Flickr stats revisited

A while ago, I posted to the blog my thoughts about how I could use Flickr to help me choose/confirm choices for up-coming competitions. Of course, this means I need to get those images on Flickr a good deal before the competitions. Failing that, I can always look back with hindsight, and see if the placings (or not) in competitions match up to the "interest" in an image, as calculated by Flickr.

Below is a screenshot of the most recent stats for my photostream on Flickr. The "star" means favourite by another user, and the bubble means "comment".



Interesting that Window and Chair and Sunset Beach Walkers are up there (in 1st and 3rd place); these did v well for me last year. Bath Crescent is second, but did nothing last year (although did get selected for the Annual Exhibition this year), so not full agreement with the Flickr stats. White Peacock bombed in a recent projected image competition, although is 4th on Flickr. Reculver got a Highly Commended, and it's 5th on Flickr, so not bad there. Shadowman, 6th on Flickr, won me Creative Image of the Year in 2007. Peek-a-boo (Flickr 21) got me a 3rd place, but it's not been on Flickr as long.

So, mixed bag. I think I need to standardise things. I need to submit all images to the same Pools (for exposure to wider Flickr) and then track the interest for a set period of time. I assume views = interest, comments = more interest and favs = more still, which is kind of how Flickr works it out. Also, I think I need to standardise the tags on images. This is almost impossible, since different images need different tags, unless I pick very generic ones (like camera model) or don't use tags at all for all images.

Interesting from yesterday's Cambridge night posts, the order is Kings 1, River Cam 2, Clare Bridge 3, which is what Dams thought too.

Still, useful to see some rough agreement between results and Flickr, but it's not statistically controlled, so impossible to draw any conclusions as currently running.

Thoughts? Suggestions?

Ivan

Vertorama tutorial

This is from a contact on Flickr, called Panorama Paul - check out his site, it's excellent.

Anyway, this link describes his way of stitching two images vertically to get a "vertorama" (his word, not mine), and simulate a kind of high resolution, medium format type approach. Some of his images done this way are nothing short of spectacular.

Check out the tutorial, and check out his photostream for inspiration. Lots of good descriptions on there and excellent shots.

Ivan

A good use for Flickr

I've just up-graded my free Flickr account to a "pro" account for about £13.

One of my reservations about Flickr is the fact that the comments are 99.99% favourable, no matter the nature of the picture. Thus, it's a bit of a back-slapping exercise, rather than a true forum for critical discussions. Unlike this venerable blog.

However, there is an upside. Well, several, I think.

Firstly, there is a lot of great stuff on there. But you have to look. If you do find great stuff, make it a favourite, and learn from seeing that work, that approach, that style. I like this aspect.

Secondly, and the main meat of this post, with the pro account comes the rather natty "Statistics" package. This breaks down all the traffic on your photostream, tells you which images have the most views, the most comments, the most favourite tags, and can do all kind of funky stuff with the data. So, while the comments on individual images are not all that useful, if you look at the total picture, it can reveal things about the images that the comments do not.

In addition, I have entered some of my photostream images to club competitions, so I know how well they have done. I can compare this to the results from my stats, and see how well it's borne out. Like in any clinical trial, individual results don't give the complete picture, but the larger the sample group is, the greater the statistical power, and the more relevant any trends become.

To test this, I have had a look at some images that I know I have submitted to club competitions and that I also have had on my photostream for some time now. Here are how they break down for a selected few, from a photostream of 44 images:

Window and Chair - 1st in a print competition; most popular in Flickr
Sunset Beach Walkers - digital pictorial image of the year; 3rd most popular in Flickr
Shadowman - creative image of the year 07/08; 6th most popular on Flickr
New York Ferry Moment - bombed in CCC; 16th on Flickr
Help - bombed in CCC; 18th on Flickr
Lady - bombed in CCC; 19th on Flickr

The others I have not submitted before, so can't compare, but I have submitted two images to the competition next week that are on Flickr, and are currently residing at 4th and 13th, which would imply that one will get something, and the other will bomb, which was my EXACT pattern last year in competitions with two entries! So, we'll see...

To complete the picture, I'm going to upload all my entries from last year, and see how they do. Clearly this is not scientifically controlled, but I'll try not to bias the descriptions or tags to favour one over the other. Also, I have submitted some images to other groups, and some not, so there's clearly a difference in the exposure of the images to the wider Flickr community. I'll try to be even with this too, but if an image gets and invite and another doesn't, that too is a measure of it's popularity, so I can't ensure exposure is even totally and neither should I.

Ivan

Flickr

Do any of you chaps use Flickr? If so, what do you think? I posted a few images on there a while back, and go very little feedback. Then, I put a couple of monos up there, and there was a flurry of comments, and "awards" (basically invites to post on a certain group and encourage you to give awards) and invitations.

Have a look at what I mean at my photostream.