fine whine

The more I play with making fake CD covers from flickr photos, the more I get frustrated by the sheer number of interesting photos that are unusable because of their creators' chosen license. I realize that if I emailed them, many wouldn't mind me using their photos for something like this, but that seemed like it would be so much trouble.

So instead I programmed a little script to find only the usable shots, and made a page for it. Took me a couple hours, give or take, to get it working.

To be more specific, here's what you might want to know about what I did.

Flickr photos can be licensed in a number of ways, either through the default "© All rights reserved" that regular copyright protection grants, or through the more flexible and friendly creative commons licensing with its many permutations, including requirements for attribution, non-commercial and commercial use, and even the licensing of derivative works. So what I needed to do was take the five hundred photos flickr picks each day for their "interestingness", and filter out anything that was marked "all rights reserved" or "no derivative works". Fortunately those happened to map to the integers 0,3, and 6, which meant filtering them out was a simple modulus test*. Calculations aside, the rest was a matter of a few hours' work and the phpFlickr documentation to get a working script that wouldn't hammer my server, nor get me banned from flickr's.

So all it does is check 3,500 of the last seven days of interesting photos, and from those, screens out the usable ones. From my quick observations, the percentage of correctly-licensed photos chosen for their interestingness is under ten percent. Which means that hopefully I can save people some time. The one that it picks, at random, is ready to use.

Click here to try it for yourself.

Go ahead and put your comments and questions below. I'm releasing this as a 0.1 version - I know there is much more I'd like to do with it, but since it works, I figured I'd get it out there for people to play with. My source code is messy, but eventually I'll get the relevant bits of it posted too.

* At the risk of sounding too nerdy, modulus is, and has been, one of my favorite mathematical operations. It's just a fancy name for "the remainder", but since that sounds like something out of elementary school, I think everybody calls it the much more impressive "modulus". It's wildly useful (or at least, I've used it a lot), and, well, I've used it a lot, probably more than multiplication and division combined, in my programs over the years.

I've uploaded more than ten of the random CD covers I've created (mentioned earlier) to a new flickr set* I made.

See it here.

As I've started spending more time and effort on making them, I've found myself going through old, uninstalled fonts and using parts of Photoshop (and the Gimp, and Illustrator, too) that I'd not used much of before. I've even re-done a few from before that looked, well, cheesy. I imagine this set will only grow as I mess with more of these.

What can I say? It's fun.

* I could finally create a third photo set having activated the "pro" subscription that Janice gave me for Christmas. Thanks Janice!

So ever since I read about them on Neatorama, I wanted to make some random CD covers. From what I read, the recipe was simple:

1. en.wikipedia.org/wiki/Special:Random The first article title on the page is the name of the band.
2. www.quotationspage.com/random.php3 The last four words of the very last quote is the album title.
3. www.flickr.com/explore/interesting/7days/ The third picture, license permitting, is the album cover.
4. The finished product belongs in the CD cover meme pool.

I turned out a few in as many hours. I tried to stick to the rules, but couldn't bring myself to use the photos that were marked "© All rights reserved" when I knew there were ones licensed (via creativecommons) for derivative works, as this would likely be considered. Though I ended up reloading a few times more than I liked, I did come across enough to make these (and a few more that I'll eventually upload).

Making fake album covers is nothing new to me. Back when I was first learning Photoshop I'd made many a cover using stock photography for a fictional band called "Spontaneous Grape", going even as far as creating a fictional record label* to release them. But coming up with the titles was often the trouble, and moreover selecting photos that I thought would be interesting even more so. Without those aspects to worry about, I can crank these things out much faster.

Eventually I'll get them up on flickr, annotated and everything else. But I've got to take a break from making them first.

* The name of the label was Ludd Records, and it was rather a bit of a dumb in-joke. One of my many online identities was that of "Luddite Industries", which I thought to be a particularly sophisticated joke, in that the Luddites would not likely be operating a web site. Here's the logo, which I drew in AutoCad, knowing it better than Photoshop at the time.Someday I'm going to make a black t-shirt with this on it in white.

I'm not finished with my bracket programming by a long stretch. There's no way to save the page, it's still gigantic, and frankly I'd much rather play with it than fix it.

I find interesting connections between the contenders. Sometimes the two share stars (I've battled over both Tim Robbins and Dustin Hoffman*, among others) or themes, or sometimes just the year they were released, but no matter which two are randomly picked, it's still fun to pick one over the other.

With that, I can add this to my ever-expanding list of programming projects I've begun and abandoned. The script that grabs the first titles, even, is among them. I've intended for some time to have all sorts of interesting lists and polls and links created from the movies I've seen and want to see, but so far have only been able to grab the raw data and effectively sit on it.

