This came up in a web forum where I was asked to say what I do about keywording….. Without being too fancy about it, my goal is:
“the maximum number of relevant keywords in the time I think it’s worth spending”
With that overall objective – intentionally a rather elastic phrase – in mind, I’ll do keywording in 3 passes.
1. List the contents
Firstly I pretend I’m describing the picture to a blind man or stranger who knows nothing about anything. So for an image like this from Arches, I don’t start off by adding keywords like “Landscape Arch” which mean nothing to my hypothetical stranger. I’ll add words like “arch”, “desert”, “sand”, “red rock”.
2. What does the picture mean?
Second, I try to add abstract or adjectival terms – eg isolation, remoteness, geology, erosion, sculptural, solidity, resistance.
3. What do I know about the picture?
Only then will I add the easy words, information that I actually know about the image’s contents – Delicate Arch, Utah etc.
In some ways, this 3 phase approach is back to front. That’s deliberate because it forces me to describe the picture’s actual contents, add as many relevant keywords as I can, and not be lazy and just add the obvious words. For example, if I started keywording this image with Swaledale, Yorkshire, England, some of my keywording time and enthusiasm is already expended by these easy terms. So will I have enough time and keep going long enough to add purely descriptive words like walls, barns, grass? And will I really get round to adding lateral or abstract words like “green” or “patterns”, which are also what the image in this post is “about” and which I or someone else may subsequently use as a criterion for choosing pictures? Adding the easy, “known information” keywords last is my way to make sure I reach my goal of the “maximum number of relevant keywords in the time”.
Other points
I tend to have 6-15 keywords per image, make a lot of use of metadata presets (so I can update location and other fields in one go), and use keyword sets for images where there might be a bit of variation from picture to picture – eg a set might include one item per major arch.
My keyword list is completely flat – no hierarchies – and I think not worrying about hierarchical structures and organisation also saves me a bit of time.
While I see keywording as something you do while images are still new, I will often add more keywords later.
What about names?
As for real names, I include them as keywords only if they are well known people, but not for members of the family or people who aren’t known outside one’s circle. Place names get added as keywords if the image is typically “Utah” or “Swaledale”.
There are other ways to keyword, but that’s how I actually do it. For what it’s (beards)worth.
Also see
See these posts on keywording.