Tim's thoughts

I'm waiting for the day when every model can have a per-pixel spherical harmonic transfer matrix applied. Yes, I want more world detail, but the biggest for me today is the poor lighting. I've already gotten used to the stencil volume/shadow buffer look. Now I want more global illumunation solutions, and algorithms to deal with soft shadows and diffuse intereflections.
 
DemoCoder, correct me if I'm wrong, but aren't radiance transfer matrices useless as soon soon as a model is deformed?

Regards,
Serge
 
All fine and dandy, but I can see the userbase complaining far more about the lack of advances or new ideas in terms of gameplay, instead of graphics. When was it the last time when a popular multiplayer game managed to fill countless of servers as much and as long like let's say Q3a or UT1 in the past?

Ideally I'd like to have both and I personally highly disagree with the notion that we've reached a certain point in terms of gameplay, now we can improve graphics. You'll have to excuse the OT.
 
(I define this being when games use charachter models equivilant to the Dawn demo)

Dawn actually still need more polygons. They need to be able to achieve something that is beyond what movie uses. Something beyond Hulk or Golem. Until then they should keep going with the progress, and we are no where close to point of diminishing return.
 
Vince said:
Perhaps. Yet untill I can run into a crowd of thousands in a GTA like game - I'm inclined to call Bullshit!. I think the PC, perhaps strongly due to it's architecture, has traveled down a path where instead of creating geometry heavy worlds that are filled with many objects; has been more inclined to have few onscreen objects, which are highly detailed.

Your comments are only supporting this ideology. Which, IMHO is almost ignorant when you think of what could be done. And when you think of the 'sub-genre' of games which are becoming popular (GTA like, open ended worlds) the limitations are only more appearent. Maybe one Dawn is pretty and cool to talk about, but make me a city out of them and then gaming will have progressed,

Dude your tone is a little rude/impolite, I have no desire to be in a discussion with someone who has no manners.

V3 said:
(I define this being when games use charachter
models equivilant to the Dawn demo)

Dawn actually still need more polygons. They need to be able to achieve something that is beyond what movie uses. Something beyond Hulk or Golem. Until then they should keep going with the progress, and we are no where close to point of diminishing return.

We definitely are! Everything in CG gives diminishing returns, be it AA quality, texture resolution, screen resolution, or geometry resolution. Diminishing returns = when any given factor is increased the returns are smaller and smaller the more it is increased. Most all computer technology yields diminishing returns.
 
Ailuros said:
All fine and dandy, but I can see the userbase complaining far more about the lack of advances or new ideas in terms of gameplay, instead of graphics. When was it the last time when a popular multiplayer game managed to fill countless of servers as much and as long like let's say Q3a or UT1 in the past?

How about Counter-Strike? AFAIK it's still going very strong. The only thing that may have slowed it down is Steam, but I don't really know if Steam/CS1.6 has "killed" CS yet or not. I suspect not. Anyway, CS arguably has very stale gameplay now that its been out for so long. Every patch basically slows the game down further and turns it into more of a camping fest.
 
Vince said:
Your comments are only supporting this ideology. Which, IMHO is almost ignorant when you think of what could be done. And when you think of the 'sub-genre' of games which are becoming popular (GTA like, open ended worlds) the limitations are only more appearent.
How about massive online games Vince? Maybe the genre doesn't get much press compared to hightech crap like FPS, but that doesn't mean they aren't doing large open ended worlds - and it has been more or less exclusive PC domain thus far.

I agree with you to a point though - I think it's likely that the future pace of MMO games will be set by consoles, inspite of current PC domination in the area, partially due to factors already mentioned by the last few posts.

psurge said:
DemoCoder, correct me if I'm wrong, but aren't radiance transfer matrices useless as soon soon as a model is deformed?
They are, but nondeformable stuff is still quite prevalent for environments at least, and it'll continue to be so for a long time to come in some types/genres.
 
my simple, humble opinion is, we have not even come remotely close to diminishing returns in realtime graphics. in CG yes, we have reached diminishing returns, but not in realtime. with realtime graphics, we are just beginning to knock on the door of CG from the 1980s.
(luxor lamp demo for instance). we have not even reached modern (within the last 10 years) lowend CG graphics seen in television shows, television commercials, games, etc. we probably have at least another 20 years of realtime graphics advancement yet to go.


