Just a thought…
What if you would set milestones for each phase with a set deadline. A couple of weeks for modelling, then submit your model, then a week for unwrapping and submit again, two weeks for texturing and again submit and so on …
This way the tasks are kept relativaly small and easier to finish. This method works great for me with other projects and assignments and keeps people from wasting all there time in a particular area and loose interest.
I know that your strongest allergy is Fantasy art, but don’t forget these other great cliches!
Marines/Army Soldier W/Stubble & Cigarette (Cigar If He Is A CO Or Non-Com)
Naked Robo-Chick With Huge Gun/Swords/Titties Of Death
Impossibly Over-Armored Space Warrior
Female “Fighter” Who Actually Just Looks Like A Stripper (Improbably Costumed For Battle)
<backontopic>
I think doing characters for a non-action (and for the sake of dodging the fantasy archetypes, non-RPG) genre or story would be pretty cool!
Unconventional genres:
Racing (Street, Sim, Futuristic)
Puzzle
Strategy
Dance/Motion
Edutainment
Sandbox (simcity, etc)
Some of these lend themselves to art better than others. Andy, your suggestion of doing Dance characters has already got my brain going…Also, milestones sound good to me too, I think many of the people here have pretty crazy schedules, and it’d be very helpful to say HEY GET YOUR TRIANGLES IN 2 WEEKS EARLY OR YOU ARE INELIGIBLE. People can still finish their complete model in time, but since its unofficial, no tears, you still have a cool thing for your portfolio.
I totally agree that doing an environment-only challenge could push a lot of people away. However, it seems kind of lame that since they are a minority, NONE of the comps are about environments? Not sure what the best solution is, but we are starting to have some REALLY talented level designers on the boards, and it’d be cool if they could participate too.
Lol - that one always cracks me up - why do robots need tits anyway?!?! Lug nuts for nips?
Sigh.
few ideas off the top of my head:
-sentient plants (killer plants or just intelligent ones)
-insectoid mecha (mecha that has to be fashioned after an insect/arachnid form)
-remake a previous 3d fighting game character (new outfit, new style, maybe new story even…)
ummmmm I got a few more, but I think they’ve been done. Maybe something with cars or environments tho, cause they are really important part of game art but rarely gets enough attention.
-KDX-
What about realistic subjects? People from a certain tribe? A specific animal or any animal from a certain family. What about politicians or people who work along the road with a few construction vehicles. How about a few highschool girlfriends without going for the stripperlook… or a maybe a complete footballteam. Some typical old people hanging around the block on a bench… maybe mom, dad and grandma (posing for a familypicture) ?!
It doesn’t have te be fantasy.
What about self-portraits. And then we gather everyone for a deathmatch and who ever wins is the winner of the GameArtComp 
Seriously though. I think it´s a matter of timing. Some contests wont have many entries because alot of people might be caught up in work, school or whatever. I usually dont get much free time for my personal projects, and sometimes I hardly have any…
Im really keen on the idea of making a stylised, charicatured model of a famous person.
Would be a good challenge of perception, as well as modeling skills.
I may have a different proposal.
There could be a set of predefined phases in the comp, with deadlines set to each of them (i.e.: concept, model, UV, texturing ), but define a week from the finish of one phase to the beginning of the next. That week could be called a “turning point”.
During turning points people could join in the comp, taking the entry they like most and continue with it (this could lead to a branching in which, for example, the same model is being textured by several people); while others could resign and leave it there.
Judging points should be assigned to each of the phases of the project and the merit of winning the comp should be averaged among the people that contributed each phase.
This could probably help people join in at the phases they like most whithout feeling compeled to do everything.
Just my 2 cents
P.S.: A contest proposal: Take concept art from any established car designer/manufacturer and create a racing game car model. This comp would focus on the fidelity to a concept not created by you, as this is a much needed skill when you have to work in a team.
Haha, Im with Ivars on this one - slightly stylised self-portraits would make for a great challenge. :applause:
I think that’s similar to the last comp…
and imho it should be more narrowed down than just “famous ppl” cause you’ll be getting entries all over place from comic books to obscure novels to television and movies to politics…too broad overall imho.
what about some static meshes???