But I don't intend to give up on this yet. Maybe I'll dub it the 'bracketeer' and devote a directory of my site for it, for anyone to come and pick from the movies I've seen, or even enter a new list to rank--basketball teams, crayon colors, anything. It's just a matter of further cleaning up the code, and making it easier to use.

Hmmm, the bracketeer. I like the ring of that, I think.

* Wag the dog beat Stray dogs.

So an idea struck me for an interesting programming project and a fun diversion: a single-elimination bracket for ranking the movies I've watched, 64 random ones at a time.

You see, it's March, and with that month comes the annual basketball tournament, and while I enjoy making the brackets I don't know about or care about the teams involved. So movies it is.

The logistics of entering and displaying the data made for an interesting challenge, and it's taken me a week to clean up my code. It worked fine for me the second day, but the code was horrible and disorganized and very, very bad.

I've rewritten it twice already, and with the magic of regular expressions* I've condensed some eighty or ninety cut-and-pasted lines down to around ten, and now I can make brackets of more arbitrary sizes (well, powers of two, unless I introduce some sort of 'bye' functionality) filled with whatever I want to tourney-ize.

So enough talk. Take a look at this example of the final output.

Yes, it's huge. Yes, it's difficult to read. I'm still working on making it legible at smaller sizes, but that's another project for another day.

Let me walk you trhough some of the more interesting match ups of that particular bracket. You may notice that The Shawshank redemption beat The Godfather, and here you can see my opinions in action. I know Godfather is probably the better movie, but I also know that any time I'd see any bit of Shawshank, from any point in the movie, on TV I'd watch the rest of it all the way through. I've ranked the winners by how much I enjoyed watching them, not necessarily how good they really are.

Moving back to round 1 (the one with 64 titles), we have some interesting bouts:

• Local hero vs. Terminator 2: This isn't the first choice that would probably get me drummed out of the armchair film school. While the former is a fantastic movie, T2 is even more fun to watch, if not one-tenth as intelligent.
• The living daylights vs. Mr. Deeds: This was not a choice to be made lightly: I didn't really enjoy either film all that much, and neither will be held up on a pedestal anytime soon. In the end Deeds got the nod because the copy I watched had Malaysian subtitles that were as informative as they were entertaining.
• The conversation vs. The last action hero: Actually this isn't that interesting. It's a blowout.
• Soylent green vs. Cube: I thought about this one longer than many others. I enjoyed both films, but in the end I liked Cube that little bit more. It's not nearly as dated as Soylent green, and nowhere near as parodied.

Later notable matchups include these:

• The Blues brothers vs. The conversation: Both of these films are the sort that should appear in the final four; I was sad to eliminate either of them so early on. Other times I've run brackets each one has 'won' at least once, but the advantage goes to the SNL movie since it's so much fun in every way that Coppola's film is serious. They both so darn great, though.
• M*A*S*H vs. The tall blonde man with one red shoe: I doubt these shared any marquees in 1972, unless Altman's film got stuck in the art houses. The latter is a small French picture made into a rollicking remake with pre-Forrest Tom Hanks, and it's a good movie too, but like most foreign movies remade here, it loses just a little in the translation. M*A*S*H, on the other hand, is too loose, too unstructured, too disorganized to grab onto without watching it five or fifteen times. I may like it more someday, but I've only watched it once so far and I'm in no hurry to see it again.

I've found this method for picking the movies I liked far better than picking favorites. I could never pick the one (or five, even) movie I enjoyed the most to yoke myself to it as a favorite. Grabbing sixty-four at once means I can subjectively pick the ones I liked most, on something of an equal playing field.

Of course, something like pitting Scary movie 3 against The Empire strikes back isn't exactly a meeting of equals, but I'm pretty sure that's how these things work out with the athletes, too.

* Regular expressions are cryptic strings of letters, numbers and other characters that make matching patterns inside text much, much easier. I wish I'd learned them in grade school, or at least high school. Read more about them here; at least, that's what I do every time I need to use them.

working walking

This entry was posted from my treadmill*.

It was only a matter of time before I'd do something like this. I've played PS2 games on the treadmill, I've read books and watched movies, and now, I've been online... at three miles per hour. The board isn't quite the right size, and I couldn't find matching clamps, but for a first attempt it was entirely adequate.

* Of course the photo was not. While I considered setting up the tripod and a timer shot, the fact that I was walking in boxers ruled out any photo featuring me. And I'd probably block some of the view of my fantastic workmanship.