Shooting Raw

Volume 10, Issue 11; 12 Feb 2007; last modified 08 Oct 2010

New cameras bring new technologies. I'm making every bit count with raw images.

The firmest line that can be drawn upon the smoothest paper is still jagged edges if seen through a microscope. This does not matter until important deductions are made on the supposition that there are no jagged edges.

Samuel Butler

Back in December, I got a new camera: I've replaced my Nikon Coolpix 5700 with a D80. I learned photography with an SLR, a Minolta SRT 101 that I got from my Uncle Virgil circa 1983I eventually switched to an X-700 (because it had the same lens mount), but I think the days of film are behind me now. I feel a litle sad about that. I fondly recall buying film in 100 foot rolls, loading my own canisters, shooting it, developing it, making contact prints, choosing images, and printing them in sizes up to 11x14. I never did color processing myself, but I did a lot of black and white. Getting from unexposed film to negatives doesn't really require much in the way of equipment. If you like photography, I highly recommend doing it at least once. Making prints requires a bit more equipment, but it's even more magical.. It feels so good to have an SLR in my hands again.

Although it was possible to store raw images with my 5700, it wasn't really practical. Saving a Jpeg took two or three seconds, saving a raw image took closer to 10. In the field, two or three seconds can feel like a lifetime. Ten seconds feels something akin to the geological time scale. My D80 can store raw images as fast as I can push the shutter (faster, in some modes).

One of the advantages of shooting raw is that you have much greater control over the image: the raw sensor data has more latitude in terms of exposure, white balance, saturation, and other qualities than the “frozen” jpeg.

This is roughly analagous to the difference between exposed but undeveloped film and a ready-to-print negative. The decisions you make about the kind of developer, temperature, length of time in the developer, etc. all have an impact on the negative that you get. You have some control over the final print when you develop it, but if you've lost detail in the negative, there's no print process on earth that will bring it back.

When your camera stores a jpeg image, it's made a whole bunch of decisions for you and you're stuck with them. When it stores a raw image, you get to make all those decisions yourself.

One of the disadvanages of shooting raw is that you have to make all these decisions about every image.

There are all sorts of commercial applications for managing digital work flow on platforms that I don't use. On the platforms I do use, I'm still exploring the work flow options. Here's what I'm doing today:

  1. I move images off the camera's media with a script that puts them on the local file system:

    1. It sorts them by date and creates date/nef for the raw files, date/jpg for the original jpegs, if there are any (so the script works with my older cameras too), and date/rdf for the RDF metadata (I can't put it in the raw images, so I'm storing it separately now).

    2. Next, it creates scaled jpegs with UFRaw and stores them in date . This conversion uses all of the camera's settings to create an image that's more-or-less what the camera would have created if I'd shot in jpeg.

      I scale the jpegs so that they're no wider than 1600px and no taller than 1200px. That makes them fit on one screen even when I'm running with a dual screen configuration.

      (You may have to build ufraw yourself, as I did, because the latest version packaged with my Ubuntu distribution doesn't include support for the D80.)

  2. I used to save most of the images I took, even the ones that were pretty much crap. But having a DSLR makes it possible to take a lot more photographs. That means a lot more crap and raw images are big. It doesn't make sense to save them all.

    The first thing I do is run through all of the scaled images with Xv, deleting the ones that don't make the cut. Then I archive the junk with another script. That script ought to delete the junk but I'm not quite brave enough to do that yet. I usually delete the archive after a couple of days.

  3. I process the images I want to publish with GIMP, using the UFRaw plugin interactively to choose the characteristics of the final image. I store this image in date/edited in the lossless XCF format then make a jpeg for publication.

  4. I wrangle the RDF metadata into the jpeg image so that my other publication tools will do the right thing.

Live and learn, we'll see how this goes. What's your work flow like?


I have a very similar workflow to yours (with a slightly older Nikon D70) -- although I can't always be bothered going through the GIMP plugin, so I sometimes just upload the auto-converted versions.

I recently worked out a work-around to some problems I'd had with copying EXIF information over to my jpeg previews. A post I did on this may be of some interest or use to you. To see the problems that I am trying to solve have a look at the output from "exiv2 pr photo.jpg" after a ufraw conversion, versus the result after adding in the extra exiv2 step.

—Posted by Malcolm Tredinnick on 12 Feb 2007 @ 11:23 UTC #

I suspect I have a somewhat different workflow. I try to minimize the need to delete at all stages: being 99% a landscapie at heart, I believe in taking control at time of shooting, understanding and choosing parameters to taste. A 256M CF card, allowing 17 shots in an afternoon, is quite enough when they almost all come out decent - and although I've gone up to a 1G card, I'll still not be running rampant with bulk over quality. (This is particularly relevant since I also shoot MF and LF film, processing b&w myself.)

When I'm shooting digitally, images may still get deleted through chimping, but that's infrequent. The rest of the workflow is simple: NEFs get offloaded to a new directory, bulk-converted to DNG, which are then edited (mostly tweaks to levels, curves, HSV, crop, optionally funny effects) in Photoshop to PSD. If I find a naff shot has made it through, I'll skip it at this stage (probably happens about once a month). Then a couple of PS actions to produce an image under 1600x1200 with borders in 8-bit sRGB, then ImageMagick to bulk-convert down to within 800x600, adding copyright text. These I upload to my general dumping-ground and leave to stew for a while (at least a week for the rose-tint to fade), before taking a selection of the best for use in the gallery. Film is scanned to TIFF and processed as PSD etc. I archive NEFs and full-size JPEGs only - no TIFFs, no PSDs, no DNGs. This is enough to allow me to upload the jpeg to photobox later if I want, or to re-process it keeping an eye on how I did it previously, and keeps wastage down too.

—Posted by Tim on 13 Feb 2007 @ 11:58 UTC #

Since I mainly photograph dogs doing what they do, most of my photography is in high-quality jpeg. However, I've started using raw for most other situations lately to benefit from the added control.

My workflow is: move images into folders based on shooting date. Then go through the pictures and delete the crap (however, many memorable crap shots may still remain ;).

After this I'll open the shots worth publishing with the Gimp + UFraw, Photoshop, or Lightroom and do what needs to be done. Sometimes I process more, sometimes less. The processed files are saved as jpg, tiff, png, or psd in a subfolder. The reason why I don't use xcf to save the processed images is Photoshop, which my wife uses.

—Posted by ramin on 15 Feb 2007 @ 04:01 UTC #