• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Why don't 2D games use vector art?

Yea, that was my point. It's possible. I don't know how practical though.

How far are you willing to dance around and twist words in order not to admit your idea is simply unfeasible? It would be prohibitively expensive to develop and it would take a much more powerful machine to render on the fly, all for hypothetical 4k usage. Saying "I don't know how practical" repeatedly after people have repeatedly told you how and why it's not practical at all is extremely rude. It's just a childish way not to concede the point when you have no arguments.

I'm not ignoring it it's a solid point so I have nothing else to add or request clarification on. Short of just quoting it and saying ok there's nothing to say. So I don't say anything.

Well, glad we cleared that up. :D
 
Boy, you crazy!

Foxy__by_copperthistle.jpg

Baduist_Monk_by_copperthistle.jpg

thanks for proving my point bb <3
 

Shin-Ra

Junior Member
n++ is an actual game with a Vector graphics engine:
http://blog.us.playstation.com/2014/05/14/n-on-ps4-graphics/
This is my favourite blog post ever.

One simple way to combat this problem is to take more than one sample per pixel — multi-sample anti-aliasing (MSAA) takes a series of samples inside each pixel, and then averages the results to produce an anti-aliased image.

Unfortunately, even lowly 4x MSAA means drawing the entire scene four times per frame (quadrupling the amount of work)… and it’s nowhere near the quality level we were shooting for.

So anyway, we decided to go crazy and write our own custom vector renderer.

The shapes we want to draw are defined geometrically, and a pixel’s color is determined by how much of each piece of geometry overlaps it.

This is a huge number of calculations, since it has to be done once for every pixel for every layer (most things in N++ are made of 4-8 layers); this also means generating a lot of geometry — each frame of N++ is made of up to 1,703,936 triangles!

The result is immaculately smooth anti-aliasing, which runs at a flawless 60fps/1080p.
As far as we can tell, it’s equivalent to 256x MSAA!! (or more — Shawn insists that it’s equivalent to 4,294,967,296x MSAA, and I guess he should know since he programmed it, but Shawn is often prone to hyperbole, and so although we actually do agree that theoretically it may be 2^32x MSAA, for the sake of avoiding any possible false advertising, we’ll go with the lower bound. Either way, the point is: it looks amazing.)

Bespoke vector rendering was at the top of our list of priorities for N++, because smooth graphics make the smooth gameplay feel even better. So, we’re really happy this worked out. Thanks Shawn!

The process for getting the geometry into the game is quite involved: working from our designs (vector shapes made in Illustrator or Flash), Shawn defines each vertex as code.
That’s right — N++ has hand-crafted artisanal vector graphics.
 

bomblord1

Banned
How far are you willing to dance around and twist words in order not to admit your idea is simply unfeasible? It would be prohibitively expensive to develop and it would take a much more powerful machine to render on the fly, all for hypothetical 4k usage. Saying "I don't know how practical" repeatedly after people have repeatedly told you how and why it's not practical at all is extremely rude. It's just a childish way not to concede the point when you have no arguments.



Well, glad we cleared that up. :D

All I was saying in the original post you are referencing was it was theoretically possible to draw that scene but I wasn't sure of how probable it was to render it in a real world scenario. I tried to make it clear that I was not sure about the technicalities only that it could be done in a hypothetical sense. The how and why was addressed after that and I agreed. Or are we no longer talking about the Muramasa pic?

Regardless, It's obvious at this point it's not probable to do a complex scene with vector graphics in a real time game engine which I have conceded I'm not sure what more you want from me.
 
thanks for proving my point bb <3

Boy you wish you had Joey Zeor's talent. I know your ass is just playing, because this is a prime example of a man with beyond exceptional talent. Now my negro Joey is a master of blending colors and creating that subtle 3-D perspective. Find a brother or sister, who can exceed him in all accounts or perspective, color, and finesse.
 

Mandoric

Banned
Honestly, I think the reason is mainly aesthetic.

