Toward a more perfect photo

No more politics, I promise. I’m going to bore you instead with digital geekspeak.

Another problem we didn’t used to have is how to manage the output from multiple cameras. On this last assignment I usually shot with two 5D bodies. It minimized the lens changes which, when I use one body on a job, I’m changing several times an hour. My sensors stayed a lot cleaner this round. But file management got a lot more complicated.

The first issue is the file numbering problem. If I could (and maybe there is a way), I would change the prefix that the camera assigns to the files. I reset the numbering on one camera so that it was nowhere near the count on the other camera. I synchronized the time setting on both cameras so that the images would sort by chronology. But I was using one camera more than the other (and I was making so many thousands of captures) that one day I had duplicate file numbers. Fortunately, my ingest program (Photo Mechanic), merely assigned a numerical suffix to the duplicates. The sequencing went all to hell though.

Now that I am in the file processing stage, I’m getting excruciatingly sensitive to the different capture quality of the different bodies. As it is, I have different calibration settings for each body and for each kind of lighting (daylight, cloudy, fluorescent) that I made using Eric Chan’s protocol. The biggest difference between the cameras is the different amount of green saturation correction required. That gets me one step closer to a match, but it requires an initial sort (using a metadata search for the serial number of the camera) and applying the calibration setting to the mass. The images subsequently get  jumbled back together, and it’s difficult to segregate the images again when making individual corrections. I’m batch correcting dozens and hundreds of images at a time, and at some point I have to say, it’s good enough. As Voltaire put it, "The perfect is the enemy of the good."

Comments

[Josh Wand]

On the sequencing problem-- why not put a timecode as part of the filename, or, alternatively, sort by capture time and then renumber the whole group?

I don't use photomechanic, but I'm sure you must be able to do one of these.

How are you downloading from the cards? Are you actually using photomechanic for that part? You might want to use a product like Breeze Downloader (for windows) or ImageIngester (for OS X) that can dynamically assign new file names AS you are downloading from the cards. No more filename collisions!

[David Adam Edelstein]

A second for Breeze downloader pro, which makes sure that all of my files are pre-named when they hit my computer. The templating is quite powerful and you can basically make any of the metadata part of the filename -- I wonder if that includes the camera s/n or something.

Mostly though I wanted to say, holy smokes, I had no idea the difference between different cameras of the same model was so much. That's wild.

[Graham R.]

I used to use downloader but i have recently switched over to Image Ingester: http://www.basepath.com/ImageIngester/

the main reason was because its cross platform and the "pro" version supports multi-camera imports.
I'm on windows right now but the Pro feature set is something that has me anxious for it to be ported.
Also, you can apply any of your Camera raw templates at the time of import - making that initial meta-data entry go faster or even applying your batch camera-1 corrections when you download.
Like breeze it will also rename, sort, and make a backup copy of the files for you.

best thing about it is that the basic version is free... and well... thats a hard price to beat.

[QT Luong]

If you learn any scripting language (such as the unix shell included in OS X), it's really not difficult to write a dozen lines of code to rename downloaded images to what you actually want.

[QT Luong]

If you learn any scripting language (such as the unix shell included in OS X), it's really not difficult to write a dozen lines of code to rename downloaded images to what you actually want (as opposed to what some programmer thought would be useful).