 |
|
|
|
|
Nikon Digital Workflow and Asset ManagementContentsIntroductionThis article describes my workflow for managing files from Nikon digital cameras. It covers the entire process from image capture to archiving. Although I will focus on Nikon cameras and software, much of the information is general and applicable to other camera systems. The goal in developing my workflow was to create an efficient process for managing thousands of existing images and rapidly incorporating new ones. My typical tasks include storing files, retrieving specific images, and preparing images for various forms of publication (fine art print, web, magazine).
This is a work in progress and as new tools become available, my workflow will undoubtedly change. If you have comments or questions please email me.
To get the highest quality digital images I shoot in RAW mode. The main idea behind recording RAW files is to capture the original sensor data before any processing occurs as many operations are irreversible. While it is convenient to have the camera directly produce publication ready images, often one has much more control and can obtain better results by processing the RAW files with dedicated software at the computer. More specifically, for my work I find there are several benefits of raw mode:
During shooting, I select the lowest ISO that will give me an acceptable combination of shutter speed and aperture. I use the following camera settings:
I use either matrix or spot metering. Generally I use matrix metering in time critical situations and spot metering for static subjects or scenes with a high contrast range.
I find that with digital capture accurate exposure is extremely important as even slight overexposure can result in blown highlights and underexposure can increase noise levels. If I am in doubt about the exposure, I will bracket around my best estimate of exposure. For dynamic subjects where bracketing isn't an option, I may slightly underexpose the image to ensure detail in the highlights.
I use several 256MB Compact Flash (CF) cards with my digital camera. Each card can store about 25 uncompressed RAW files and roughly double this number with compressed RAW files. On short assignments, I bring my Image Tank G2 portable hard drive (Figure 1). This is a small self contained unit that houses a hard drive, CF card reader, and a battery. I place a full CF card into the reader slot and press the copy button. It takes about 100 seconds to copy a full 256MB Card to disk.  Figure 1: Image Tank G2. I try to copy full CF cards to the hard drive immediately. However, if I am not able to do so, I flip the card and insert it backwards into the case as a reminder to copy it at a later time. The main advantage of a portable hard drive is its cost compared with buying enough CF cards to last a shooting session (which may be a whole day). At this point (November, 2003) 1GB CF cards are about $250. A 20GB portable drive is about $300. Although a portable hard drive is heavier than carrying a couple of compact flash cards, it is far lighter than a laptop. A CF card weighs about 10 grams, my Image Tank weighs 341 grams (about 3/4 of a lb), and my laptop weighs nearly 2 kilograms (4 lbs).
On extended trips, I bring both a portable hard-drive and a laptop. The portable drive comes with me during the day and the laptop stays at my place of residence. During the day, I will back up cards to the drive and at night I will copy all of the files to the laptop. Although carrying the laptop adds significant weight, it substantially reduces the probability of having a catastrophic loss: I have at least two copies of the files during the trip (more if I have time to burn CD backups). If I lose one device because of theft or failure I still have a backup.
I transfer all of the RAW files to my working computer and I place them in directory named after the event and date. For example, the pictures I took of BABA's book arts jam go into a directory labelled book-arts-jam-2003-10-18.
On the computer, Nikon RAW files are stored with a .NEF extension, which stands for either Nikon Electronic Format or Nikon Electronic File (both are used in Nikon literature).
Digital cameras store the color information in an a scene by recording the level of red, green, and blue light that falls on the imaging sensor. However, in order to display the image a computer must take these numbers and send them to an output device such as a monitor or printer. The problem is that because of physical device differences a monitor might display red=128, green=97, and blue=217 as shade of violet while the printer might render it as blue. Without a standard scale, output devices have no way of knowing what absolute color the numbers should create. This obviously causes problems if you want to view the same image on the different devices. I'd like to make sure that the red in my image that I'm viewing on my Sony monitor is exactly the same shade of red that a photobuyer will view on her computer system which is using a Lacie monitor. Furthermore, there is no sense in color correcting an image unless I'm certain that other people on different systems will see the exact same output. The solution is to use profiles that describe how each device renders colors. Given these profiles, display software can accurately determine the correct numbers to send to a device to obtain a specific color. On my computer, I used the software package Photocal and a Colorvision Spyder (a colorimeter) to generate a monitor profile. Photocal first calibrates the monitor so that it responds in linear fashion with respect to the gamma setting, which relates the voltage of the video signal to the light intensity on the monitor. Photocal then profiles the monitor to generate accurate measurements of how the monitor actually renders the video signal into specific colors.
Note that before doing any profiling and color critical work, you should turn on the monitor and let it warm up for at least 15 minutes.
