Archived

This topic is now archived and is closed to further replies.

SpittingTrashcan

Graphical sophistication is bounded

Recommended Posts

To put my thoughts simply: - Photorealistic graphics have been achieved in cinema - Gollum for instance. - If Moore''s Law holds, the technology which drives cinematic photorealism will be in our homes and capable of real time rendering in the foreseeable future. - Once graphical sophistication reaches the level where further refinement cannot be perceived by the human eye, "better graphics" in the sense of more realistic or powerful graphics engines will not be a selling point. - At this point, games will be appraised on other criteria - the creativity of their use of graphics hardware, the attention to detail, the use of graphics to convey the overall mood or aesthetic sense of the game, and of course gameplay. - Therefore, to create a game that will last through the ages, stop focusing on the technical sophistication of your graphics system and start focusing on aesthetics and gameplay. Summary: In ten years, Doom 3 will look like Tron - outdated technology supporting a lack of substance. Comment, counterargue, but please do NOT mistake this for a "gameplay vs. graphics" post. The issue is whether sophisticated graphics will continue to be a selling point. - STC --------------------------------------------------- -SpittingTrashcan You can''t have "civilization" without "civil".

Share this post


Link to post
Share on other sites
Moore''s law maybe for CPUs, but its definitly not for GPUs.

A GPU, like a CPU, can only render as much as in the bandwidth of its memory. Memory bandwidth isn''t cheap, so there will always be a finite amount of it. Juggling data to the point where we can fill, but not overfill, this finite amount will be a problem for the foreseeable future.

So even though we can create photo-realistic characters, we cannot have a infinite amount of them, so their placement will always be engine driven, and therefore, sophisticated engines will be important for the foreseable future...

Share this post


Link to post
Share on other sites
It''s not a trivial prospect to model something so accurately that it can be called photorealistic. To have a photorealistic Gollum requires modelling accurately down to the centimeter, and using displacement mapping to model down to the millimeter. If textures are to look realistic, they must not have a discernable pattern (I assert that Gollum still has a discernable pattern to his skin), which requires at bare minimum pseudorandom alterations (this may not be enough).

Animations can still become more realistic. Veins that stick out when a character flexes his muscles have not been realistically modeled as far as I know. Hand movement (see Final Fantasy) needs some work.

There''s another consideration in that computers use fundamentally different techniques for rendering, and there may be an upper limit in quality for it (probably not). Also, while computers may be able to render one Gollum realistically, I doubt I''ll be able to render an army.

Let''s say I being to realistically render a billion Gollum''s at photorealistic quality. What can I do with a billion monsters? To have them act independently in interesting ways is just as processor intensive. What will probably happen is that whatever obsession there is with rendering billions of polygons will transfer into making billions of different independent actors (Doom 47: Your overcrowded planetful of enemies chases you down from all four dimensions! Fortunately, the BFG 15 Billion has a mind of its own and it''s plowing a path to rescue you.). It''s not much of an improvement.

Or maybe it''ll transfer to physics (Freespace 39: Real string physics! Not only are the projectiles superluminous, they tear your opponent to pieces even while he''s in a local hyperspace bubble. Realistically models the ships and interstellar dust to the string level. No longer will collisions with muon clouds fail to produce neutrino effects.)

Share this post


Link to post
Share on other sites
Personally, what I wish to see are games that are designed from playability and fun-factor first. Come up with a great idea for a game (or even find ways to enhance old games) and work out how you will implement all of the game features you need and want. Use the extra CPU cycles for your graphics

Plus, extra benefits being that if you program it correctly then the graphics can easily be made to accommodate new hardware. Game logic, AI, etc however is set in stone once it is sold.

Look at all the MMORPGs out there. They can generally increase their graphics through infinity, however, try to change the fighting system or the quest system and you are talking about a totaly new game.

So, make the game first and the graphics later. It should be clear by now that flashy graphics can only take a game so far if the game doesn''t have something else to offer.

Now to remark more closely to your original question. I think there will come a point where graphics will be able to reproduce very lifelike results and I think it could happen within our lifetimes. However, one thing I also know. Many people play games to escape from reality not fight rats in it. So, why is it even all that important that we keep pushing the limits on graphics. There is so much more than just graphics that makes the game fun.

Webby

Share this post


Link to post
Share on other sites
Sheesh, it seems like all of my posts are multiposting for some reason. Feel free to delete the extras while I see if I can figure out what''s going on.

Webby

(Crossing fingers for single post this time)

Share this post


Link to post
Share on other sites
After we get ''perfect'' graphics on a 2d monitor, we still have the world of 3d displays to move to, and after that, we can move to nerual interfaces which will probably make everything we''ve done up to that point useles. Displays and graphics still have a long way to go.

