C2D 6750(@2.66Ghz), 2Gb RAM, 8800 GTS 512, XP32

I think that if a level is well designed it's probably good. By well designed I mean it looks good, plays good (r_speeds aside) and doesn't leak. Vis blocking can be used at the players discretion nowadays (ooh, I'm gonna take some shit for saying that), but a level must be fully vised. If it's not fully vised then I see that as poor mapping. I mean for a finished map, all other things aside, the level must be visually optimised. If a mapper can't get it to vis then they should go back to the drawing board. Also I think there should be no areas where packet overflows occur (probably guilty of that too).

OK . . . . 

If it takes weeks if it doesn't make it that much better for low end comps anyway and with high end comps it's still playable with fastvis.
I have a low end comp (1.8 GHz) and can't play em anyway. 

all of these maps. Sickbase was made on an Athlon 1.7, and I was playing at 1024x768. Something like WarpC has much better vis-blocking than Sickbase so it should run faster.

What is your GPU Bambuz? 

But if people really want to stray far from what quake was built for and considering not even vising map you're really better off throwing away the standard .bsp compability and mod the engine for something that's more suitable for what you're doing(q3 .bsp might be good enough in many cases). 

r_speeds be damned, quick VIS passes, huge levels - do you really want to be making Quake levels? I think maybe you're mapping for the wrong engine. 

 

 

All maps have to be vised, no matter what. Even with a powerful machine, a large map will run poorly unvised. Just try r_novis on one. If vis takes weeks to complete, then so be it. It's basically the mappers fault for making the map too open, intentionally or by accident.
There are few occasions where fastvis is enough, simple space maps for instance - in this case, there's often no real difference in the vis result compared to a fullvis, except for the duration of the process. Remember the long delay of the release of Travail? That was due to my stupidity of not realizing that a fullvis was totally unnecessary for qtfin2. :P
But in most other cases it is and should be regarded as mandatory.

High(er) r_speeds, on the other hand, is somewhat acceptable nowadays, since most people's machines can handle it easily. Back then, the rule of thumb was trying to keep r_speeds under 600, maybe 800 if justified. Today, I'd say up to 2000 is ok for a single player map, though one should still try to tweak it as much as possible, of course. 

If you release a map without doing a full vis on it, you're an asshole. 

Slave on a fast vis was >10000 in places (!)

It took 60 hours to vis, and r_speeds are more than halved.

It's still a good example of what you DONT really want.

I look up to stuff like Marcher Fortress, Insomnia, Aderlass2, er WARPSPASM etc. These are examples of maps which look visually impressive in many places but have good r_speeds.

I think the only map which I have released which did have good r_speeds was the very first one, dvstart. ! :-|

Thehand was good except for the final area, looking towards and beyond the GK door was a bit hefty.

ALL maps were full vised though.

e1m4rmx will be better. Vis-blocking is already done in the layout, so I cant take much credit for that, but atleast it should have good speeds! :P 

This and other recent discussions have given me inspiration for my next map. It's a bold, grand concept, so brace yourself!

I'm going to make a map. Not just any map though. It's going to be epic. At least 1,000 monsters! 100,000 brushes! I'll know I'm on the right track when I start breaking all of the standard Quake engine limits, but at that point I'll just be warming up!

I really couldn't be bothered with tedious stuff like vis blocking. It doesn't matter though, because I'm not going to vis it anyway!

I'm just going to keep adding stuff until the compiler won't build the map anymore. At that point, I'll delete a few brushes and call it done! By this time I'm sure the map won't run in any engine known to man (another sign that I'm doing well), so I'll have to get my own special .exe with massively increased limits so I can run the map. Maybe I can convince someone to make me a special version of QBSP as well, so that I can add even more stuff!

Common sense? Self-control? Testing? Compatibility? Fuck it all! It's all going out the window! This is ART, man. Don't oppress me with your bullshit limits. If you don't like it, you just don't understand it. You just don't understand ME.

Guess what? It doesn't matter if you don't appreciate my art anyway, because the map will only run in my very own special engine... and you can't have it! 

post of the month

But some engines can run Q3 maps amirite? So why not use that format? 

 

To make a map which was right next to all of the limits.

Hard to do though!

The thing is that one could make a HUGE map, and stay within the limits. 64 lightmaps is quite a limiting factor, but if parts of the map are made up of rocks then scaling the textures up would increase the possible size of the map.