My goal in editing is to select a set of high quality images that have strong visual impact. Each image that I keep must have acceptable exposure, focus, and composition. I browse the thumbnails of the RAW files in Nikon View. At this time, I rotate the any images as needed (the D100 does not have an orientation sensor) and I evaluate images by viewing them in Nikon Capture. I immediately delete any photographs that are not technically satisfactory: i.e., pictures with gross and unrecoverable exposure errors, blurry subjects, or poor composition. As I browse, usually a few photographs will immediately stand out as keepers and I mark these with a check in Nikon View. As I edit, I view the images fullscreen in Nikon Capture and I often perform adjustments to correct exposure and white balance as needed. Sometimes I will even see how the image looks in black & white before making a final decision. I save the corrections to the RAW file. Note that saving these changes to NEF files does not alter the original image data but instead records a set of processing instructions stored in the header of the file. I often have a set of similars from which I want to select the best shot. I will compare similars side by side in Nikon Capture and I examine details such as focus in critical areas, cropping, and color balance. When I decide to keep an image, I mark it with a check. I go through the entire directory deleting images until only those I plan to keep remain. If I do not have an immediate deadline, I will often wait before editing and I may do the evaluation over the course of several days. The images that make the cut get their own unique number. I use an accession system where each new image is assigned a new sequential number. Another alternative is a classification system where each image is assigned a subject category. Ernest Robl has an excellent article on numbering systems for images that describes these two approaches. In my numbering system, slight variations, such as from a bracketed exposure, get an additional letter appended. Variations such as the same image shot wide open and stopped well down get different numbers.
Overall, I find the digital editing process much slower than editing slides and I'm looking for ways to speed this up. With slides, I can quickly view an image with a loupe and immediately tell if it is sharp, whereas with digital I need to open the file with a viewing program and adjust the magnification appropriately (this takes about 5 to 10 seconds on my computer). While this might not seem like much time for an individual image, it greatly increases the time I need to process images from a large project which may have as many as a thousand pictures.
I perform most global image adjustments in Nikon Capture (Figure 2), which I use for
exposure and color adjustment. Typically I start by selecting an appropriate white balance setting and then adjust the exposure and contrast. I save all changes in the original raw file.
There are several other programs available for manipulating RAW files. These include Adobe RAW, Bibble, PhaseOne, and Qimage. Thom Hogan has a detailed comparison of these programs with Nikon Capture After processing the images in Nikon Capture, I open the images with Photoshop and I edit in Adobe RGB color space. In Photoshop, I do additional processing which could include:
I generally edit in 16 bit mode and once I am done with adjustments, I save the image as an 8 bit TIFF file with the ICC profile.
I do not sharpen images at this stage and I reserve this operation for the last step before making a final version for publication in a specific medium. I tend to sharpen differently depending on the specific output device
I use the directory structure to store and manage different versions of my images. Each directory stores a single type of file: For example, I store my images in the following directories:
I further divide each directory into additional folders which each store up to 1000 images. In theory, modern operating systems should have no problems with an arbitrary number of files in a directory. However, in practice having too many files can cause problems, especially with applications that are poorly written.
I use IMatch (Figure 3) to keep track of my images and their associated information. IMatch is an image database that stores pointers to where the image files are stored on disk and associated information that the user defines such as captions, date, location, and so forth. In addition, IMatch also supports IPTC (International Press Telecommunications Council) and EXIF (Exchangeable Image File Format) fields. IPTC fields are often used to store caption and author information, while EXIF is typically used for storing camera settings such as exposure or lens information. With each image I store the following fields
IMatch has some nice features which can greatly reduce entry time and facilitate searching and browsing. Most importantly, IMatch supports batch labelling multiple images. I often have many different images of the same subject (shot from different angles) and batch labelling saves me a tremendous amount of time entering the caption information which would be identical. IMatch also supports scripting but so far I haven't needed this feature. To facilitate image retrieval IMatch has features such as search, color tagging, and hierarchical categories.
One drawback of IMatch is that it doesn't have a good way of tracking different versions of the same image and storing the associated information. So I only store the processed version. If I haven't yet processed an image, I simply use a small unadjusted version and then update it a later time. I mark each image with version information telling me whether it has been processed yet or on my website. I also use IMatch to store information on my slides and negatives. Again, as with digital, each image gets a unique identifier. If I haven't yet scanned a film image I use a default image which has the words "not available".
IMatch has excellent export facilities. In order to get images onto my website, I export the relevant fields from my IMatch database to a text file. I run the text file through a Perl script to extract the relevant information and I FTP the textfile to my database server (MySQL running Apache) and import the file.
On my website my images are 600 x 400 pixels and my thumbnails are 200 x 133 pixels. For large images, I process them as follows:
For thumbnails, I use the following steps:
To reduce processing time, I automate these steps using photoshop actions. The large versions require my input at several stages as I will change the sharpening settings and the placement of the copyright and URL text depending on the image. For thumbnails I always use the exact same settings.
I consider backups to be critical. I have owned several different computer systems with perhaps a dozen hardrives among them. Of that dozen, I have experienced two hard drives failures. I archive my RAW files by creating three CD copies with the "write verify" option. Of these three, one copy is stored offsite. Ideally, one should use at least two different kinds of CD media to prevent a common mode failure.
I also maintain a backup of all my files on an external firewire hard-drive. I simply copy all of my files on the computer using the supplied software (Dantz Retrospect).
I have Retrospect set to use a plain copy (as opposed to a compressed archive) so that my files will be accessible from any computer.
The software compares timestamps and only updates new files. I also use the write verify option.
In this article, I presented one example of a successful workflow. Since my goals and objectives could be different from yours, some of my suggestions may not be relevant for your situation and you should carefully consider what portions of the presented workflow will be effective for you.
|
