Lunaran

Yes! I stated the polycount as 663 but I forgot to triple some quads so it's actually 828. i am a terrible artist. Glad everyone likes it, though. (also: Honda CB350)

something different o/ 128 x 128 x 663. not flawless, but I've had my fun.

I've finally had the time (oh games industry ) to write up the information on how all the script I wrote for this works. http://www.lunaran.com/page.php?id=220 Includes Carl Sagan reference.

Thanks for all the kind words, everyone Writing it all up this week in fact. I, uh ... didn't comment any of it. This completely bit me in the ass every time I stopped working on the script to play Minecraft for three months straight. and now you CAN'T ANYMORE. last goddamn time I work on anything with you 4 srs

With KungFuSquirrel on layout and gameplay, and John Bonham on drums. Brushwork is procedurally generated with some crazy maya python I wrote, and lit with an HDRI dome of sunlights generated from the skybox images. Verbose-mortems, downloads, and more pretty pictures at our websites: http://www.lunaran.com/page.php?id=218 http://www.button-masher.net/content.php?i=1379

Thank you! I, uh ... have since realized that the layout kfs and I came up with is just too big to support that much geometry, so that map is kind of dead. It played great as a blockout (dur) but the detailing was just driving the framerate into the floor. We're gonna keep the layout and I'll do something else with it, and come up with a more tournament-sized, less open layout to cubify. So it will live, just not in current form. Looks like a Pinto or a Gremlin, and a Chevy Caprice.

it's at 20mb at samplesize 16 half-detailed atm, and I want to build the final at samplesize 4. so, math says ... uh, 640mb? oh shit that cannot be right. basically I know deep down that this map will never work and it's sort of a race to see which limit I'm going to hit first. :derp: I've already switched to netradiant-q3map2 to get around MAX_EDGE_LINES and raised com_hunkmegs so q3 has room in memory for it. MAX_LIGHTMAPS may be next.

It's not really that. If you pull down the white point then you wind up with a lot of intense contrast in the texture that looks inappropriate in-game, and if you pull up the black point you get that awful grey haze. You're better off going back to the photosources you started with and repeating your work with a different target in mind. I'll shut up and post a screenshot now.

The histogram of three of those four textures peaks at 45/255, which is about 16%, and the bottom-right one peaks at around 85. I noticed while poring over Valve textures that a lot of theirs will peak just below 128, with a curve something like this: This makes a lot of sense for working with overbright lightmaps: everything can be double-exposed up to near white without losing saturation or turning yellow, and dark shadowy areas still retain a lot of information. You can separate textures a surprising amount just playing around within 96-136 or so, and it lets you exploit the full range of lightmap values when doing your lighting without ever coming close to big blobby hotspots. this post got a little but I do try to stick to the above when texturing and deviate only with a particular purpose. It's hard to fix a too-dark texture without it either looking too grey or too contrasty unless you just redo it, and then you've got to re-light the whole map to restore the original look (or just compile with -scale 0.5 I guess), so it's probably not something you want to tackle at this point I suppose.

@Hipshot: How dark are your textures? The interiors have lots of splotches that are visibly at full lightmap saturation, but still look dim/dull. Or do you roll with overbrightbits off?

Q3Map2 radiosity, skyboxes, and higher texture resolutions help make Quake3 look a lot better than it used to, which guys like Shadow keep proving, but I always kind of look at Q3 screenshots and wonder what next step we're missing. I want to take every shader in that screenshot, give each one a heavy vertex-alpha-based filth blend and put 0% brushes in every groove and corner - basically treat architecture the way Simon O'Callaghan does his rocks - but the manual brush splitting would take ages and not be very creatively rewarding to do. If Q3map2 did that by itself, now, that'd be cool, but my fu ain't strong enough to poke around in the compiler at all effectively. That code still reads like black mathemagic to me. Part of me wants to try another map after cubespew where I assault it with every tool q3map2 provides - static meshes, projected decals, ludicrous lightmap scale, the whole enchilada - just to see how visually competitive vanilla ioQuake3 can still be when supported by an onslaught of environment art. (that part of me is clearly not in charge of managing my time)

what, no way, why is this not finished

Let's see what I've got here. Unfinished industrialized art pass on a map I designed specially for the Quake4 team to play Rocket Arena 3 during lunch hour.

Func_msgboard did this once but reduced it to 768 because the Quakeman is only 48 units high. It seems generally 1024 in Halflife will map to about 16 mans in your game of choice. Therefore, if we're doing this proportional to the player we should stop calling it 1024 and go with "The 16 Men Contest." I don't see anything wrong with that name.

Voting for the door just because I think you're going to be really surprised.