So has anyone ever made a map which was at all of the limit?

Which map was the closest? (AguirRe?) Breakfast at Twilight must have been close! 

Awesome post, you've captured my method of mapping exactly, and it's the way plenty of others operate nowadays too. Lol.

I've found that ever since I started using Hammer instead of old WC, that the "automatic scale" of things I build is bigger. I think it's the 3d accelerated, quick-moving camera. The ability to get around areas in 2 seconds in-editor in smooth 3d makes it really difficult to stick to rooms 'n hallways. That said I've always hated rooms 'n hallways.

I do vis my freaking maps though. Not to do so would be a jackass thing to do as stated above, even if it does take 2 weeks (as one of my new maps does, although Masque was only a few days). It's the mapper's fault/choice/responsibility. If something looked awesome and it was unvised there's no way I'd download it. 

"That said I've always hated rooms 'n hallways."

Again - are you sure you want to make Quake maps? 

What about all of those coagula maps? The ones where everything is inside the void? 

that although people want to make maps that are probably better suited to other engines, those other engines dont have games that supply the same fun gameplay as quake. 

That featured the earth in real scale. So you just dropped for minutes until you hit the ground.

Maybe an alternative to traditional vising could be developed for huge maps for fast computers. Some coarser way of boxing visible areas. 

But none of those ways include using an eleven year old engine that had a performance target of being able to render ~800 polygons per frame.

Quake seriously is best when you play on the game's strenghts. Squeezing out all these shitty huge maps just because you can doesn't make them good.

I hate 

It's not like there's an overly large amount of limit-breaking maps already, and I doubt there will be that many in the future, so much of the sarcasm and hating seems out of place. 

I don't understand how making large maps is _offensive_ (no one's forcing anyone to play them). It's kinda hard to argue against Marcher and Breakfast at Twilight dude, though I suppose you do need to make a hater comment on every thread.

And yes, Willem, I do (or did, when I was building the architecture I'm now setting up gameplay for) want to make Quake maps even though I don't like maps that are pure rooms and hallways. The Shadow Over Innsmouth was a fucking great example of the kind of map I've always wanted (and generally failed) to make - something which pushes Quake into new, outdoorsish territory, yet retains the dark atmosphere and netherworldly feel. 

[Isn't The Point] that although people want to make maps that are probably better suited to other engines, those other engines dont have games that supply the same fun gameplay as quake.

quoted for copious quantities of sheer unadulterated TRUTH.

It's not like there's an overly large amount of limit-breaking maps already, and I doubt there will be that many in the future, so much of the sarcasm and hating seems out of place.

as above. 

my private opinion is that building intelligently, with vis blocking and some basic limits (I really mean the standard protocol here) in mind, and with basic compatibility and portability in mind, is just good practice (tm). And when you build according to good practice, which any good craftsman does, you will have reasonable r_speeds.

However, when you're building Warpspasm or Nehahra, it might be that other rules apply. But yeah, that's not the norm.

The problem with Warpspasm-type maps I think are not puny limits like static entities, not even huge open areas, but the standard protocol. There are some limits that you can't break without breaking the standard protocol, I think the model count for example.

So r_speeds aren't really the problem I guess, although I still think that a well-built map will automatically have bearable r_speeds. Personally when a map goes over 1000 all the time, I start to ask stupid questions :-) but not many maps do. Not even Ijed's. His maps play well indeed, even for me on my old laptop without hardware accel. I mean, even Marcher runs OK for me (!!!) so if a map requires "gamer hardware" then sorry, the map must be faulty.

It's probably the amount of stuff crammed into a map. The obvious solution is to split it up. Map loading time practically isn't a factor in Quake... and a hub system would even let you carry keys across maps (Quoth does.)

JPL did it with 5rivers. That wasn't so bad I thought. It would be nice if mappers could disable the status screen between parts, though. Or maybe that's already possible.

Neg!ke split his Zer map up, too, I'm sure there are more. It's like, definitely do-able :-) You don't go blind from doing it :-)

That's another benefit of hub-type maps, a lot of limits don't apply anymore. Sure you can have 1000 monsters and 300 torches, in a 5-part hub. No problem. It will even run in Winquake.

Think about it. 

we want PORNO!!!