I would not mind seeing something CELL-like on the Desktop, maybe even paired to a nice GPU ( the two are not mutually exclusive even if for cheaper systems they could find it cheaper buyt still ok to work with embedded rasterizer and the CELL-like processor ).


ahhhh, the best of both worlds. 8)
 
Fafalada said:
psurge said:
DemoCoder, correct me if I'm wrong, but aren't radiance transfer matrices useless as soon soon as a model is deformed?
They are, but nondeformable stuff is still quite prevalent for environments at least, and it'll continue to be so for a long time to come in some types/genres.

Deformable PRT is currently being worked on as we speak.

Peter Pike Sloan seems fairly confident its 'doable' and he's knows a few things about the subject :)
 
Fafalada said:
Cool :) I knew about it being worked on - but it's good to hear that it's actually deemed doable as well :p
Perhaps slightly related to this was a presentation at Siggraph where they used some "complicated interpolation" between various models to manage both complex responses to forces as well (IIRC) lighting. Unfortunately I can't remember the paper's title but I think it was in dynamics session.
 
Someone mentioned MMORPG's and graphic evolution if you will. I feel this is a area of games that is growning faster than people think or give credit.

Game play mixed with social interactions is sort of scary. Those with better engines that can help the player visualize will have the most subscribers. FPS in these games is not a priority. However, world and character detail is.

So I can see a world with thousands of objects at once. Characters with well defined features, hair strands that move, chinks in armor move when I breath...

I would think data movement or bandwidth if you will, will play a bigger role moving forward. Something about people escaping reality and new ways to socialize is going to change the gaming community as the player base appears to be growing.

Dunno - just a thought.
 
saf1 said:
So I can see a world with thousands of objects at once. Characters with well defined features, hair strands that move, chinks in armor move when I breath...
From a graphics standpoint, you don't need to render all of those things in high detail at once. Only the very nearby characters would need full detail.
 
nobie said:
Dude your tone is a little rude/impolite, I have no desire to be in a discussion with someone who has no manners.

Don't worry about it. A few of the "regulars" always bring this (on-line, no risk) machismo with them wherever they go around here. It's nothing personal, really. :LOL: This is just what happens to a board after about 20 knock-down drag-out flame wars in a year.

On a serious note, I'm curious about this "primitive rasterizer - uber-parallel CPU" architectural paradigm. I'm not the sharpest tool in this shed by a long shot yet I can already see that it doesn't seem very robust at all. Here's an example: Let's say we take the calculate the angle between the view vector and the incoming light vector and use this to lookup a 1D texture which encodes a Fresnel reflectance function (since this can be a pain to calculate). We then look up an environment map to get the color value which we will reflect to the viewer. Cutting it short - we do several cycles of fragment calculations followed by a texture lookup dependent on those calcs. Now, since the "shader" units are physically on the CPU and the "TMUs" (ugly 3dfx term) are physically on the GPU we are swamping our CPU-GPU backboard bus with loads of traffic* every pixel. This, even with an ultrafast bus, will be a nasty bottleneck to say the least.

Or am I missing something obvious here... :?

*Shader, 5 int8 texture lookups
Passing 5 texcoords: 5 coords * 8 bytes = 40 bytes
Retreiving 5 int8 color values 5 colors * 4 bytes = 20 bytes
40+20 = 60 bytes per pixel * 360 Mpixels/sec = 21.6GB/sec bus traffic
 
That would definitely be interesting to know more about... I guess my other question is then : how are you supposed to go about computing accurate incident radiance to all these pixels? And how are high frequency lighting effects handled?

Basically... why this approach versus say photon mapping?

Regards,
Serge
 
I seriously doubt that a CPU can do what a GPU does along with it's other duties. try this at home


Load up moto racer under windows 2K or XP and choose the software rendering option and try to play the game... it is next to impossible with anything less than the fastest CPUs available. I have tried it with my AMD rig running at 1.8G and the game play is so bad..........
 
PatrickL,

Thanks for the link. Rather interesting if I may say so. It would be really cool if it was possible to get more accuate numbers outside press releases. However, it is not :(

Still - rather interesting to see how it looks.

Couple things to note - there are a lot of people willing to shell out a monthly fee for:
A. Dedicated servers
B. No Cheating
C. Continued Content updates

Amazing.
 
Online games are addictives. Since i first played to UO (7 or 8 years ago) i just can t stand offline games, i am bored very fast.

Playing in guilds, building a communauty, adds so much to a game that people enjoy paing for that. In fact one server on Everquest had a special $ 40 monthly fee and was full :)

I heard in taiwan a game had 7 % off the population playing it and so on.

5 december in europe and 9 december in US, Horizons was launched and, despite being a half finished game, servers are crowded due to players looking desesperatly for novelty.

And you know what, noone ever "optimize " for them, noone benchmlark them :(
 
Back
Top