UPDIG Guidelines, Version 3.0

Digital Image Workflow

No single workflow suits all photographers or all clients. A good digital workflow is the most efficient and automated way to get the job done while capturing the most image information in the widest gamut for the widest possible uses in the future. It should satisfy the client’s needs, embed necessary information (metadata), embed color profiles, save derivative files, and archive and back up files. A good workflow saves time and protects against both loss of images and loss of work done to the images. Choosing the best software to achieve these aims is key. Because digital photographic imaging is relatively new, photographers must regularly explain to clients the trade-offs between quality and cost in different workflows.

A film-based workflow is simple. Photographers deliver film, designers or art directors decide how pictures will be used, and offset printers and prepress houses handle conversion of the film to printing plates.

Digital cameras, along with scans from film by photographers and agencies, are now replacing historical workflows. While clients have been quick to embrace the speed and convenience of digital capture and delivery, they do not universally understand what is required to achieve the same quality levels that they used to expect from film. With the exception of those involved in high-volume, quick-turnaround workflows, most photographers must decide how to handle file preparation. Some photographers want to avoid the distraction of file preparation. Others have embraced it, because it allows powerful control over the reproduction of their images. Profiles and soft-proofing allow photographers to see how their files will look as display or press prints. Looking at soft proofs on their calibrated monitors, those who embrace file preparation can deliver to printers files that will reliably (if not precisely) reproduce on paper the optimized image files the photographers see on their monitors.

With some photographers embracing and others avoiding the details of file preparation, clients are understandably confused about who does this work and how much it costs. Many have tried preparing image files without first calibrating their monitors, yielding disastrous results. Some assume printers are accustomed to working with RGB files. However, a survey released in February 2005 by Printing Color Digest found that, of some 800 offset printers queried, only four accepted RGB files in their workflow. When you send an RGB file to an offset printer, the biggest risk is that a pre-press worker will open it in the wrong RGB color space, altering the color, then lock in the mistake by immediately converting the file to CMYK. If the printer receives an untagged RGB file, and there has been neither communication with the printer nor the inclusion of a ReadMe file that indicates the color space of the files, prepress will probably open the file in the shop’s default RGB space, which may or may not be the space in which the file was optimized. Even when an image file is correctly tagged and its profile preserved when it’s opened, there may be problems if a printer uses a RIP for CMYK conversion. Most RIP software does not use black-point compensation, and without it, some conversions can appear flat and unsaturated (muddy).

The workflows below are general in nature. They aim to illuminate options and possibilities rather than offer step-by-step guides. The ideal workflow is a moving target. It varies based on the nature of the project, the photographer’s capabilities, the client’s needs, and perhaps most importantly, the overall budget and time constraints. Photographers should adopt and adapt workflows to fit project parameters.

High Volume, Quick Turnaround Workflow

Examples: Most photojournalism, sports for newspapers and magazines, events, and most public relations.

The focus is on speed. There is a calculated trade-off between ultimate quality and speed of delivery. Consequently, such workflows were the first to replace film with quick digital capture. Some issues and solutions include:

  • High quality JPEGs, captured in either Adobe RGB for print or sRGB for web, are suitable in many situations.

  • Professional digital SLR cameras offer white-balance controls (with surprisingly accurate auto settings), in-camera sharpening and custom tone curves that can be applied at capture to generate JPEGs that are often adequate — if not optimal — for reproduction.

  • Sometimes JPEGs must be reduced or compressed in the camera or afterward to meet transmission limits of bandwidth and time.

  • It is expected that news organizations and printers have staff trained to receive transmitted photos and prepare them for either print, broadcast or web.

  • Software such as PhotoMechanic allows embedding IPTC data on-the-fly as image files transfer from memory card to computer, saving time and ensuring that the file recipients have all necessary captioning, cataloging and crediting information.

High Volume, Moderately Quick Turnaround Workflow

Examples: Monthly magazines, institutional brochures, annual reports, web sites, theater and dance performances, and some types of advertising.

There is more focus on balancing competing needs for speed, quality and cost. This leads to many questions about who should do what. Some approaches and concerns include:

  • Whether to capture RAW files or high-quality JPEGs. A single assignment can generate thousands of images that could take many hours to process from RAW files. If the photography is straightforward, the lighting is good (or at least not mixed), and the intended use is web, broadcast, newspapers or magazines printed on web presses, then JPEG capture may be fine. If the intended use is higher-quality, sheet-fed printing, if the lighting is mixed or requires a high ISO setting, if the final output will be full page or larger, or if there is a strong chance for higher-quality uses later, then capturing RAW files provides the best quality.