-KDX-
How do people feel about this? Would they make their mesh/models open for the taking and using? Some people might be protective of their work.
i think making that mandatory would push some people away. it might be an interesting option, but its a tad linuxy for my tastes
I like Andy’s idea with the elves and orcs!!! All in favor! j/k.
I do think Neil has a point. Go back and look at the people that start a comp and then drop out. Partly because they get frustrated but partly it is the fact that some of us dont have as much “pull” as say Bentllama or Andy H or other people that have shown off their work. They get the posts and watched threads and when we newbies dont get the critiques we bow out. I am not bashing anyone - just saying its hard to compete with the shoguns here sometimes.
As far as any ideas that come to mind, I do think a specific genre of people or era would be cool. I think there was a famous actor/actress comp, but if not, that would be cool. Maybe a worker comp - plumber,carpenter,industrialists,cubicle guy/gal,nurse, - pretty much anything you could think of, but that does keep it a little under a specific type.
Just my 2
I dont think that many people drop out because they dont get posts (though it can be a little saddening) but because they just run out of time/interest. As far as what we can do to keep people interested, I think that if there is some kind of prize (like Bentllama is trying to get) than more people will stay in for the length of the competition
Just another Idea, but what if (with the designers permission) we require all models to be rigged, and then actually put the models into an engine, and distribute the game as an open source freebee to the people of CG Talk? Not only will more people finish their model to see it inside of a game, but it would also add to their resume, saying they helped model for an open-source project.
[i][b]Dont take anything I suggest too seriously, as I am just tossing ideas out there
[/b][/i]
Let’s say that, when entering the comp, each contestant must declare its entry as public domain (anyone can take over at turning points), restricted (you should ask for permission to take over) or proprietary (only owner can finish).
Or maybe upon signing up you must declare at which phase you are going to stop and set the entry to the public domain. This way, modellers can do their work and leave it there for others to texture.
In this latter case, every entry that reaches its announced finish state must be considered finished and be judged.
And it would let Adam Atomic go as Windowy as he likes 
I have only one comment.
Let the winner of last competition decide on the topic, which is rightfully so (unless I missed out something).
PKD: We’re just openly talking about what we’d like to see and expect from it. If it’s ALWAYS just the winner doing whatever the hell THEY want, then it runs the chance of less people participating.
I agree that the last UGAC went on for too long, I don’t feel the poly count had much to do with it though.
I like the sound of team entries, though getting a team together in the first place might be difficult, let alone staying together and coordinating.
Doing one thing is a bad idea in my oppinion, I hate texturing (infact I’ve never done it propperly at all), and I’m sure there are other’s that don’t like modeling, or rigging. If the brief involves every process then you’ll please most of the people most of the time.
Making something like sceenery sounds interesting, not everyone will get the chance to make the next Lara Croft or Nissan Skyline. Often the un-noticed things are what makes or breaks a game.
I like the ‘do a bit and pass it on’ concept, although I don’t see how this could be judged, unless you get teams together (see first bit of post) and you set the brief like ‘make a robot’, and then someone concepts it, someone else models it, someone else unwraps it, someone else textures it, someone else rigs it, someone else animates it. That just sounds like a team project to me though.
Going with that last though, you could do team projects, but have a ‘winning team’ and then winners in each process.
^^ Just me rambling ^^
Let me just say 2 things because it’s coming up a lot:
a) why do people say they won’t finish somethign if it involves a certain part they don’t liek doing (insert = texturing, rigging, modeling…), I thought the point of this comp is to LEARN and challenge yourself?
b) why is everyone so concerned with the judging. it’s not like you a get medal, nor do you put it on your resume. It’s just a little fun. The judging is just meant as a way to say congrats and pat on the back. There doesn’t need to be a scientific scoring system or regulations of qualificatiion. I mean if 3 people contributed to a model, or 1 person, we will still say “nice work” or “nice try, it’s bad”… doesn’t really matter.