Funny article on graphics "advancements"

We've been working with shadowmaps, and I have to say, it is the first real graphics advancement I have seen in several years. Until now, there hasn't been any lighting in real-time graphics, just a lot of fake effects. Some of the features on the GE Force 8800 really are worth using, and I am glad the manufacturers are finally moving forward.

I came across this article the other day, and thought it was funny how they tried to sell these simple post-processing effects as if they were the next big thing:
http://www.bit-tech.net/gaming/2005/12/09/source_film_effects/1


Your eyes already do that when you look at the monitor. That's not a holy grail.


A noise texture drawn over the screen adds complexity?

I remember when HDR was being pushed as the next step in graphics, and I kept thinking, we don't even have lighting that doesn't suck?!

I'm glad that phase is mostly over. The stuff Crytek is doing actually looks new and better than the same stuff that's been around for the last four years.

I have an 8800 GTX.

My shadow-mapping is second-to-none - I own. I also recently declared that HDR was pants.

I am the greatest. I lesson all.

They were the next big thing; the article's from 2005..

That's what I am saying.

2003-present has mostly been one long period of non-advancement. A few crappy techniques like stencil shadows came out, but now those are being abandoned. And at the same time, I have seen so much hype about how great graphics were during that time.

'Cause we all know, colours weren't so intense back then...

They're all fake effects - generated by your computer - on a flat projection screen. Any notion that any technique is "real" is absurb.

Hahah, priceless.

If you go back and look at a game from 2003 and compare it to a game from the present day it won't look pretty. Heck, even Resident Evil 4 which was highly praised when it was released as a graphical powerhouse has been relegated to the 'meh' graphics category of today. Therefore, I think advancements made just weren't your cup of tea perhaps but overall they created a more tangible looking experience just the same.

meh. DOD:Source is kickass imo, snipers and rifleman ftw!

Dynamic HDR is underappreciated and misused...

Today's common displays are incapable of giving much intensity at all with their brightest white, so intense colours have to be created via other means. Dynamic HDR is the way to do that!

Most games use it terribly, just as a pointless feature to sound cool, but if you want to see it done right, take a look at the reactor core scene in HL2 Episode 1. Without dynamic HDR, it looks like a regular old boring scene with lots of white and a floating blue ball.
With HDR turned on full, the intensity of that floating blue ball is really highlighted by how it contrasts with the environment.

That effect was previously impossible because simulating its intensity would require either the ball or the environment to be coloured to some strange extremity, which degrades the rest of the experience. (One of the two objects then looks terrible, and everything looks really weird unless looked at from the right angle).
With dynamic HDR, that same scene is possible because the ball's contrast with the environment is simulated in real time, thus neither the ball nor the wall are by default ugly and washed out. The effect happens when it needs to, creating the same experience that could theoretically be done statically, except preserving the quality of the scene as it changes.

The difference in graphics between 1999 and 2003 was much greater than the difference between 2003 and 2007, and the difference between 2007-2011 will be much more than 2003-2007.

Graphics have been in a slump where the hardware can't quite handle what we need it to, but now we are beginning to get some power to do basic things like lighting.

It's always easier coming up with more revolutionary looking stuff on a first sight when a technology is rather young. Working on details and getting things faster is very hard work to do and needed as otherwise you would be sitting on a cool concept which isn't useable in realworld projects.

Shadowmaps like Stencil Shadows have their own disadvantages. There is no golden bullet at the moment which does it right for all. Depending on your plans there might be a technique which is better suited to you but that's it.

What i will see as a new stage for graphics will be when fast realtime raytracing will be around in everyones computer system and not only on some dedicated hardware. Till then i guess we'll see nice improvements here and there but nothing will look really revolutionary because the way you program gfx cards just is so open and you can do so different stuff with it. And this mostly makes it a speed problem.

If there is enough power then you can do fantastic stuff. If you have a slow card then you can't. That's all about it. I don't see this as less revolution than before and i think it's wrong when i look at the sums they invest into research for gfx cards too.

It would be nice to see a shift away from polygons to solid objects, proper raytracing, realtime radiosity lighting which would automatically mean correct shadows, both hard and soft, and perhaps also more along the lines of fractal detail. But we've been waiting for this since the 80's.

