Topic: Why is the console so small?

[KnowsNothing]

I'd like Oblivion to be on the Rev.  That's the only request I have.  It's the only game coming out for hte 360 that I'm interested in (although it's also coming out for the PC) and it would rock with the Remote.

Kinda random  MaryJane reminded me of it, because she pretty much perfectly described it.

[BigJim]

FWIW, I know someone at EA in art, several months ago he and a few others designed and lit an entire scene (in HD) for an original game they're working on. He was excited to be working on next-gen. I asked him how long it took them to do all that, and it wasn't any longer than it would take for a typical current gen platform, because they had much better tools at their disposal and better hardware to play it on.

Possibly anecdotal, but it's a little bit of evidence that HD in itself doesn't automatically blow up expenses. There are a lot of elements that go into costs. He still curses PS2 because they were constantly adjusting to accomodate the weaker hardware. Literally spent weeks screwing with it, and he wasn't getting home until way after midnight (this was at the time of the EA spouse fiasco). And products were getting bumped a month or two. If that's not a time consumer/cost spender, I don't know what is. So costs that may go into HD could also possibly be negated by less hardware hassle and better tools in general.

Or if you have crappy tools (cough, PS3) then your costs will go up no matter what.

[KDR_11k]

I didn't say things couldn't be detailed, couldn't be beautiful. I'm saying there is a line when it's entirely superfluous.

That line is NOT obvious. Let's take FPSes: When were graphics sufficient? Did Wolf3D need to use VGA or was Catacomb Abyss good enough in using EGA? Does the detail level of Quake suffice for all games? After all it can convey anything you'd possibly want. Quake 3 Arena already added superfluous polygons, even though by today's standards the game is blocky as hell. Unreal Tournament 2004? That's the level the current generation is at. But it uses thousands of polygons per character and gigantic textures. Doom 3? OMG, normalmaps! While the polygon numbers went back down again we suddently have much more and much larger textures. Unreal Engine 3.0? Polygon numbers that we haven't seen outside of fighting games combined with all kinds of textures including faked SSS to make skin appear like skin instead of plastic. Which one is the first past the line?

Other examples would be almost every game on the Gamecube. Did Wind Waker need to use more detailled characters than Ocarina? Why does Sunshine have these water shaders that severely impact the performance? Couldn't SSBM have used the graphics from SSB while merely updating the gameplay mechanics?

If there is such a line it's constantly past current gen and in front of next gen games because that's how people always complain.

You can stop your slippery slope arguments about ASCII characters and other nonsense.

Have you ever heard Nethack fans arguing about how superfluous graphics are and how the imagination is much greater than any graphics can be? Seriously, there are people who say Diablo is just a shallow Nethack clone that is only successful because of its graphics.

The reality is, you'll need to write the code which will tell the game which texture to load based on the output.

Modern graphics chips can choose the proper MIP levels themselves.

I guess that brings up an important question. How big are the initial "big textures" to begin with? If they're already big enough then it doesn't matter.

Depends on the company and artist, usually they keep the final resolution in mind (e.g. they won't add small details to a texture for a PS2 game since that'll look bad downsampled like that anyway). Some draw at the final res right away (if it has been decided upon, that is), others prefer to work larger.

But anyway, PC developers produce multiple 1024x1024 textures per character these days, for a comparison the GTA main characters had one 256x256 texture at max. Larger texture resolutions need more time but since many are using modified photos for levels and normalmaps for characters these days resolution isn't as much of an issue anymore.

Dual development of SD and HD also requires you to burn both versions of the game on the same disk taking up space.

Wrong. The only difference between HD and SD is a single instruction. You DO NOT NEED TO RECOMPILE YOUR GAME FOR SOMETHING THIS TRIVIAL. PC games support 20+ resolutions each and they do it with one binary.

I just hope Nintendo uses dual layer high density dvd's as they originally said. I really want to play a 200hr rpg.

Baldur's Gate 2 took up four CDs (as opposed to the five of Baldur's Gate 1) and provided 200 hours of gameplay with meaningful choices that affect the plot development.

I asked him how long it took them to do all that, and it wasn't any longer than it would take for a typical current gen platform, because they had much better tools at their disposal and better hardware to play it on.

