n-Space Blog

Nintendo 3DS observations…

Posted in Analysis by Dan O'Leary on July 26, 2010

It’s been a little over a month since E3 2010 and Nintendo’s announcement of their successor to the DS we’ve come to know and love.  The 3DS was an “instant” hit, that, like most overnight success stories, was years in the making for Nintendo.

As long time Nintendo developers with over 20 DS, Wii and Gamecube titles under our belt, it occurs to me that we should take a minute to reflect on this new handheld.  This and subsequent reports on the 3DS are based entirely on our hands-on experience with the demos at E3, along with information we’ve gathered from publicly available online reports.

Stereoscopic 3D

Magical is the word that has most often been used in describing the 3DS’ signature glasses-free stereoscopic 3D display.  I agree.  It’s something your brain struggles to reconcile – the sense of visible depth created on a flat screen – but only briefly, before giving in to the pure child-like joy that follows.  For me it hearkens back to the Viewmaster or storybooks with scratchy lenticular cover graphics I had as a child, only animated and interactive.  It is effortless and amazing.

I won’t dwell on this now as it’s been well covered elsewhere.  Suffice it to say that seeing is believing.  If 13 of 15 of IGN’s editorial staff agree on anything, it must be pretty damn cool.  More importantly, even without this impressive, attention grabbing feature, the 3DS is a hot piece of kit.

What’s in a Name?

After a long line of incremental DS-branded handhelds, from the original DS to DS Lite, DSi and on to the biggest and best DSi XL, it would be easy to assume that 3DS is just more of the same with a stereoscopic 3D display.  Nothing could be farther from the truth.  In reality, a review of the published specifications forces us to conclude that the 3DS is more like a “mini 3D Wii plus” than any DS.  Allow me to break that down:

  • mini – well, it fits in your pocket
  • 3D – stereoscopic 3D display
  • Wii – graphical performance, motion sensor and gyroscope input devices
  • plus – programmable shaders, rather than a fixed function pipeline, allowing features like per-pixel lighting, procedural textures, refraction mapping, subdivision primitives and gaseous object rendering

But that’s only half the story.  Let’s put a few of these numbers in perspective…

Many Tiny Pixels

In a world of 1080p HD and cell-powered processors, the idea of a Wii-powered handheld with a 400 x 240 resolution (800 x 240 interleaved per eye) may not strike many as next-gen, but resolution alone isn’t sufficient to evaluate the clarity and detail that a given display is capable of producing.  Let’s compare the pixel density of various game systems with the 3DS:

  • State of the art next gen console in a typical living room
    • 1080p signal displayed on a 36″ display
    • 1920 x 1080 pixels = 2202 pixels on the diagonal
    • 2202 pixels / 36″ = 61 pixels per inch
  • Wii on the same display
    • 640 x 480 pixels = 800 pixels on the diagonal
    • 800 pixels / 36″ = 22 pixels per inch
  • 3DS
    • 400 x 240 on a 3.5″ screen
    • 466 pixels / 3.5″ = 133 pixels per inch

When it comes to pixel density, the 3DS is king!  To get similar pixel density from a 1080p source you’d have to play on a 16″ screen.  So there are lots of tiny (7 thousandths of an inch) pixels, which allow for very crisp detail and smooth gradients.  But the DS does well in this analysis too, with about 100 pixels per inch on the diagonal.  Yet nobody considers it a graphical powerhouse… what gives?

Polygon Parity

The Nintendo DS can display only 2048 polygons per frame, with a display resolution of 256 x 192.  Under perfect conditions the average polygon would consist of 24 pixels; far from the ideal of 1 polygon per pixel, even in the best conditions.  In real-world scenes with overdraw, etc. it gets much worse.  Contrast that with the 3DS, whose DMP PICA200 GPU is spec’ed (PDF) at 15.3 million triangles per second (or twice that in some reports).  Even dividing that by a factor of 8 to allow for a combination of real world conditions, dual screens and inflated specifications, it equates to over 60k triangles per frame on the top screen.  That’s roughly 1.5 pixels per triangle, not far from the theoretical 1:1 optimum where texture isn’t even required and simple face color gives you all you need to create a perfect rendition of the scene at the display resolution of the device.

So it seems that the 3DS has a lot of very tiny pixels and can draw more than enough polygons every frame at 30+fps (CPU and pixel fill-rate willing) to use those pixels very, very well.  When you add the PICA200 pixel shaders to the mix, it gets pretty exciting graphically, especially when you consider the battery life and likely pricing of the 3DS, not to mention all the other features this article takes for granted.

What do you think?  Based on your response we’ll share more of our thoughts on this and related topics.

%d bloggers like this: