Talk:Vector graphics

Scaling Illustration

Added an illustration showing the effect of scaling on vector and bitmap images; this illustration demonstrates a key advantage of vector-based graphics (the ability to scale them to any size without degradation).

Agateller 06:21, 13 January 2006 (UTC)[reply]

General

Please see my comments under the "raster graphics" page.

Presenting "raster graphics" and "vector graphics" as two alternative ways of doing 2D computer graphics made sense in the '70s, but that view is no longer valid or useful today. They are not alternative ways but succesive stages of 2D computer graphics.

What this page calls "vector graphics" is better called (2D) "geometric modeling", and what it means by "raster graphics" are nowadays called "digital images" (as mathematical objects) or "frame buffers" (as concrete data structures). Rendering a geometric model produces a digital image, and virtually all output devices today can only print/display digital images. (In fact, I believe that even large format plotters have been largely replaced by raster-based devices -- essentially wide-bodied printers, using inkjet, electrostatic, or laser technology).

Changing the name "vector graphics" to "geometric modeling" would also allow us to unify the 2D and 3D geometric modeling pages. After all, the basic idea is the same, and many primitives (polygons, splines, etc.) are the same.

Moreover, it seems that many traditional applications of 2D graphics, such as commercial artwork, are being taken over by 3D graphics -- which provide effects like shadows and shading. At the very least, the boundary between 2D and 3D graphics seems to be getting blurred.

Rendering is still often called "rasterizing" , both because of the original vector/raster nomenclature, and because the first algorithms did it in real time, by sweeping the "vector" representation with a horizontal line in sync with the CRT electron bean path ("raster"). That solution was imposed by the lack of a frame buffer. Presently, however, scanline-sweep is only one among many methods for converting the geometric description into an image. For instance, the popular Z-buffer method generates the pixels in random order. Thus "rasterizing", while still in wide use, is a rather old-fashioned word.

Recently I wrote Portable_Document_Format#Layers. I wonder how this would translate into your terminology.--Patrick 12:45, 5 Mar 2004 (UTC)

Well, what about this:

A PDF file often combines geometric primitives, text, and digital images. E.g., the general reference map of the US [1] uses:

polylines or splines for coastlines, administrative boundaries, rivers, highways;
polygons or circles for lakes, cities, and Interstate highway symbols;
text for labels;
digital images for mountain relief.

In the Postscript/PDF imaging model, these graphic primitives are sequentially painted on a "virtual canvas", in the order in which they occur in the file; so one can have arbitrarily many layers of text, geometric shapes, and images, interleaved in any order. Images can also be clipped to a boundary of arbitrary shape.

In zooming in, the outlines of geometric primitives and text remain sharp and smooth; whereas image data breaks up into separate pixels, rendered as uniformly colored rectangles. An example of a PDF map done entirely with geometric primitives is [2]. In map [3], the colors are provided by a single background image, while coastlines are represented as polylines; the mismatch between the two layers is noticeable when the map is highly zoomed in.

The PDF page still has plenty of room for improvement. The Postscript language and its imaging model certainly deserve a separate page and a better description, and part of this information belongs there. Also, for instance, images are not merely plopped down on to the canvas, but can be dithered according to a "dithering screen" -- which may be either the printer's default or a client-provided one. The Postscript model also allowed one to treat each pixel as a geometric object, and (e.g.) paint each pixel as a tiny hexagon rathr than a square. I don't know how much of that freedom is retained in the PDF model. I am not sure, but I bellieve that Postscript also could perform smooth interpolation of a given image; in that case your remark about zoomingin, above, would have to be qualified. Jorge Stolfi 21:38, 5 Mar 2004 (UTC)

image!

Someone make a vector equivalent of this:

Show that it still holds the same shape at larger sizes, and somehow demonstrate the presence of discrete objects instead of pixels. - Omegatron 00:39, Apr 5, 2005 (UTC)

Why do you want this done? --InvaderJim42 00:58, 12 August 2005 (UTC)[reply]

path

How the "brain" and "eye" work?

The second paragraph relates vector/bitmap graphics to the way a human "brain" or "eye" works. Is there any source for this information? It certainly seems intriguing but I wonder exactly how much solid research it's based on. Especially when it refers to ""recent studies," which I don't see any link to.--Eraboin 03:52, 4 February 2006 (UTC)[reply]