People have deeply formed associations with very, very specific dot art styles. A 16x32 (2-tile) character sprite with constant full coverage in that box is seen as completely different from a character who mostly occupies the middle 16x32 of a 32x48 (6-tile) box, and styles are so absolutely formalized that for things like FF5/6 iOS a 48x96 character drawn to the same proportions as that old 16x32, by the artist who popularized the style, will be called out for looking like a cheap knockoff because of just how prevalent hi-res copies of her style became in cheap PC games.

Meanwhile, the vector look is linked to, at best, a few Amiga titles, and at worst Newgrounds ca. 2004 or /f/ porn from 2008.
 
Please keep in mind that all a "Vector" style game is really just a flat (or 2D) 3D style game.

It uses a lot of the same math and techniques.

It's just created and rendered in pure 2D.
 
From another perspective, if you take all the textures out of a 3D game, and remove the Z axis rendering you would in effect have a Vector style game.
 

Atomski

Member
Vector is a pain in the ass to use. Especially for colors.
As someone who has worked with vector art a lot I have no idea what you're talking about.


Anyways I imagine the actual reason is vector art is made of layers and layers of geometric shapes. Rendering a pixel based animation would be much easier than calculating hundreds of small shapes and keeping them all organized and such.

So technically it would be much harder than pixel based art and it also would use much more power than a typical 2d game as well.

I love vector art thou and would love to see it in more games, just need some talented developers to do it justice.

Edit: I also wish more films would do vector.. it was amazing in Scanner Darkly
 
Boy you wish you had Joey Zeor's talent. I know your ass is just playing, because this is a prime example of a man with beyond exceptional talent. Now my negro Joey is a master of blending colors and creating that subtle 3-D perspective. Find a brother or sister, who can exceed him in all accounts or perspective, color, and finesse.

i just playing b

henri regnault
 
Keep in mind that Flash is mostly just a vector tool, so a lot of Flash games used Vector art.

Technically Vector Art is used in a ton of games, just mostly web based games. This is due to a host of factors such as it's less taxing than 3D but still infinitely scalable.

A few of the reasons it's not used in major AAA games is mostly market demand, tool availability, and has a certain "feel" that doesn't work in most art styles, especially more "realistic" ones.
 

bomblord1

Banned
Keep in mind that Flash is mostly just a vector tool, so a lot of Flash games used Vector art.

Technically Vector Art is used in a ton of games, just mostly web based games. This is due to a host of factors such as it's less taxing than 3D but still infinitely scalable.

A few of the reasons it's not used in major AAA games is mostly market demand, tool availability, and has a certain "feel" that doesn't work in most art styles, especially more "realistic" ones.

So is it or is not prohibitively taxing on the hardware to render vector graphics? I understand the lack of tools as that was addressed here in a post about it not working with a AAA workflow.
 

Orayn

Member
So is it or is not prohibitively taxing on the hardware to render vector graphics? I understand the lack of tools as that was addressed here in a post about it not working with a AAA workflow.

Like Raging Spaniard said, the main reason isn't hardware or lack of tools, just efficiency of workflow. Raster art is easier, faster, and gives the artists more control, while vectors wind up having relatively few major advantages in the high-profile games that do go with 2D visuals.
 

bomblord1

Banned
Cheap plug for Psynchrony, my cinematic rhythm action game in development...

BrushingFast.gif


1423471186874

That's pretty cool so thats some kind of vector rendering engine?

Like Raging Spaniard said, the main reason isn't hardware or lack of tools, just efficiency of workflow. Raster art is easier, faster, and gives the artists more control and doing it all as vectors has relatively few major advantages in the high-profile games that do go with 2D visuals.

That makes sense.
 

Palculator

Unconfirmed Member
Like Raging Spaniard said, the main reason isn't hardware or lack of tools, just efficiency of workflow. Raster art is easier, faster, and gives the artists more control, while vectors wind up having relatively few major advantages in the high-profile games that do go with 2D visuals.

