Exponential Growth vs Diminishing Returns: What the State of 3D Gaming Has to Say About Technological Progress

I am old enough to remember a time when there was a differentiation between a video card and a 3D graphics card – that’s right you whipper snappers, there was a time when specialized hardware had to be purchased to either feed information to a monitor that looked something like this OR perform the calculations necessary for 3D gaming that looked something like this.

We have come a long way since those early days of 3D gaming, and it serves as a perfect case study of the accelerating technological improvements that we futurists are so fascinated with. Note that I am not making any arguments in terms of whether or how much games have improved in overall quality, I am focusing strictly on the quality of 3D rendering. Here we can use the evolution of Playstation game console I will give this quick example of kind of gaming evolution that we went through:

(1) Playstation 1 – released 1994 – The dualshock controller marked a breakthrough for 3D gaming, allowing the ability to simultaneously control movement of a character and the where that character is looking. The graphical advancements over the course of the PS1 lifetime were consistent but modest.

File:Twisted Metal 2- World Tour screenshot.jpg

 Twisted Metal 2 – released 1994

Gran Turismo 2 – released 1999

(2) Playstation 2 released 2000 – The release of the PS2 was the beginning of what I think of as the age of accelerated gains in graphical wow factor. The release games with the PS2 were really only an iteration above the best games on the PS1, but this changed greatly over the lifetime of the system. The jump to Grand Theft Auto III level graphics was nothing less than a revelation at the time, it was like nothing we had ever seen before.

Midnight Club Racing – released 2000

GTA III Vice City – Released 2002

Gran Turismo 4 – Released 2004

(3) Playstation 3released in US 2007 – By the end of the era of the PS2 games were actually becoming pretty hardware limited (although games continued to be released for PS2 for many years after the release of PS3). The release of the playstation 3 marked a significant graphical advance. Over the course of the system’s lifetime (which is not yet even over), there were still consistent gains in graphical quality but nowhere near what we saw on the PS2.

 Screenshot

Need for Speed Carbon – released 2006

GTA IV – Released 2008

Gran Turismo 5 – Released 2010

GTA V – Released 2013

(4) Playstation 4 – Released 2013 – Just recently released, it is yet to be seen how far the PS4 will push graphical advancements. Launch games do look quite nice as shown below.

Need for Speed Rivals – Released 2013

So what is the point in all of this? Well first I just wanted to illustrate just how impressively far we have come in just 20 years. Since 1994, we have gone from a basic 3D with buildings which are rendered as simple boxes with some crude textures on them, to the much more realistic landscapes of modern games such as GTA V or NFS Rivals. 3D gaming serves as a perfect example of the kinds of breathtaking changes we have and will continue to see across the tech world.

Still, there is something more here. Notice the rapid improvements that were realized between 1999 and 2006. Now compare that to the difference between 2006 and 2013. Despite being a longer time, there just does not seem to be the same kind of leaps and bounds advancement we saw earlier. Yes, the games have undoubtedly improved, with more details, more believeable character modeling, better physics, but overall the improvements have not been a large as they were earlier.

By way of another example. Here are three FPS games which really pushed the graphical boundaries with their respective releases as they came out at about equal intervals.

Halo: Combat Evolved – Released 2001

Crysis – released 2007

Crysis 3 – released 2013

There is approximately the same time gap between the release of Halo and Crysis, and between Crysis and Crysis 3. Certainly there is an obvious improvement across all 3 screenshots, but in my opinion the difference between Halo and Crysis is far more apparent than the difference between Crysis and Crysis 3.

So it would seem that 3D games, while still consistently showing improvement year over year, subjectively they seem to be improving at a slower rate than they were previously. If I were to plot a graph based completely on my subjective asessment of the quality of 3D rendering of games it would likely look something like this.

Quality of Games

The shape of the curve above is what is known as a sigmoidal curve (as in shaped sort of like an S). Often times what appears to be an exponential relationship is actually the beginning of a sigmoidal curve.

In the case of games, of course they continue to improve today but the overall rate of change between successive releases has begun to slow. While I don’t predict video games are going to stop improving altogether, I simply think we have entered an age of diminishing returns in terms of the quality of 3D. This means that as we put in about the same amount of effort in improving video hardware each year, we don’t get quite as much subjective improvement in the overall visual result.

The reason that the quality of 3D games is not improving at the startling rates that it once did is likely related to a number of factors, but I think one of the simplest explanations is the diminishing visual improvement from logarithmic increases in the number of triangles used to render a particular scene.

Diminishing returns on logarithmic increases in complexity

In the early days, when games when from thousands to millions of triangles to render a particular scene, the difference was astounding. We ended up with something on the order of the improvement we saw between Twisted Metal and GTA III, and it was breathtaking (bordering on unbelievable) to see at the time. Games just kept getting better and better. Surely this would go on forever.

