100% image size)
  • import PNGs into BookSmart layout
  • upload to site

First proof came back 6 days after sending it. Examined it,

  • looked at binding [OK]
  • color [OK]
  • trims [aggressive to outside margins].

Readjusted live so it was 1/2″ from trim, fixed pages that needed it, re-uploaded it.


I worked on this book 6 days a week, 8hrs a day from Dec 29 to January 21. It was my job when there was no immediately visible work. I decided I needed to get a project up and running that might have a wide/wider reach that would kickstart other opportunities.

The color is OK as a proof. Nothing matches ink hitting paper. However the advantage of creating crossovers with impunity is big fun.

I’m looking forward to my next book.

Every Pixel Tells A Story, Don’t It?


Quality halftone image reproduction is central to the print process. I’ll set aside vector art for the moment, as those are generated under a different protocol. And yes, numbers help, but they don’t tell the whole story.

Digitally-created images are unlikely to ever see the bright light of a scanner. But they are bound by the same rules as scanned files.

Scans are made from either flat art, or negatives and chromes. Flat art is scanned on flatbed scanners of varying horsepower, or drum-scanners. Film files can be scanned on a variety of scanners ranging from insanely excellent down to consumer-borderline competent. All scanners have their own proprietary anomalies.

For the purposes of higher-end print production, placed image is typically 350dpi at 100% in the layout. This is true across the board: Photoshop, Quark, InDesign and Adobe Illustrator.

(Note: In Quark and InDesign it is far easier to verify an image’s enlargement/resolution ratio than in Illustrator, where you have to dig. Read why Illustrator is unqualified as a page-layout application).

Scenario 1: High and Low

If you import an image which is 350dpi at its native 100% into a page layout scenario like Quark or InDesign at 100%, the ratio is 1:1, and everything is golden. However, if I import the same file and then enlarge it 125%, the image inversely decreases to 280 lpi.

350 / 1.25 = 280dpi

When I’ve encountered this numeric, I’ll flag it as a potential problem. When I’ve called advertisers about this problem, reactions vary, ranging from “uh-oh” to “it worked when I sent this identical file to Rolling Stone/NY Times Sunday Magazine” etc. This is when I tell them that the difference between art-book publishing and web-offset periodical printing is considerable. Images go soft, banding starts to appear, and image degradation becomes more evident. Oh.

However I’ve seen rare scenarios where images fell short numerically, and were saved by accidental or intentional bold contrasts, even halftones, and overall image quality as evidenced by the customer’s target proofs. The layout was given a pass and off it went to be wet-proofed.

But that is the exception. Like the drunk driver going over a canyon rail and walking away from the wreck.

The opposite doesn’t play out the same way. I’ve seen layouts where the original image is 350dpi, but is imported into a layout at 125%, equalling 437.5dpi.

350 x 1.25 = 437.5dpi

This is excessive, as current printing typically doesn’t require this.

On the higher end of grotesque, I’ve looked at layouts where a placed image is 15% of original, and the customer’s target proofs are 13×19. And printed out on warm toothy art paper, which typically will add 5% yellow and warm things up quite nicely. Which is fine—until the wet-proofs come back.


  • Under-resolution images need to be replaced by higher quality files
  • Over-rez files get reduced in either pixel dimensions (Quark) or percentage of import (InDesign)

My G5 Is A Beast, Why Bother?

An eternal production question. Because the monstrous spool file you are generating every time you print files that are massive is costing you time and money. Every additional calculation you make with an image (rotation, skew, horizontal/vertical flips) is an additional burden on the document and a clear sign to the pre-press guy or gal that You Are Ignorant.

Scenario 2: Real and Junk Pixels

For example: I open a customer file which is an 8-1/2 x 11″ page document that has the following image in it, complete with bleed. The placed image bleeds 4 sides in the layout.


Its starting to look like a good day. The file’s native dimensions are identical to the layout requirements.

Now look at the Image Size palette for this document:


Numerically the file is good to go.

Junk Pixels

So what happens when you get a file that numerically checks out, but just looks plain weird?

Imagine an 11 x 17 spread, and an extreme closeup of a young toddler. The child’s face looked like it had been put through a Photoshop Mosaic Tile filter. This is what the previous image would look like if it had been treated the same way;

low-rez scan

My first reaction was “Yipes!”

Was this…

  • a stylistic statement (I’d been to a lot of galleries lately)
  • an inadvertent mistake by a designer (ie, low-rez jpeg not swapped out)
  • a basic technical error.

The advertiser was contacted. We found out that the designer had put the original 2 x 2-1/4″ neg on a mid-level flatbed transparency scanner, and then enlarged it to fill the space available.

The pixel dimensions are identical, but the outcome is considerably different. Seems pretty obvious, but this is a continual travelling partner in the World of Pre-Press.

The Advertiser resent their file with a high-resolution scan and the results were considerably improved.


Image management is a combination of metrics and alchemy. The guidelines I’ve mentioned will go a long way in getting an image to print to the best of its original capacity. It has nothing to do with aesthetics—I’ve looked at many files containing images with proper numerics and zero content. The alert designer, pre-press operator, and whoever else that comes into contact with the file are all obliged to do their homework and make sure nothing falls between the cracks.

Cheap Scans, Cheap Printers… Oh My!

swap meet art

These are good examples of analog-induced banding.

Cheap scans and cheap printers are problematic. I remember examining a file and proof from an advertiser which utilized both. It led to nothing but disappointment and heartbreak as we were always chasing the last mistake.

To wit:

  • It became evident in the conversation that the negs the customer used were not drum scanned, but scanned at a local lab using their med-format neg scanner [perhaps set at a lower scan resolution by the operator, etc].
    I remember that the files were mechanically within range (ie 300dpi), but it is unknown if/and how much they’d been rez’d up.
  • The customer was proofing these images at home with a really low-end Epson Photo Stylus, and thought it all looked great.
  • The custormer began to wonder if they did something wrong in the quality of their scans and proofs.

I explained that there are levels of magnitude in resolution between the original proofs, the returned wet proofs, and what they see on screen. Right there there are three modalities. Which one is right, and which one is producible is the big question.

Cheap printers interpolate the living daylights out of whatever they’re fed. They are designed to hide banding and other low-resolution file deficiencies— like digital snapshots and other casual documents. These organic defects will lie in wait and re-emerge when proofed from a high-caliber proofing device.