Mark Rein (Epic Games) stated that they need 1.5x as much time and ressources for a next gen game as they do for a current gen one.

[Strell]

Where's the line for superfluous graphics.  God, I'm sure glad I haven't explained that about 5 times in this thread and you still don't get it.  Ok.  I'll do it once more.  Madden.  Cheerleaders on the sidelines, fully rendered and given AI.  Ok.  I'll let you fail to understand that again.  There's more examples but I don't need to churn them out.

I like how you completely missed the ASCII characters thing by bringing up one example of a game that, for the most part, was developed with graphical limitations in mind.

ALSO, I like how you bring up Nethack, since it so perfectly illustrates my opinion to begin with.

You tell me first that graphics may or may not be superfluous, something that I'm arguing against by saying gameplay is what matters, and then you bring up Nethack.  So good job destroying your entire argument.  

You can't argue that there's no line and then talk about Nethack.  That's like saying roster upgrades don't make a game, but then talking about how Madden is so amazingly well done year after year.

I gaurantee the differences in HD versus SD cause development times.  Gaurantee it.  If you honestly think it doesn't, you literally have no clue about programming.  None.  Increased resolution = more detail on the screen = takes more juice = computer has to know how to handle that increase versus SD = the rendering engine has to respond differently = must be programmed to do so.  There's no magical switch that says HD = true, and suddenly the engine just knows what to do.

I doubt it is a huge monolithic thing to program, but it most certainly deserves to be taken into consideration.  In fact, I'll ask my programmer friend, who is currently writing a 3D engine from scratch, what he has to say on the matter.  And yes, I know this last paragraph looks like bs and everyone and their mom has used this argument before, but I gain nothing from saying things I can't deliver on.



 

[KDR_11k]

Madden. Cheerleaders on the sidelines, fully rendered and given AI.

I thought you said immersion shouldn't be hurt in the process? These details are important if you want it to appear believable. Having an audience that is completely oblivious to the match hurts the immersion. Seeing that the audience is just a row of cardboard standups hurts the immersion. I still don't see what your arbitrary definition of superfluous is.

I gaurantee the differences in HD versus SD cause development times. Gaurantee it.

I guarantee that it's not more than 2%.

Increased resolution = more detail on the screen = takes more juice = computer has to know how to handle that increase versus SD = the rendering engine has to respond differently = must be programmed to do so.

Increased resolution DOES NOT mean more detail. Most of that detail would be added in SD as well since in 3d an object does not have a fixed screen size and you don't want things to look bad when you get close if you can avoid it. You will ALWAYS use all the ressources the machine offers, no matter what resolution your game is running at. Look at SSBM, those characters are very detailled despite not being larger than 100px most of the time. There is no difference between optimizing for SD and for HD, you will always make sure to have as much detail as possible without dripping the framerate. HD just changes a tiny bit of the testing environment.

I can't imagine a single scenario where HD would require optimization and SD doesn't.

[Ceric]

I know for a fact that the Cube only has 1 resolution.  1 and if memory serves me right it's bigger than the screen resolution.  Then you have interlaced or progressive.  I mean why we're going done this road Progressive is a hit too cause now your rendering double the data.  To simply keep compatibilty and to change the least fundamental hardware to drive down Nintendos costs in R&D it makes since for them to stick with having this 1 resolution only.  In fact be forcing it to be progressive they truelly give you only 1 pipe way to get your stuff out.  It's pretty easy to split a progressive image to interlaced.  Harder to do the other but still doable.  With the graphic chip upgrade I'm sure Nintendo include a hardware solution to upmixing to Progressive, probably just an extra screen buffer really.  

So why would you want only 1 resolution?  Graphics, for the most part, are represented with a matrix.  There has been a lot of study around this particular data structure.  Weird ways to optimize.  Having a fix resolution let you embed some of those optimization.  We all can agree that hardware will always beat software when doing the same task when designed at the same level.    So having 1 resolution will allow Nintendo to leverage this like they can in the handheld market.

