ETA: For the uninitiated, LMAT stands for "Let's Make A Tileset," and represents community projects to make tilesets, in which anyone may participate. Here is the list of past LMAT sets.

We're currently between LMAT projects, assuming that Valentine's Day will be released soon and that no one ever wants to think about our failed beach set again. The LMAT success rate is actually surprisingly high, but it's quite slow, so I thought we should try inputting a little bit more structure into the proceedings and see what happens. This will mean a tileset a little less abstract than, say, Kansas or Aftermath, but I think we can collectively come up with enough eyecandy to make it interesting. Here's what I propose, although this is open to discussion and adaptation:

Phase 1: Collectively decide on a theme. Here are a bunch of tileset ideas and we can come up with some more here if necessary and ultimately try to agree on one. If no consensus can be reached I will do some private consultation and emerge with a decision. Name may be decided during this phase or may emerge later.
Phase 2: Come up with more detailed concepts. Concept discussion will be created by asking people not to start drawing for a couple days after a theme has been decided upon. This can include concept art (in any medium), screenshots of other games, written ideas, etc. Stuff that develops the idea beyond a general theme and gives us specific items, floor types, layers, whatever, to work on. For Aftermath, for example, things like the lockers (ruined and otherwise), rotating warp pillar, and nuclear background refinery might be speculated here.
Phase 3: Initial Artwork. This is mostly for the most important tiles, such as the basic ground and whatever else is essential to the particular theme. Background layers, maybe, if we're feeling fancy. The basic ground and other such tiles determine the artistic feeling of the entire tileset so it's important to get those done first before breaking off in other directions. As has happened in some previous LMAT sets, this is a good time for multiple people to try drawing the same thing, so that we can decide what looks best rather than being stuck with the first few tiles forever.
Phase 4: Fleshing out the set. Basically the equivalent to an ordinary LMAT development, with people drawing the rest of the tileset, ideally in a visually consistent style. One idea I've been toying with in my head is people being assigned individual parts and then presenting them for criticism and editing, rather than people doing things at random without prior warning.
Phase 5: Editing. Again, a basic tileset procedure: changing up the tile order to be more user-friendly, discovering missing tiles (particularly connecting tiles) and adding them in, and so on. Will probably have a blurry border with phase 4.
Phase 6: Postprocessing. Palette(s?), mask, example level. To forestall what has happened with Valentine's Day, here, as with everything else, if someone is assigned to do something and then simply never does it, someone else should take over instead.

Thoughts? The main point is that we come up with ideas before we start drawing, everything else is just formalization. One theme I'd like to see is a good shipboard tileset, with sails and cannons and all that good stuff. Maybe some island tiles as well, just to be nifty. What are your ideas?