AEnoch ...
I think in terms of approximation, 1 foot equalling 8 units is about right.
7ft for the player doesn't sound to outlandish. Dolph Lundgren stands about 6ft 6in and I would have expected the player to be about the same physique as him.
And I would also expect someone of this physique to be able to jump about 4 feet.
-Mic
SketchUp Converter ...
The reason I'm asking for this is because I'm currently working on an .skp (Google SketchUp) -> .map converter, but the maps when loaded after conversion are too small because SU works in inches & feet and 1 foot doesn't map to 1 Quake unit. Therefore I need some sort of scale to work with.
-Mic
Re: Willem + Looping
#8837 posted by necros on 2009/07/14 20:12:45
If you jump off an unsealed level and loop a few times, you can still land on the level itself and have proper collision
this actually isn't the case. if you loop around, the geometry is drawn as if you had looped, but everything is non solid and monsters cannot attack you, even though they look as if they are able to.
any geometry that goes beyond the boundary though, will be solid to you when you loop over. but it won't be visible.
it's like the client is running on floats but the server is running with doubles.
Re Units
#8838 posted by necros on 2009/07/14 20:14:13
maybe this is for hl/hl2, but i believe they do 1 unit = 1 inch which works out good for like stair heights and such, but i think the quake player is shorter than the hl2 player.
Quake Guy
#8839 posted by spy on 2009/07/14 20:20:17
even shrter than the q2 guy
#8840 posted by JneeraZ on 2009/07/14 20:24:46
necros
Oh yeah? Weird, I could swear I remember doing that but maybe I never actually tried it.
#8841 posted by metlslime on 2009/07/14 23:39:21
it's like the client is running on floats but the server is running with doubles.
Literally, the server is using floats and the nework protocol sends them as 16-bit fixed-point numbers, with 13 bits for the integer part, and 3 bits for the fractional part (so the smallest representable difference is 1/8 of a unit, and the largest possible value is 4095.875) The client converts them back into floats, but of course you've already lost the extra range at that point.
Also:
#8842 posted by metlslime on 2009/07/15 00:25:28
you can see this 1/8 thing if you stand on a slope (like the beginning of e1m1) and zoom in on the wall nearby. As you slowly slide, your view ratchets forwards and downwards in noticable increments.
When working on protocol 666, i tried various techniques to add range to the coordinate system without increasing packet size. They were all very complicated and still made packets bigger. The simplest is just to always send floats, but this makes packets much bigger and I didn't want to pollute the protocol that way (as it is now, the packet sizes are about the same as 15 until you need to send special data like alpha, 2-byte modelindex, etc.)
I still have plans to support this in the future, I just wanted to get a functional version out and that requires locking down the feature set at some point.
The packet size limit is the toughest part, and there's not much I can do about it (but I should add a cvar so you can attempt larger packets on a LAN, which might have higher MTU than the internet at large.)
#8843 posted by necros on 2009/07/15 00:36:12
angles are another thing where the precision is incredibly low.
the hanging corpses that slowly rotate in quoth are animated in the mdl because the engine's angles round to like, nearest 5/360th instead of nearest 1/360th degree. when you start doing rotating brushes, and the brushes are large (or are placed far from the origin) it can get incredibly ugly with this gigantic gaps between rotations.
it would be nice to beef up whatever handles angles if possible.
Necros:
#8844 posted by metlslime on 2009/07/15 00:57:13
protocol 666 does have high precision client angles (so aiming your gun is accurate) but not on rotations for entities.
I noticed the problem with hanging corpses but I thought the rotation was a model animation (so they could be static ents) rather than entity rotation. I guess that's a good feature to add to the next protocol.
Though, It seems like something that mappers/modders should avoid relying on, otherwise you'll have the huge ugly bmodel rotations you talk about happening in all other stock engines. I had this problem with a rotate train in rubicon2 and the fix was to spawn it near the world origin.
Monster Action Measures
#8845 posted by madfox on 2009/07/15 04:06:35
maximum distance a player can jump without falling into water, lava, or traps
225 units
maximum distance a player can jump straight up and get up to another object:
42.5 units
maximum distance a player can fall and not be injured:
275 units
minimum gap in the floor spacing, slatted bridge or floating squares of lava:
35 units
maximum height that a step can be before a player must jump:
17 units
Units ?