----------
Almost typo-ified using Extrarius'' AUTOMATIC Typo Generator, but I decided to be nice =-)

Share this post


Link to post
Share on other sites
Graphics are more than just look, they are animation as well.

Even though Gollum seemed realistic, his movements had character to them.

In the Harry Potter movies, they intentionally make things look a little less realistic to keep it within bounds of a children's story.

Then we get movies like the Matrix, to take us to a new reality.

It will be interesting to see the new kinds of imagery that movies generate for us. Perhaps a new form of visual entertainment that has no human characters? Shapes that move about and tell a story that way?

On the note of graphical sophistication being bounded, there really isn't any point in making an image more real than what is real, is there? How can you make what's on a screen more real than what isn't on the screen? Somehow, I don't think that will stop anyone from trying.


[edited by - Waverider on March 4, 2003 12:26:40 PM]

Share this post


Link to post
Share on other sites
I expect that at some point developers will actually clammer to depart from realism. cel-shading is just the first example of this, but it demonstrates the underlying thought well.

For fantastic games, we don''t want real, we want fantastic!

Just a side note, Websitewill mentioned that games should be designed from premise to engine, also known as top-down design. I think that the underlying theme here would be that given a premise which requires photorealism, and realistic simulation. Can we provide it?

George D. Filiotis

I am a signature virus. Please add me to your signature so that I may multiply.

Share this post


Link to post
Share on other sites
quote:
Original post by Symphonic
[...]For fantastic games, we don''t want real , we want fantastic ![...]


I''d like a fantasy game more if the graphics were photorealistic and the physics was really close to realistic and you could interact with any object, but I would like going ''beyond'' realism and adding in things like magic and the ability to jump 30 feet if you have X stat up to Y etc. I want reality augmented with fantasy =-)

----------
Almost typo-ified using Extrarius'' AUTOMATIC Typo Generator, but I decided to be nice =-)

Share this post


Link to post
Share on other sites
To rebut some arguments:

- "The graphics pipeline has physical limits."
But as you mention, reasonably clever algorithms can and do get around these limits. My timeline is meant to encompass advances in software as well as hardware.

- "We still have further to go in photorealism."
Yes, but after a certain point it stops mattering. Once you can''t tell the difference with the human eye, there''s no need to get any better. 1600x1200 screens approach the limit of the human eye to resolve objects - there''s no use in creating subliminal pixels.

- "Certain aspects of reality still aren''t modeled."
That''s often more of an artistic rather than a technical problem. Hand movements in FF: Spirits Within, for instance, look odd because human hands are constantly in surprisingly quick motion, something the modelers failed to note on an artistic level. I agree there''s plenty of room left for increased artistry and more aesthetic algorithms - but at a certain level realism for its own sake becomes gratuitous, see above.

- "You can''t create a billion Gollums."
I wanted to address this one directly because it simply isn''t true. The army of Orcs at Helms Deep was created through computer algorithms, all ten thousand of them, moving in formation but asynchronously. Fooled you, didn''t they? On the real-time rendering end, the Serious Sam 2 engine supports hundreds of enemies on screen at once, at passable if not photorealistic quality.

- "Graphics encompasses more than technical sophistication."
My point exactly. To coin a phrase, the size of your chip doesn''t matter, it''s what you do with it. I''m a tremendous fan of non-photo-realistic rendering and other innovative graphics techniques. I also like games which use their graphics in a consistent and well-thought-out way to convey a mood or aesthetic sensibility, such as Devil May Cry, Jet Grind Radio, or Mario Sunshine.

Extrarius: A comprehensive physics model and photorealistic graphics do not have to go hand in hand. While I agree that a more comprehensive physics/world model would be a great boon to roleplaying games, I actually think that non-photo-realistic rendering and artistic modeling convey a fantastic and exciting world much better. See also: Fable, which hopes to redefine the term RPG with a stunningly comprehensive world model - and no visible stats!

Overall, I guess the point I''m trying to make is that at some point, better graphics alone will not sell a game. For example, there are many many FPSes currently available which sell chiefly based on their improved graphics. There will come a time when the graphics get about as good as they possibly can. I''m saying that time is not far away, and that we as developers should plan for it by placing more weight on features other than graphical sophistication.

Please do continue the discussion, I''m finding it very interesting!

-STC

---------------------------------------------------
-SpittingTrashcan

You can''t have "civilization" without "civil".

Share this post


Link to post
Share on other sites
I think we still have a long way to go, even though graphics are getting close to photorealistic. One big problem is the way objects and lighting etc are calculated. I don''t think we''ll have perfectly photorealistic rendering until we stop using the diffuse/specular/ambient lighting model hack and stop using polygons to represent objects (why use polygons anyways when you''re using so many that each one is less than a pixel big on average... might as well use voxels or something like that).

