Here’s that post I promised explaining the content you get on Inventing Situations.

1. Why Public Domain?
   Yes indeed, other cap sites and caption galleries use copyrighted images and back when this site’s forbearer was running “underground”, even I used them. But, I’m not really convinced that if I serve up 200 images or so from a film it suddenly becomes fair use or parody because other people add funny comments. The fact that other sites aren’t getting nailed doesn’t mean that I won’t. I really just don’t want to sit around worrying about if and when I’m going to have to fight off a cease and desist.
2. So, there must be plenty of places to get good public domain images on the web?
   You’d be surprised. Most sites offering PD content don’t make it easy to pull down lots of images, and if they do, the content is, on average, more monotonous and less cap-worthy than what you’re getting now.
3. So, how do you pick the images?
   The initial stockpile of images was just taken from the automatically generated thumbnail pages at archive.org (like this one). This was a necessity to get enough images to start with. Depending on my mood, I exerted varying degrees of effort to remove the real junk images when I first put them into the pile. Since then, new images are usually added by downloading the entire film, watching (or fast-forwarding) it and picking what I think are the best grabs. With longer films, I do the hand picking a little less carefully, and just automatically grab about 50 to 100% more images than I want, then throw out the losers.
4. So, that’s why the movies don’t play back in real time?
   That’s part of it. Sometimes I may get over 30 good images for a 10 minute short, or 250 grabs for a 90 minute feature. With the preferred timer interval of 30 seconds, these are both guaranteed to run over. Other times I get the opposite ‘problem’ and a 10 minute short will only give 6 good grabs. The other big issue in play back speed is that two things have to happen before a new image is served up - the timer has to hit zero, then somebody has to ask for a new image. I’ve got no control over how long the second step will take. Which means that when I run features, I either have to trim out a lot more potentially cap-able images, be prepared to let a movie run for three hours straight, or cut the movie off midway and start it again later. I’m going to start doing the last one, as I explain in the next answer.
5. What’s the composition of the collection? How are those chunks getting rotated?
   

   Source	Image Count
   The Prelinger Archives & The AV Geeks Archive	24768
   Stills from LOC, COE and other sources	2400
   Feature films (162 films)	21522
   Other shorts (cartoons, trailers, serials, etc)	1840

   The stuff from the Prelinger archives is divided into two chunks of roughly equal size - the stuff that’s been in rotation in day one, and all the stuff that’s been added since then. Those two chunks plus the AV Geeks material are essentially rotated endlessly. While the order of material within the big chunks remains more or less the same, sometimes the new stuff is followed by old stuff then followed by AV Geeks, sometimes it goes old, new, AV, and sometimes I drop the old stuff for one pass through the rotation, but you get the idea. I just recently re-introduced the stills, features and “other shorts” listed above into rotation, and I’m trying to do so in a way that’ll keep them fresher without completely burying them. My new schedule is four one hour blocks: 1 PM & 11 PM eastern - Randomly ordered still images from the Library of Congress et al, 4 PM & 10 PM - features and other shorts. The theatrical material will be shown in order. If 5 PM strikes and we’re in the middle of a movie, you’ll have to wait until 10 PM to cap the rest. If you can’t make it at 10, you’ll have to wait a few weeks til it comes back up in the rotation.

6. Why don’t we see more new images?
   Just at the moment, I’m on one of my active sprees where I’m putting one ‘new’ movie into the rotation most days. Of course, it’s usually not really a new movie, it’s just handpicking grabs from one of the movies that’s been in rotation since day one with automatically generated grabs (I take the original images out of rotation). Of course, since we show about 500 images a day and I’m adding between 20 and 40 on a good day, the difference may not be immediately noticeable. I’m trying to increase the effect by putting some of the “new” movies into the older version’s spot in the rotation, instead of in with the “new” material. This way you have a chance of seeing some new grabs while we’re in the ‘dreary’ chunk of the rotation.

This entry was posted by GersonK on Tuesday, May 16th, 2006 at 3:28 pm and is filed under Caption Sites, Inventing Situations. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.