Also, it's not like programs that are used to produce bitmap images only offer tools that are exclusive to raster graphics. Photoshop and Gimp both provide geometric shapes, brushes and that pen tool to draw a Bézier curve, among other things I'm likely forgetting right now. But using these tools artists get at least some of the benefit of vector graphics even though the result will be rasterized.
 
So is it or is not prohibitively taxing on the hardware to render vector graphics? I understand the lack of tools as that was addressed here in a post about it not working with a AAA workflow.

Not anymore taxing than a 3D would be (and probably less).

I could be mistaken, but as far as I know there is not a lot of artistic tools and programming libraries readily available for this in regards to games outside of Flash (performant with animation capabilities and a focus on games).

It really just comes primarily down to artistic style. Vector art has a certain look. Some like it, some don't.
 

bomblord1

Banned
Not anymore taxing than a 3D would be (and probably less).

I could be mistaken, but as far as I know there is not a lot of artistic tools and programming libraries readily available for this in regards to games outside of Flash (performant with animation capabilities and a focus on games).

It really just comes primarily down to artistic style. Vector art has a certain look. Some like it, some don't.

Alright, thank you.
 

dr_rus

Member
If I were to guess I would say that because pixel art is a specific form of art which is very different from vector art. Colors, ugly, etc. - that's not the point really but I do think that doing vector is more time consuming in general. Soooo it's a different type of art, it's more time consuming and until recently it was quite a bit more h/w resource intensive. Basically this is a legacy of old gaming platforms I'd say.
 

Palculator

Unconfirmed Member
Not anymore taxing than a 3D would be (and probably less).

I could be mistaken, but as far as I know there is not a lot of artistic tools and programming libraries readily available for this in regards to games outside of Flash (performant with animation capabilities and a focus on games).

It really just comes primarily down to artistic style. Vector art has a certain look. Some like it, some don't.

Where's this misconception coming from? How could you use polygons to create a circle with infinite accuracy like you can easily do in vector graphics? And graphics cards are tailored to polygonal and bitmap graphics so vector art would have to be rasterized by the CPU first and then buffered into video memory, or the vector data is first loaded into video memory and then a special shader or something like a CUDA program rasterizes it using the graphics card. But neither are really comparable to polygonal art because those are only capable of representing things with a finite amount of triangles.
 

Dunkley

Member
Much of JSRF's beautiful texture work could be vector/polygonised, as well as the HUD, effectively future-proofing it.



You'd need very robust LoD scaling for when you're standing one end of a large environment looking across hundreds of polygonised 'textures' though.

I don't see what you would need that for though if you are producing for a single platform with a locked resolution, it would just waste processing on something that can be prerendered for the game with absolutely zero quality difference.
 

Barryman

Member
Where's this misconception coming from? How could you use polygons to create a circle with infinite accuracy like you can easily do in vector graphics? And graphics cards are tailored to polygonal and bitmap graphics so vector art would have to be rasterized by the CPU first and then buffered into video memory, or the vector data is first loaded into video memory and then a special shader or something like a CUDA program rasterizes it using the graphics card. But neither are really comparable to polygonal art because those are only capable of representing things with a finite amount of triangles.

Yes and no.

Most 3D polygonal art is trying to approximate a smooth surface or a surface with otherwise infinite detail -- hence the craze over tessellation when geometry shaders were introduced into Direct X. So in this way, rendering a 3D model is no different than rendering 2D vector art -- you're using spatial points (or...vectors..) to display an approximation of a certain form.

Take your example of rendering a Vector circle with polygons. What if we were to render a sphere with polygons, too? Would the sphere be any better "represented" by its polygonal rendering than the polygonal circle, just because it happens to be a 3D model? So to me, I don't see much of a difference between rendering vector art with polygons and rendering 3D models.