  • Presenting images to the client, art director or editor. Although there are many good image file browsers, including Photoshop’s companion browser Adobe Bridge, speed and convenience often require posting a web gallery for editing. If not captured in sRGB, which is the default color space for most web browsers, copies of image files should be converted to that narrow-gamut space before posting to a web site. Usually, it’s also best to reduce the resolution and compress the files to fit the viewer’s screen and load quickly. Typically, proof images should be 500 to 1,200 pixels at their largest dimension (width or height), unless an editor or designer needs to see minute details. If you are using a DNG workflow, you can include an sRGB JPEG preview in the DNG file which can be used for creating HTML proofs, so you don’t have to run a separate batch just for proofing JPEGs.

  • Delivering files for reproduction. Clients have two basic choices: 1) order optimized image files from the photographer; or 2) receive camera files and take responsibility for optimizing their reproduction. Either way, if good reproduction is the goal, someone who understands color correction, contrast enhancement and sharpening techniques will need to optimize the files. Some formats for file delivery include:

    1. Photographer delivers uncorrected, high-resolution JPEGs or TIFFs. At the very least, the photographer should embed and note the color space of the files, unless instructed otherwise by the printer. If the delivery method (bandwidth, size of media) can accommodate large files and the client has the needed software, hardware and expertise to deal with them, then 16-bit (48-bit RGB) TIFF files are probably best.

    2. Photographer delivers proprietary RAW files. Although not recommended, this is now an option, since Photoshop’s Adobe Camera RAW supports most camera RAW formats. However, the photographer should include .XMP files (the small “sidecar” text files that describe how the RAW file will be processed). Without them, the client won’t likely know the photographer’s intentions for color and tone. Working with RAW files also requires high-end computer equipment, recent software, a color-managed workflow and a good understanding of digital photographic processing.

    3. Photographer delivers DNG files. If the photographer made corrections for color, exposure, etc., and then created DNG files with the full size JPEG preview option enabled, the recipient of the file would have the flexibility of a RAW file and the intentions of the photographer. Until more RAW processing software supports DNG format, this would limit conversion options primarily to Adobe Camera RAW and a few others, such as DXO Optics Pro and Silkypix Developer Studio. Other RAW developers, such as Phase One, have promised support, so the list should expand soon. (Most of these products now can only read a DNG file and convert it to TIFF or JPEG. Full support, with capability to read and write a DNG, may take longer to evolve.) The same requirements as for proprietary RAW files — high-end computers, current software, color-managed workflow — would apply.

    4. Photographer delivers optimized, high-resolution TIFFs or JPEGs in an agreed RGB color space tagged with the appropriate color profile. Budget requirements — along with the variance of exposure, contrast and color between images — dictate whether and how much the photographer optimizes each file, or whether he or she batch processes them. Tradeoffs between cost and quality, along with the client’s responsibility for adjusting final output, must be clearly understood by all. Some issues to consider include:

      1. Whether to deliver 16-bit TIFFs to allow the widest possible adjustment by others further along the workflow. Client must have storage space, knowledge and processing power for such files.
      2. Whether to sharpen the image before delivery. Sharpening is only appropriate for images delivered at final reproduction size and when the photographer fully understands the sharpening needs for the final use. If an image file is sharpened, this should be noted in an accompanying ReadMe file, on the delivery medium (e.g., the CD-R label), on the delivery memo, or on the border of the image. Sharpening a duplicated, composite layer within an image file (TIFF or PSD only) allows a client better resizing options.
      3. Whether, how much and when to compress images as JPEGs. JPEG files are handy when many images must be delivered or delivery must take place with limited bandwidth or storage. But each time a file is saved as a JPEG, it loses image data and gains digital artifacts. The more a file is compressed, the more damage is done. When editing a JPEG file (whether created by the camera, scanner, or in post-processing), it’s best to save the changed image as a TIFF or PSD (native Photoshop) file. Ideally, a photographer should only deliver JPEG files that are compressed slightly (10, 11 or 12 quality), optimized and sized precisely for output.
    5. Photographer delivers RGB master file (show footnote). If the final use is offset printing, consider creating and archiving RGB master files. An RGB master file is used to create targeted CMYK files, which are sized, sharpened and profiled for the intended offset printer or press. The photographer can help ensure things go smoothly by delivering a “cross-rendered” guide print, as described in the earlier section on guide prints and proofs.

Low-Volume, High-Quality Workflow:

Examples: High-end advertising, photo-illustration, portraits, high-end editorial, architecture.

High-resolution cameras capturing RAW files, or high-quality film scans (usually from medium- or large-format film), are standard. Retouching, blending images and other special Photoshop procedures are common. Some photographers perform this work themselves, while others hire digital assistants or retouchers. Time and effort invested in postproduction often equals or exceeds that spent on image capture. Generally, post-production work yields an RGB master file (show footnote). How such files are prepared for final uses depends on the clients’ production needs. (See the discussion of sharpening for more info.) In some cases, it’s best to deliver several different file versions derived from an RGB master file, such as a high-resolution Adobe RGB file, a screen-resolution sRGB file and an CMYK file optimized for the expected press conditions, possibly sized and sharpened.