Now just a sample.  I can't think of what this is named but back when I was taking Algorithms we were taught an alternate way to do matrix multiplication but it only worked on Square matrices and the had to be of a certian size (I beleive they had to be some form of 70x70 matrix, 140X140, 210X210, etc).  Dr. Kosa then proceeded to explain that this wouldn't give you neglible venefit if it was really small like two 5x5.  But it would greatly benefit you if you went not that much larger then that.  Now heres the kicker.  The difference was that you replaced a multiplication with 2 additions.  For a human that would have you no speed benefit whatsoever since multiplication for most people takes about as much time as addition.  On most computer architecture (Von Neumann(sp?)), I think Harvard it doesn't but accessing memory is also 1 cycle operation to, multiplication takes longer to do then addition because it requires storage and more operations.  It is much faster because of that underlying difference.  Computer Science is full of that stuff.

Also could you think about the virtual console?  Most HDTV have built into them a pretty good system to upconvert SDTV so why reinvent the wheel?  If Nintendo went HD then the best solution, at least in my mind, would be to black bar the virtual console games.  They already have to compensate for being progressive which will make some of the "hacks" to make effects on older games not work correctly.

At the end of the day HD would just add one more big developement item for Nintendo onto a laundry list that they already have pretty full.  Because we've already seen that if Nintendo doesn't support it then no one else is going to either, in the context of there own consoles (Disclaimer made).

The next console will have it.  In fact Nintendo may even release a Revolution later with basic support for it.  At this junction I think they are just really concentrating on working the kinks out of there new toys.  Much like when CD-Roms first came on the seen.  They were "over-engineered" old age is about the only thing that could kill an early CD-Rom drive when it was fighting against rewriteable media.  It will even read the newest formats that some new drives won't.  Same concept here.  Get the iffy stuff solid to win people over.  Then you can't break what's not broken to make it better.  I'm sure that as soon as the Rev is out and it's settled down some Nintendos R&D will be back to working on the next iteration with HD in mind.

[KDR_11k]

So why would you want only 1 resolution? Graphics, for the most part, are represented with a matrix. There has been a lot of study around this particular data structure. Weird ways to optimize. Having a fix resolution let you embed some of those optimization.

What matrix? Do you mean the transform matrices used by the 3d engine or just the pixel array that is the screen (not really a matrix since no matrix operations are used on it, only blits and blending)? The screen is as optimized as it can be, the difficulties come from the 3d rendering and most of that is resolution independent. The matrices in 3d engines are always 4x4 and the hardware is already optimized for that. They are 4x4 at 320x200, they are 4x4 at 1920x1080. And they are really not that big of a problem.

[Strell]

KDR, you just suggested a higher resolution doesn't put more detail on the screen.

At this point I am asking myself why I would even bother responding to something like that.

[trip1eX]

Well DVDs on my 480i  Toshiba TV look awesome (and much more lifelike and detailed than 360 games) so I don't think think 480i/p will be much of a hindrance in the good looking game department.  

[NinGurl69 *huggles]

Quote
Originally posted by: Strell
KDR, you just suggested a higher resolution doesn't put more detail on the screen.

At this point I am asking myself why I would even bother responding to something like that.

I agree with KDR here.  Imagine a simple, yellow 3D-rendered polygon, a triangle (triforce, lol), on a black background rendered at the wonderfully low-rez NES standard.  The pixels are horrendously huge, and the jaggies are big enough to climb up.

Now, maintain the same viewing/monitor size, and increase the resolution, rendering the triangle again.  The most obvious effect of an increased resolution is the perceivable flaws at pixel borders are reduced (jaggies), providing some "cleanliness."  Bumping up the resolution DIDN'T increase the "detail" -- it's still a simple yellow triangle.  What's being confused is higher resolutions increase the POTENTIAL for greater detail -- more pixels to play with -- BUT it's still up to the developer (the image source) to decide how to occupy those extra dots.

I can see how the Megapixel rating in Digital Cameras imply "more rez -> more detail," but that works simply because the image source provides an insanely high level of detail; that source being the photons in visible light.  Look at the "CG" art Nintendo provides for some its games, like Metroid Prime.  A lot of the art are basically high-rez renders of in-game models w/ in-game textures.  The jaggies are mostly gone and the lighting is a little cleaner, but the textures hardly express new/extra details.

[Strell]

Meh.  I'm going to talk and see what my friend says.