With typical "vector art," there *is* the additional fact that there are mathematical formulas that specify *exactly* where the added detail would go, if you were to add it. With a hand-crafted 3D model, this becomes a little trickier. But unless you're going for an intentional flat-shaded, low-poly look, your 3D models will always have the same errors in representation as renderings of vector art.
 

Shin-Ra

Junior Member
I don't see what you would need that for though if you are producing for a single platform with a locked resolution, it would just waste processing on something that can be prerendered for the game with absolutely zero quality difference.
There's never a guarantee a future version of the game will be rasterised from vector source assets so you want a PC/console-likely-to-be-emulated-in-the-future version with fully scaleable (or as scaleable as possible) graphics.
 
Where's this misconception coming from? How could you use polygons to create a circle with infinite accuracy like you can easily do in vector graphics? And graphics cards are tailored to polygonal and bitmap graphics so vector art would have to be rasterized by the CPU first and then buffered into video memory, or the vector data is first loaded into video memory and then a special shader or something like a CUDA program rasterizes it using the graphics card. But neither are really comparable to polygonal art because those are only capable of representing things with a finite amount of triangles.

It's not a misconception.

It's true that GPUs are tailed to polygonal and bitmap graphics, but polygonal graphics use most of the same math since polygons really are just vectors.

Also Vector art is solidly in the 2D realm which takes out a whole dimension and extra set of math which makes calculations simpler.
 

Palculator

Unconfirmed Member
Yes and no.

Most 3D polygonal art is trying to approximate a smooth surface or a surface with otherwise infinite detail -- hence the craze over tessellation when geometry shaders were introduced into Direct X. So in this way, rendering a 3D model is no different than rendering 2D vector art -- you're using spatial points (or...vectors..) to display an approximation of a certain form.

With typical "vector art," there is the additional fact that there are mathematical formulas that specify *exactly* where the added detail would go, if you were to add it. With a hand-crafted 3D model, this becomes a little trickier.

I'm not that familiar with tesselation. Do they allow infinite precision? Like, could you define a plane warped like a Bézier curve and have the tesselation approximate its shape at any zoom level?

Also, I was just confused by this notion that vector art would somehow be less taxing when the majority of video hardware is meant for regular 3D polygonal rendering based on approximations of shapes, whereas any vector graphic describe a perfect representation of whatever they're supposed to be: A circle is a collection of triangles arranged around a center (the amount of which might be increased dynamically due to tesselation) while a circle in vector art is a circle.
 
If it's not clear by now, let me point out that polygonal art and vector are are the same thing.

Polygonal art and graphics are simply a SUBSET of vector art and graphics.

The only real difference in conventional terms is that Polygonal art is usually in reference to 3D rendering, whereas Vector art is usually in reference to 2D rendering.

Generally when people are talking Vector art they are referencing 2D (such as works created with Illustrator), and when they are talking Polygonal art they are referencing 3D (such as works created with Max or Maya).

The funny thing is this convention is a socially constructed one. You could do full 3D Vector art (which would be very expensive to render). Polygons are just vectors and where chosen for 3D for the fact that they are the simplest and most basic vector shape.

My comments are in regards to 2D Vector art which I assumed was the focus of the discussion.
 

Palculator

Unconfirmed Member
It's not a misconception.

It's true that GPUs are tailed to polygonal and bitmap graphics, but polygonal graphics use most of the same math since polygons really are just vectors.

Also Vector art is solidly in the 2D realm which takes out a whole dimension and extra set of math which makes calculations simpler.

But polygons are only a subset of vector graphics. I know that literally being a collection of points (which are a collection of scalar values) make them vectors and thus vector graphics. But polygons, like pixels, only approximate the shape the vector art itself describes into a more hardware friendly format. Like I said, you can't have the detail of a vector circle (namely, the point cloud described by x^2 + y^2 -1 = 0) with pixels or polygons. As was pointed out, tesselation does help here, but tesselation only switches approximations on the fly, improving the approximation, but still losing precision.