But if you look at the difference between 6000 and 60000 triangles in the image above you can see what the problem is. Even if you go through a completely new hardware iteration and manage to improve the rendering by a factor of 10, subjectively the experience does not improve that much. At this point you have reached an era where even exponential growth in the power of hardware starts to result in diminishing returns.

There have been a number of responses to the argument that increasing numbers of triangles for rendering really should provide more details in models (as here) but I think the fact still seems to stand, we are past the inflection point in the ability to render convincing 3D scenes.

Will we see major improvements in video games in the future? Yes, without a doubt. Will the improvement in raw 3D rendering ever be as astounding as the kind of jump between 2000 and 2010? Not likely.

So what can this tell us about technological change more generally? Do we see this kind of sigmoidal curve in other technologies? Absolutely.

I would say that the improvement of operating perspectives from the perspective of usability and productivity would be another good case. DOS was a wicked operating system, compared to it’s predecessors. Similarly first release of Windows was an absolute revelation in terms of computer usability. Successive generations of windows got better and better, with 95 and eventually XP offering exponential improvements. But really, since XP, changes have been more either in the background, or iterative tweaks to the general user interface. The advance of operating system usability is well into the age of diminishing returns.

How about something more hardware related? Well if you look at the history of digital storage (ie the SD card) it seems to have a similar trend. Using the Way Back Machine, I noticed that the maximum size of an SD card went from 1gb in 2004 to 8gb in 2007, by 2010 it was 64 gb, and now in 2013 it still is at 64gb. The 128gb SD card seems to be on the near horizon, but it has taken at least 3 years to double once. It would seem that flash memory has entered the age of diminishing returns, and in its current form it is unlikely to ever reach 1TB capacity.

In the science lab we see sogmoidal curves all the time, in fact it is rare that we see any natural system where there would be a exponential relationship without some sort of sigmoidal nature. The interaction of antibodies with targets, the reaction of an enzyme with its target, the growth of bacteria – all are exponential phenomenons that eventually hit physical limits and enter a plateau phase.

Those exponential relationships so adored by futurists are always just the beginning of sigmoidal ones. 

Diminishing returns are a fact of life, and we futurists are just going to have to deal with it. The advances of graphical wow factor are slowing down, improvements in SD capacity are diminishing, and eventually even Moore’s law will hit a fundamental barrier and we will see that what was exponential will become sigmoidal. Will new technology replace with new exponential curves eventually supplant the technologies driving Moore’s law? I hope so, but they too will eventually find their own limits.

Everything has limits.

So is this all doom and gloom then? Will we never get to the exciting futures that we like to envision? I am still confident that we will because we have yet to hit one very important inflection point. By my subjective observations the returns that we have achieved from scientific investment continue to pay accelerating returns.

We have yet to envision a limit to the exponential power of human ingenuity and I hope it stays that way. 

—————————-

Hey fans of TI, I am looking for contributors and guest bloggers. If you have an interest in futurism and you would like to try your hand at writing for a growing blog, send me an email at thought.infected@gmail.com. If you would like to support TI in other ways check out the new support page (go bitcoin!).

Advertisements

11 thoughts on “Exponential Growth vs Diminishing Returns: What the State of 3D Gaming Has to Say About Technological Progress

  1. The rate of increase in graphical resolution is decreasing in video games because game makers have changed their focus to AI, environmental reactivity, and story. Probably because of the diminishing return you mentioned.

    But the cutting edge of graphics has moved to Hollywood CGI. Directors make great use of closeups, which makes rendering beads of sweat or muscles under skin worthwhile.

    • Oddly, not listed on the Sandisk site. I am only basing my analysis on what was on the SD site for those years (although larger ones were surely available otherwise)

  2. While all you have written is true, the coming VR revolution will blow everything out of the water. In a few years, eye tracking and foveated rendering will allow path tracing to become viable in headmounted displays. I expect near-photorealism by 2020.

    • Yeah the ray-tracing stuff is totally a breakthrough. Here’s hoping that we will see it implemented in games in the near future. If things look anything like what the Brigade engine test shows I would totally admit that that would represent a huge leap in terms of quality.

  3. It was hard to find your posts in google search results.
    I found it on 21 place, you have to build a lot of quality backlinks , it will help you to get more visitors.
    I know how to help you, just search in google – k2 seo
    tips

  4. you’re truly a good webmaster. The web site loading velocity is amazing.
    It kind of feels that you’re doing any distinctive trick.
    Furthermore, The contents are masterwork. you’ve performed a fantastic activity in this subject!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s