Originally published at: https://boingboing.net/2019/01/11/why-video-games-are-made-of-tr.html
…
…smaller and smaller triangles
…until you reach the turtles.
Which, coincidentally, is a triangle.
Because triangles define a plane unambiguously.
A twisted quad of the same 4 points can be interpreted in two different ways depending on how you place the crease. Choosing a way and triangulating it disambiguates the surface.
Ask me!
Not only is the triangle the easiest instrument to play in music. It is also the easiest for a computer to project in three dimensions.
Back in my day, video games were made from SQUARES, not TRIANGLES. (Yes, they address this in the video.)
Also, I guess I shouldn’t be surprised, but I find it interesting that a Vox video has an ad for Skillshare embedded in it. I expect this from an independent blogger’s videos but not necessarily an online media org.
The new media are turning into the old media.
Good to see that Keiry Knightley is still finding work these days.
Easier than the fakeblock woodblock?
When they went to triangles everything started to get too complex. Back to squares, I say! (yells at pixellated cloud)
GameHut is a lovely channel on Youtube from a former developer at Traveler’s Tales. One such video discusses how the Sega Saturn used rectangles instead of triangles, and consequently caused no end of woes:
Thanks, I knew there was a 1-2 sentence summary that would save me from wasting my time watching video!
Not particularly relevant to the video explainer here, but as it starts with pixels, it’s always good to bear in mind that pixels have “subatomic” components in current hardware that can now be manipulated directly, or with cunning plans
As an explainer, it ended just as it was getting interesting.
The real question is: why polygons rather than voxels/point clouds?
We’ve had voxel-based games for quite a while, and they’re only now taking off.
I recollect it could be done indirectly, at least, with the Apple IIe. Depending on where a pixel was placed within a sprite, and relative to other colors, the actual color of an individual pixel could be manipulated somewhat (as well as having the illusion of third colors by placing two colors near each other). Orange could turn dark green, for instance.
I’m not that up on current game engine things, but I think one advantage of triangles is they define a larger area, you can get a face normal by the cross product of two sides, and they can be processed in long connected strips so that due to shared edges, each new triangle is determined by a single new point as you go down the strip, and game engines are set up to do that very fast.
Didn’t get far into this video before I needed to comment:
As Alvy Ray Smith says, A Pixel Is Not A Little Square PDF
But aren’t squares just two right-angle isosceles triangles stuck together?
In the low-poly era, it made sense, as a small number of polys could convincingly describe a large area, with low-resolution, repeating textures faking additional surface details. That worked really well for representing very square environments - corridors, rooms, nice square buildings, etc. There was a preference for those kinds of environments for various reasons - natural (i.e. unbounded) outdoor areas were problematic in terms of storage and rendering, and part of the development history of games involved basing everything on gridded environments. Nice square environments of any size were easier for developers to model and light, too.
But voxels combine texture and model information, so when they reached a certain level of complexity (and for objects of more organic appearance), polys stopped making sense. Also, in terms of data management, voxels aren’t so different.
At this point, though, the hardware rendered polygons really fast because they’ve been designed to render polygons very fast (because games were using polygons). Even voxel-based games/data visualizers are rendering their voxels as polygons. But it didn’t need to go that way.