Raster v. Vector graphics: Thoughts/facts?

[getter77]

As I've began to delve into the world of 2D game asset creation,  I've somehow wound up overly and unusually curious/ponder'ish in terms of the above mentioned.

Near as I can tell from my surface readings:

-Raster is far more commonly used, owing to what I'd suppose are various factors besides monitor technology itself being a bit biased in favor of(?)
-There is quite a gulf in terms of the sea of development software between the 2 camps, again with Raster tools coming away with the bulk of the lot.  Vector side just about literally seems to be a pair of very expensive Adobe/Corel offerings, a much more affordable Xara suite, and Inkscape nipping at the respective heels one way or another down in the Land of the Free.  Or have I missed a ton of other highly functional/actively developed ones?
-Fundamentally different "constructions" in terms of what an image per se is to each.

So, Arcen folk and others, what are some schools of thought to consider in the grand scheme of things?   I would imagine something of a choice would have to be made early on in a project as to which to hew towards if not some sort of mixture?  Do the differences affect the likes of how they perform/can be utilized in terms of Unity and/or other general engines?   Do there look to be any signs of movement/hope within those W3C Vector standards they are working on kicking about such that the balance will even out more in the future?  How might their nature's situate with the "real" 3D of the (near) future?  Any dumb luck with taking a shine to one style easier than the other when starting out fresh?   

What is currently on my plate is solely in the Raster domain, and I surely intend to see it through Tablet/Pen in hand as these next days/several months roll on, but the developing future beyond that is as of yet unknown.  If there's signs of benefit on various ends to try to make time to cultivate both sides of this, I'd reckon some versatility would not be a bad idea...

[Diazo]

Not being a game developer, I have not done much with raster graphics, but I do deal with vector graphics at work all day. Well, vector drawings anyway as I am a draftsman.

And so, in my opinion, vector graphics are not something I'd want to see in a game. The entire point of vector graphics is that they are accurate regardless of what your current zoom level is relative to what the zoom level was when it was created. A game does not require that sort of accuracy at all, especially since the zoom level is fixed, or there's 3 zoom levels at most.

The overhead costs of vector graphics is not worth it for a game. Both in terms of creating the graphics and the engine power the game would require to display the graphics.

Really, if you want to "move up" from 2D raster graphics, move to a true 3D engine. It's probably actually simpler to do a 3D engine game as their is lots of support out there for them, where-as I don't think there's much, if any support for making a game in vector graphics.

As always, take my opinion for what it's worth to you.

Actually, I can't think of any game I have ever played that had vector graphics. They've all been raster (sprite) or 3D.

D.

[getter77]

Well, a brief Wiki tells me there was at least a slate of them Arcade and such wise back in the 80's.  Beyond that, I believe my beloved Triangle Wizard either uses it or some sort of homage to the style, as well as a substantial amount of Flash games/projects.

That there is seemingly such a lack of it, even with Flash considered, while modern hardware and engines have evolved tremendously even just within these last 5 years---somehow it just doesn't make sense to me.  I would think something marginally closer to a balanced pie in terms of ASCII vs 3D vs Raster vs Vector (Or am I forgetting a slice?)

There is a strange example of Zoom on the Xara site they use as a bullet point versus Raster in terms of size and resource constraints---image of a Microscope zoomed in until it is a lady in a bathtub.  Might you be able to...explain...that to me in any sort of practical terms as a layman?  It SEEMS like something that could offer 3D perception and depth tricks without actually gettting into 3D, which has been a useful thing for various 2d games over the years...

:Dives back into assembled his pile of 16 or so top free/modern Raster art related programs to study the gist of everything via their features/helpfiles:

[Fiskbit]

Well, when I hear 'vector vs raster', I definitely think late 70s and early 80s arcade games. My arcade collection is about half vector games, all of which are really fantastic games. Being able to have crisp, smooth diagonals on a vector monitor is a huge plus, and you could do cool things like leave the beam on something longer to make it brighter. If you've ever played Asteroids, you may have noticed how much brighter and more brilliant the shots are compared to everything else. There are also color vector monitors, though I don't really understand how they work. They look a lot more like raster monitors; the diagonals aren't as nice-looking as on black and white monitors.

There are games nowadays that try to recreate that vector style. Gravity Crash and Gravitron 2 (which "borrow" heavily from Gravitar) both have vectory graphics and look awesome. It would be cooler if they ran on real vector monitors, but so it goes.

[Diazo]

Might you be able to...explain...that to me in any sort of practical terms as a layman?  It SEEMS like something that could offer 3D perception and depth tricks without actually gettting into 3D, which has been a useful thing for various 2d games over the years...

Really, I think it's because of the extra calculations vector graphics have that raster graphics don't.

True, it's hardly an issue today with the pure horsepower modern computers have, but back a few years ago, that power wasn't there and so you went with the raster graphics because they were less processing load on the system. (I'm generalizing I know).

It's the basic format of the graphics that cause this.

Take a top down game like AI war where models only rotate around the one axis.

Take a very basic ship made out of 6 triangles. (I know AI war has more, but keeping it simple.)  

For raster, you have to draw separate images for each angle you want to show the ship at. but at a specific angle, it's load the image and display it, move on to the next item.

For vector graphics it's a lot more complicated. 6 triangles have 18 vertexes and 18 lines. Assume a bunch of triangles land on the same points so we have 9 vertexes and 9 lines on our 6 triangle ship model. The engine has to determine the angle the ship is at, to whatever precision math is used, determine where the 9 vertexes are on the co-ordinates of the screen, then draw the 9 lines between the vertexes, then fill in the triangles so the model is solid. Now you can move onto the next item.

So vector is much more horsepower intensive, but more accurate. In this example of a ship turning, raster can only display the angles that an artist has drawn during game development and will 'jump' between these angles. Vector would have perfectly smooth rotation, as the expense of a more complicated display program.

Again, note that I haven't actually programmed with either type of graphic. I do deal with graphics directly in my job so I know this, but it's your call which you want to use for what you are trying to do.

At work, we use vector because I can draw a room with equipment in it that is an exact representation of reality, down to the millimeter and I can zoom it however I want and it will stay accurate. Show the whole room, or zoom in on a junction box on the wall, the vector drawing will stay accurate.

Using raster, this drawing of the room is scaled to fit on a page. The junction box on the wall might be 10 pixels high at 600 dpi. Good luck showing any detail of the junction box with only 10 pixels to work with.

Take that FWIW.

Hmm, for curiosity's sake, I checked out AI war, looks like it is using raster graphics.

D.

[x4000]

One thing to bear in mind is hardware acceleration. On a modern computer, all it's gpu can hardware accelerate is drawing in 3d. So, for doing fancy effects in 2d, or drawing a ton on the screen like in any of Arcen's games, that's what you do: you draw 2d in 3d. You can do that with either dieect raw texture drawing, or via an orthographic camera, in directx or OpenGL. It's the technique that most (I'm tempted to say all) non-flash non-handheld 2d games use these days. And most iPhone games use that technique, too.

The hardware left 2d behind, but fortunately it cam still tag along.