While to you, a cartoony (or other non-photorealistic rendering method) might look best for a fantasy setting, I would prefer photorealism for any type of game. If I''m fighting a dragon, I want it to look realistic(since fantasy dragons never existed), and I want the ground it stands on to look real. I want my sword to gleam just right in the setting sun and I want my armor to look real and blind people with its reflection when they try to shoot me from far away.

I do agree that some game styles are best done in 2D (or a 3D world with movement restricted along one axis with a fixed camera), but I think even those games would be better with realistic images. Just think how cool the old school classics would be with realistic images. Instead of a really bad looking space ship, you could see every window on it and watch as the light of the ''slow-light lasers'' zoom by and light up your hull etc.

----------
Almost typo-ified using Extrarius'' AUTOMATIC Typo Generator, but I decided to be nice =-)

Share this post


Link to post
Share on other sites
Even though you think that the photo realistic graphics will be capable there will be able to create graphics that is better by making the movements more realistic the textures and the way they are effected by the enviroment and just make them plain better.

Share this post


Link to post
Share on other sites
The day we can''t excell in graphics anymore will be the best day in videogame history. From that day forward game developers will actually have to do thier job and not only make games graphics but work on gameplay also this will be a time of genre breaking and creating something new. like back in the olden days of nintendo. When you dont put boundries on someone they dont push to get out.

OoMMMoO

Share this post


Link to post
Share on other sites
quote:
Original post by SpittingTrashcan
Overall, I guess the point I''m trying to make is that at some point, better graphics alone will not sell a game. For example, there are many many FPSes currently available which sell chiefly based on their improved graphics. There will come a time when the graphics get about as good as they possibly can. I''m saying that time is not far away, and that we as developers should plan for it by placing more weight on features other than graphical sophistication.

I think that on some level everyone connected with the video game industry is both aware and supportive of this observation, but commercial participants are burdened with the pressure for profitability (you can''t tell your shareholders, if you''re a publicly held firm, that your game didn''t sell because you eschewed mere graphical advancements in favor of experimenting with new models of emotional content and so forth). I''m stating the obvious here, but it''s sometimes necessary to reiterate things to keep it fresh on our minds (before people start bashing "all the game companies out there").

That said, I particularly find the efforts at formalizing game design by formulating a common, standard vocabulary, identifying patterns and so on very heartening. I also think that there is an on-going broadening of implementational assumptions, such as the decision to cel-shade Splinter Cell (using unconventional graphical technology within a genre to homogenize the gamer experience, which is A Good Thing&trade. I''ve read a few articles from Warren Specter and Will Wright, as well as perused the columns on game design in Game Developer Magazine and at GamaSutra, and am quite impressed by what I see. I think the public, including those of us who consider ourselves amateur or hobbyist game developers, are less aware of these efforts because they are still in their infancy and haven''t begun to affect the content we see on store shelves in a major way.

What am I saying? I think that this discussion is very fruitful, but what I would like to see is more dissemination of these formal tools and constructs in the amateur community to raise the level of discourse and design, as well as the development and adoption of more (low-cost) high-level tools to facilitate prototyping and exploratory/experimental design.

quote:
You can''t have "civilization" without "civil".

Nor "civil war."

Share this post


Link to post
Share on other sites
Graphics will always be a selling point to some extent, even when true photorealism is achieved, however, the selling point will mainly be artistic, not technological. In photorealistic terms this means that attractive or interesting looking "human" characters and interesting environments will sell the game, kind of in the same way specific actors can sell movies.

Although the wow factor that graphics generate will drop dramatically in the next 10 years and will probably not be an issue in the next 20 years, other things like the most realistic AI and most believable characters will take it''s place. Whether or not that''s a good thing or a bad thing is up to you. A lot of people here seem to want "simple" games like in the "good old days." Photorealistic graphics, animation and physics are not conducive to that.

I believe that once we get to that point either games will all be produced only by a few major studios with staffs similar to effects companies (because the art requirements will be ridiculous) or games will be made mostly out of liscensed material. That is, there will be companies dedicated to making parts of a rendering, parts of the physics, parts of the AI, specific characters, environments, etc. To build a game you would either liscense an engine, write your own, or mix and match, like things are nowadays. Once you have your enging you would also liscense a lot of the content. Characters, environments, etc. are all liscensed. The job of the game developer is to take all the liscened material and build a game out of it. This could either help or hurt the industry. Either we get a bunch of games that all look and play the same, or we get better gameplay because companies no longer have to focus on technology. My guess is both will happen.

Share this post


Link to post
Share on other sites