When I say "fake effects" I mean a technique that doesn't work as you would expect it to under different conditions.

Shadow mapping is the first good lighting I have ever worked with. It has a few foibles, but I think this is the method that will dominate for the next five years, and it will only get better as hardware improves. So to me it's like the last five years can just be forgotten. At last, we have simple direct lighting that works! It's kind of sad that for all the hype, we really haven't even been able to do lighting, but at least it is making real progress, instead of "film grain". In ten years, people will look back at 2000-2007 and think "They couldn't even do lighting back then". I mean, taken in the broad scope of things, graphics are pretty pathetic right now, but getting better.

This is the first time I have really had 3D graphics with lighting!

Recently I came across this great comparison from Microsoft itself between DirectX 9 and DirectX 10:
http://www.gamesforwindows.com/en-US/AboutGFW/Pages/DirectX10.aspx

The first page there is pretty dumb.
They're comparing an already awful looking old game made for an old game console to a brand new, good looking game using DirectX 10 and specifically built for Windows PCs.
Everyone knows Halo 1 looks ugly compared to newer graphics (or PC graphics of its time), so how is that a viable comparison of how much better DX 10 is? That's just an example of how well good artwork works.

Yeah, that's what I meant with 'great'. :)

Basically, it looks like the best you could do with DirectX 9 is Halo 1. And they took one of the worst shots of it, nonetheless. While DirectX 10 is all Crysis, of course!

Haha, a comparison between a console game released in 2001 and a technical preview from a game that hasn't been released yet. Not even an in-game screenshot.

Knowing now that the quality of my 3D graphics would decrease so significantly with the next version of their graphics API, I think they've convinced me to stick with OpenGL (whose releases seem a lot more comfortable with the old ones).

This DX 10 hype is like a commercial for a cleaning product, except the "previous leading brand" is the exact same as the one being advertised.

I saw the full version of that article earlier, it had more pictures (including some lake landscape), which were all determined to be fake. If that was ever an original Microsoft article (at least this one and the one I saw were not), they are really desperate.

But Pickelsworth is right, OpenGL is the best graphics API, why would someone use an API which works only on Windows, since a big number of people would never use Windows. And Windows has really nothing which would give it the right to be the one-and-only operating system, although Microsoft seems to want us to believe so, since they make basically no effort to write software for Linux and Mac.

I just read the same article I think you are talking about:
http://windowsvistablog.com/blogs/windowsvista/articles/447226.aspx



Anyone who knows computer graphics can see that the second one is just a photoshopped version of the first:

They forgot the lens flare, I fixed it for them :



Anyhow, regarding graphical advances, is it me, or is it that anything worthwhile that does actually come up, it chokes most of the current video cards in their performance if combined with any other current effects?

Seems that a new round of vid cards has to be produced to accel in that new effect routine for it to actually be plausible to use in your modern-day game. Heck, I can't even put AntiAliasing on any of my more recent games cause it craps out on my card when combined with anything fancy.

They were saving that for DX11, of course.

Needs some film grain on that lens flare...

Here's your film grain:


Bumpmapping + HDR:

Now you're talking.

Good question ... perhaps because Linux will never get the drivers to see unified shader support at an acceptable speed until they understand two things? :-)
1. Intel does not, I repeat, NOT create 3D cards. Until recently they sent vertex shader stuff through bus to the CPU to emulate hell. So the fact that they offer opensource drivers actually means nothing to "3D support"
2. Until the dispute on binary drivers is fully resolved within the "rule maker teams", the chances that AMD or NV will feature a highly opted driver isn't that likely. Both offer some drivers but lets face the truth, the best of their unix drivers is actually not downloadable but fixed to some specific product series.

(ok there is still Mac which most likely will outrule Desktop Linux usage in the forseable future thanks to the KDE4 project for OSX as they will be able to have the best of both worlds which includes working 3d drivers actually.
Linux just tries to be too much like Windows and OSX without giving the needed stuff to compete with them on the productivity level *I've had to learn that once again when tryting to get Ubuntu 7.04 connecting to my D-LINK WPA2 wireless network which just won't work. XP SP2 does without any problems*)