#8846 posted by Ron on 2009/07/15 07:28:21
What is a unit ?
I always have trouble making lifts go the right distance, most of it I guess and then adjust.
When you speak of units, how do they relate to the grid in Worldcraft ?
#8847 posted by RickyT33 on 2009/07/15 09:53:33
(makes "shoot myself in the head signal")
FFS, its just been repeatedly explained not a couple of posts previously (like yesterday)
1:1, one grid unit = 1 unit, one 1:1-texture-pixel
More On Real World Unit Size ...
After researching this more, I have come across the Quake Editing Primer at ...
http://gamers.org/dEngine/quake/QDP/QPrimer.html
... which suggests 1 Quake unit is about 3cm or 1 foot is approx 10 Quake units.
This seems to make more sense now, especially with what others posters have suggested above.
The only thing that still puzzles me is the standard height of a step being 16 units. At 8 units equalling 1 foot, 16 units or 2 foot is far to high I feel, and although 1.6 foot (using 10 units per foot) is getting nearer the mark, I feel this is still too high.
5 ft 6 for the player size, and an 80 x 80 room equalling 8ft x 8ft rather than 10ft by 10ft perhaps seems more reasonable.
What do you guys think ?
-Mic
More ...
It also suggests that vertical and horizontal scales are different ...
Vertically : 1 unit = 3cm, 10 units = 1 ft
Horizontally: 1 unit = 2cm, 16 units = 1 ft
-Mic
Don't Take It Too Seriously
#8850 posted by negke on 2009/07/15 13:21:07
I always had the impression 64 units roughly corresponded to 1 meter. Then 16 unit steps would be about 20-25cm which kind of makes sense for the world. The scale for entities is totally different, though.
In the end, it's no use translating RL scale to Quake units - the dimensions of the environment isn't determined by logic, just by the textures.
<- Face It
#8851 posted by generic on 2009/07/15 14:10:09
The scale in Quake is shot: a 16 unit step/stair just doesn't look right with a player model next to it and larger monsters look ridiculously small when viewed from above. Go with what feels right and don't get too caught up in the unit conversions.
#8852 posted by JneeraZ on 2009/07/15 14:57:05
" the dimensions of the environment isn't determined by logic, just by the textures."
^^ This. Just work with it until it feels right.
Like I Said
#8853 posted by AEnoch on 2009/07/15 17:24:52
It's a matter of perception. Do what looks/feels right, and adjust from there.
It Is Relative....
#8854 posted by JPL on 2009/07/15 17:29:53
... to players size, that is 16 units tall if I remember well :)
To Add Just A Bit More...
#8855 posted by pjw on 2009/07/15 20:02:26
When you're dealing with dimensions/proportions in a game engine, you're almost always going to have problems if you try to rely on exact conversion numbers and translation of stuff directly from the Real World.
In FP games in particular, the camera height and FOV are going to differ at least a bit (and often significantly) from what you're used to seeing in real life, so some things will feel right and some won't.
And then there's the problem of stuff being on a small flat screen, rather than actually being in front of you...
Btw
#8856 posted by necros on 2009/07/17 04:27:58
there is a limit after all to how far out you can build. it appears to be 8192 (doubles?) where, anything after that just doesn't draw.
Necros:
#8857 posted by metlslime on 2009/07/17 05:58:55
Hmm, I guess there are probably some client functions that assume the original bounds. For example:
- the far clipping plane
- the skybox size
- the trace vector used to light entities based on floor lighting (was 2000 units in original quake, now 8192 in fitzquake)
#8858 posted by necros on 2009/07/23 08:08:28
you guys ever go back and play some of your really old stuff?
i was playing nesp09 and i discovered my lame sense of humour hasn't changed at all.
#8859 posted by necros on 2009/07/23 08:10:31
i guess what i was really driving at was: do you enjoy your own maps after sufficient time has passed that some or most of it is almost like new?
|