These notes about Quake3 were posted on Blue's News this morning. I've bolded parts that are TDFX relevant.
==========
John Carmack gave his talk yesterday afternoon on Quake III: Arena (yes, one of the things he mentioned is that is the title, and not simply Quake Arena). Here is my best effort at what he talked about being included (or not), or planned for the game at this point (this is in the order they came up, not in order of perceived importance):
+ Per-pixel specular light maps, described as good for metallic textures, and also helpful in creating natural lighting in brighter areas (so levels will not need "dark" design to look good).
+ Morphing between multiple textures (good for explosions).
+ A generalized mirror interface (can be used to create a rotating mirror, for instance).
+ Levels will feature non-rational Bezier curves, which, especially when combined with the specular lighting, can be used to create more realistic curved surfaces.
+ There will be no curved surfaces on characters, but it is conceivable that a mod could add this.
+ L.O.D. (Level of Detail) has been implemented (not infinite), so the face counts of objects will drop as they get further away.
+ The editor will no longer break single polygons into many. Under most circumstances, one poly in the editor is one poly in the game.
+ Bezier curves can be used to create outdoor textures like rolling hills (though nothing like mountains).
+ There will be no terrain renderer in the editor--outdoor areas will not be a strength of Q3A. The format of a level is currently ASCII text (it's uncertain if this will remain this way). This offers flexibility, but adds about 50% to the size of the levels (less of a concern since Q3A''s multiplayer nature will result in smaller maps anyway).
+ Arenas will be specifically targeted at two, four, and eight player games, etc. Larger maps are possible, and may be included, but will be "unsupported" since it's going to be hard to load 64 bots into a level.
+ The player model will be broken into two pieces, allowing separate movement of the torso and legs, allowing effects like running while firing, and a more realistic depiction of where a character is looking (looking up and down will no longer result in tipping back and forth on your toes).
+ Models will have "tabs" where the weapon models attach. The same weapon model is used that you see in first person, in someone else's hands, or to be picked up (to create a new weapon for the game, a mod author needs to create a single frame of the model).
+ Weapon changes are really fast: six frames or so, it's apparently pretty fun to watch someone (or even yourself) whipping through the choices.
+ Muzzle flashes for the weapons return (the addition of frame interpolation in Quake II "broke" this).
+ The game will likely be the best looking visual game of its type when it's released, but this is not the top priority, QA3 is not an effort to push the envelope significantly in this way.
+ BSP, Vis and Light: BSP is essentially the same. BSP will stop sub-dividing polygons. Light is not yet determined: He's exploring "photon tracing," but this may be scratched because it is so CPU intensive, and also, determining photon collisions with curved surfaces problematic. Lighting will not be the "right thing" (tm) but will be a visual improvement (the numbers demand a trade-off on lighting). John would like to get to where all light is sourced from real objects in the world, but the level designers are fighting this.
+ 3D Audio has never been a high priority, but sound cues are more vital in multiplayer play, so this 3D audio may end up included.
+ He will probably include voice communication. This could have been in Quake II, but NT does not support DirectSound input. If this is included, it will likely be uncompressed for high-bandwidth users.
+ Why no bump mapping in Quake III Arena: John feels this is an effect you need to stop and examine to appreciate, something that is less likely to happen in Q3A's multiplayer environment. B.M.'s carry a large overhead (and he doesn't want to ad it as a spot effect: he feels rendering effects should be fully implemented), and creating fully bump-mapped textures would more than double the work for the artists.
+ Also, four pass rendering will not be possible on any but the fastest next-generation accelerators (TNT and Rendition's 3300 were cited as examples), not appropriate for Q3A's emphasis on fast, fun multiplayer play.
+ On cheating. It is impossible to completely prevent cheating, but every effort will be made, including possibly pursuing legal options (making a strict EULA and prosecuting some offending cheat author). Interestingly, John feels that some of the cheats he's seen could possibly be the result of reverse engineering, but he suspects that someone has had some illegal access to some source code at some point to create some of the things he's seen.
+ Q3A will include CTF out of the box, and Zoid will create the first add-on (undetermined: the example he threw out was "CTF Pro")
What's next?
The success of this project will dictate id's next project, a success will probably spawn similar titles (QuakeLeague was given as an example). He would like to continue pursuing multiplayer play and leave single player to licensees.
Surface Voxel technology is not as far off as many think. It will not be very long before you can render "Toy Story" quality images in real time (though he points out the frightening implications of how many artists and how much time was required to make all that movie's backgrounds).
Doom and Wolfenstien 3-D sequels are still possible, perhaps done by another company, but he realizes that these projects are quite compelling for many. He intriguingly dropped the title Doom2000.
===========
Chip |