High-Volume, High-Quality Workflow:

Examples: Product and catalog photography.

Methods are often similar to low-volume, high-quality photography. However, to speed production, the large number of image files often requires help from a digital service bureau or pre-press house. Some studio workflows — particularly for catalogs — where all conditions can be carefully controlled, can benefit from a JPEG workflow.

Workflow Tools and the UPDIG Guidelines

Digital capture proved to be revolutionary in the sense that it provided options beyond film. Now that digital capture has become such a large factor in the creation of photographs, a second revolution is occurring, and that is the proliferation of software tools photographers can use to produce, edit, process and manage their digital image files. We are seeing a shift from Photoshop-style pixel editing to metadata-based image editing. This shift is occurring largely because of the nature of RAW files themselves. RAW files are not image files in the same way that JPEG or TIFF files are. RAW files are raw camera sensor data and do not become image files until they are processed according to a set of instructions created by RAW processing software in conjunction with user input.

Preserving appearance

“Digital images should look the same as they transfer between devices, platforms and vendors.”

Unlike JPEG or TIFF files, which (if properly color-managed) always look the same as they transfer between devices, platforms and vendors, RAW files will look different depending on the RAW processor and the user’s inputs.

The challenge, as UPDIG defines it, is to seek the emergence of a RAW file processing ecology that will allow the photographer or image creator to fix a RAW file rendering over applications and over time. As we discussed in the File Formats section, there are currently two ways to transfer a RAW file’s rendering intent: the use of some type of sidecar file (e.g., XMP, BIB), or the inclusion of a JPEG preview (the DNG approach). The trouble with sidecars is that they are separate files and might get lost, deleted, overwritten, or detached from their original RAW files. Additionally, most cataloging software — those who do not consider Apple’s Aperture or Adobe Lightroom to be effective cataloging solutions yet would say “all” — cannot read the sidecar files, with the result that the rendering intent is not reflected in those programs. Even if other applications could be made to read a variety of sidecar files, rendering intent would still very much depend on the rendering engine being used. Both the Mac OS and the new Windows Vista OS will have RAW file rendering built in, but the rendering results probably will be different. The implication is that a RAW or RAW + sidecar would render differently depending on the OS, browser, or cataloging software. This type of subjective rendering is precisely what UPDIG was created to combat in the digital file realm. If we adhere to UPDIG principles, the DNG format is the only RAW format that currently fixes rendering across applications and platforms, and hopefully over time, although that is by no means assured at this point. All the other RAW file formats, so far, operate in closed proprietary systems, so that fixed rendering requires conversion to another file format for proofing, cataloging, or delivery.

Preserving metadata

“Digital images should have metadata embedded that conforms to the IPTC standards, making the images searchable, providing usage and contact information, and stating their creators or copyright owners.”

Competing with fixed rendering and the safe insertion of rendering metadata in RAW files is the need to insert the other kinds of metadata outlined above. It has been a consistent problem that there are no standards among the proprietary RAW file formats as to where this metadata should be put and how other applications can discover and use this information. Some proprietary RAW processing software actually removes such metadata, creating a workflow problem and requiring this information to be re-inserted in derivative files.

Adobe’s DNG file format is the only RAW file format that safely contains this kind of metadata without using a sidecar file. Those who want to use RAW file processors that don't at least read DNG will have to continue to archive and manage proprietary RAW files. They will have to keep their metadata in sidecar files or database libraries. They will have to be mindful of the need to arrange their workflow so that metadata is either preserved in the conversion to derivative files or, if necessary, added afterward to the derivative files. In addition, photographers whose work depends on processing thousands of files quickly may well settle on a different set of workflow software tools than those who seldom shoot or process more than 30 or 40 image files per project. You should evaluate your software tools based on your needs (and the capabilities of your computer equipment) so that you can make the correct compromises among perceived quality of rendering, speed of processing and ease of adding metadata.

There is quite a bit of confusion concerning the suitability of the emerging RAW file handling software such as Apple’s Aperture, Adobe’s Lightroom and Nikon’s Capture NX. Some people report that these all-in-one programs unify and improve their workflow, while others complain that they are slow, unstable, or unable to handle batch processing. Much of this conflicting information has to do with whether the program is being used on tens of files or thousands of files, and whether they are installed on the fastest dual-core processor computers with 4+ GB of memory or on older hardware. Its no wonder that a favorite phrase on digital photography forums is “your mileage may vary.”

 

login
Last update: September 14, 2007