If it's not clear by now, let me point out that polygonal art and vector are are the same thing.

Polygonal art and graphics are simply a SUBSET of vector art and graphics. (...)

? These are contradictory statements.
 
I'm not that familiar with tesselation. Do they allow infinite precision? Like, could you define a plane warped like a Bézier curve and have the tesselation approximate its shape at any zoom level?

Also, I was just confused by this notion that vector art would somehow be less taxing when the majority of video hardware is meant for regular 3D polygonal rendering based on approximations of shapes, whereas any vector graphic describe a perfect representation of whatever they're supposed to be: A circle is a collection of triangles arranged around a center (the amount of which might be increased dynamically due to tesselation) while a circle in vector art is a circle.

All tessellation does is add more polygons by dividing up the existing polygons to better represent the shape that is being attempted. In affect you are throwing more of them at the problem. So realistically no, it would not allow for infinite precision (at least with current hardware though in theory you could get it up to it).

All a polygon is is 3 points in space (2D or 3D) which are 3 vectors. You could warp them via a Bézier curve if you wanted to (2D or 3D).
 

bomblord1

Banned
But polygons are only a subset of vector graphics. I know that literally being a collection of points (which are a collection of scalar values) make them vectors and thus vector graphics. But polygons, like pixels, only approximate the shape the vector art itself describes into a more hardware friendly format. Like I said, you can't have the detail of a vector circle (namely, the point cloud described by x^2 + y^2 -1 = 0) with pixels or polygons. As was pointed out, tesselation does help here, but tesselation only switches approximations on the fly, improving the approximation, but still losing precision.

Excuse me for jumping in here but I think what he is saying is the processors used to draw a 3d polygon should be able to draw a 2d vector graphic since the math is the same just without a 3rd dimension involved.
 

Palculator

Unconfirmed Member
Excuse me for jumping in here but I think what he is saying is the processors used to draw a 3d polygon should be able to draw a 2d vector graphic since the math is the same just without a 3rd dimension involved.

But the math isn't the same. Polygons can only describe a subset of shapes that vector graphics can by approximating them. They simply are not as expressive.
 

pixlexic

Banned
Vector art with spline paths is more taxing than triangle polygons.

Let's not forget that most vector art have high use of bezier curves Not straight lines.
 
Disappointed N++ hadn't been mentioned, then clicked over to next page and bam! Right at the top. I've personally looked into vector-based art and animation years ago for a potential 2D fighting engine some of my friends at... Well let's just say the end result of that game was something else altogether. I must bow down to Metanet Software for their efforts in this regard!
 

Palculator

Unconfirmed Member
None of those games use vectors in real time. It's all high resolution 2D images generated from vectors. Castle Crashers does not look much sharper at 1080p, for example.

Guacamelee does for the characters/enemies. They are *literally* flash files: http://steamcommunity.com/sharedfiles/filedetails/?id=166995993

vector is UGLY


Vector is stunning in motion, and at high res: http://zippy.gfycat.com/SplendidOnlyFairyfly.webm
 
But polygons are only a subset of vector graphics. I know that literally being a collection of points (which are a collection of scalar values) make them vectors and thus vector graphics. But polygons, like pixels, only approximate the shape the vector art itself describes into a more hardware friendly format. Like I said, you can't have the detail of a vector circle (namely, the point cloud described by x^2 + y^2 -1 = 0) with pixels or polygons. As was pointed out, tesselation does help here, but tesselation only switches approximations on the fly, improving the approximation, but still losing precision.



? These are contradictory statements.

I think you're forgetting that everything needs to be mapped to pixels. Polygons are perfect representations inside the computer but need to be mapped to your screen. Same with all other vectors. Same with all other art.

All I'm saying is with todays hardware a reasonable 2D vector game will generally not be more taxing then a reasonable 3D polygonal game.

Obviously if you don't optimize your game, or handle things properly, you won't get the results you want, but that should go without saying.
 
Top Bottom