Talk:Jaggies

This article had some material that was completely wrong, equating JPEG compression artifacts with jaggies.

I have trimmed out two wrong paragraphs, corrected them, and put them at the end of JPEG, then added a line here to note that jaggies are not compression artifacts. This article could do with a couple of images to show

• an image
• the same image sub-sampled with a good a/a filter
• the image sub-sampled again, this time with no a/a filter, and therefore jaggies

-- The Anome 08:33 30 Jun 2003 (UTC)

I don't think it needs expansion, it's about the right length and contains all it needs apart from the suggested pictures. The Real Walrus 07:48, 17 June 2006 (UTC)

You know the Windows XP interface has NO anti-aliasing. Look at the "rounded" edges of your windows, or WMP, or any of the WMP skins. There's a reason why your My Computer icon has jaggies, especially the monitor part of the icon. --24.87.126.13 17:41, 30 August 2006 (UTC)

There is some wrong information regarding the PlayStation2. The PS2's video chip does support a form of anti-aliasing, it's just not too practical, and thus not used. It requires geometry to be rendered from back to front. One of the problems is the vector unit's limited memory, so sorting would have to be performed on the cpu. Another problem is that this hardware is just far more efficient rendering triangle strips, which rules out sorting. I also don't understand why vertex shaders need to be mentioned in the article. Btw. Vector unit does not equal vertex shader (although it can be programmed to do the same thing). The article also makes it sound like the Emotion Engine is the video chip of the PS2, but it's not, it is the cpu. Then the article goes on to say that current consoles "can render high quality graphics using separately custom graphics processing units (GPU]])." The PS2 also has a custom video chip. GPU is just the new name being used for video chips. Also Gamecube and Dreamcast are not current consoles. —Preceding unsigned comment added by Klehocz (talk • contribs) 10:42, 16 October 2008 (UTC)

If you were referring to my edits, I didn't write that statement, I just tidied it up for quality. Rilak (talk) 05:03, 18 October 2008 (UTC)

Incoherent

"GPU manufactures of consoles such as ATI and Nvidia do not have jaggies, indicating the slow conformity in the market to PCs in general. In the future software problems are most likely going to be eliminated for the needs of the consumer."

Why is this article so obscurely written and worded when it's covering basically the same material as the articles on antialiasing / aliasing? Is there a need for a seperate article on the subject? Shouldn't "jaggies" just redirect? 152.91.9.219 (talk) 05:09, 17 December 2008 (UTC)

Current?

I don't think the xbox, gamecube, and dreamcast are current. I mean really, the dream cast? C'mon guys. It should just be the wii, xbox 360, and ps3. No sixth generation consoles. Only seventh. —Preceding unsigned comment added by 69.120.117.12 (talk) 17:55, 20 January 2009 (UTC)

Attack Site

According to Firefox, the first reference is a known attack site. Is this still a reliable source? --Blacklemon67 (talk) 21:06, 3 June 2009 (UTC)

McAfee SiteAdvisor says the site is OK. But it's possible that the site has been recently compromised, and that Firefox/Google has more up-to-date information than SiteAdvisor. In any case, the reference itself is fine, since it's to a SIGGRAPH paper - the links are just for the reader's convenience and we could remove them without removing the reference. In fact, I've just done this. --Zundark (talk) 07:44, 4 June 2009 (UTC)

Sigh...yes sorry, my website was attacked by Chinese hackers (using various .cn sites). I hope to have that resolved today. DonPMitchell (talk) 17:53, 4 June 2009 (UTC)

I've restored the links now that Google is no longer blacklisting the site. (Firefox may continue to warn about the site for a while, until it gets the updated blacklist.) --Zundark (talk) 10:35, 5 June 2009 (UTC)

Jaggies As PDA Screen Recognition Error

"Jaggies" is also a known term for a screen recognition error on PDAs and Smartphones, originating (AFAIK) in the Apple Newton community. I added a one-sentence addition for the use of the term and added a blogref. It could certainly use a better reference than a blog, and possibly expansion. But as I originally came here to look up the term under this entry to link to it, and it was absent, there you go! (User LaughingVulcan operating in anon mode.) 98.228.92.5 (talk) 16:53, 27 December 2009 (UTC)

Causes section is mostly wrong.

The first sentence talks about the "staircase effect", which is on the right track, but doesn't really get into the technical aspects. The problem is naive (and therefore computationally cheap) methods where a pixel is assigned to a single line, polygon, etc., even if that pixel actually covers multiple objects (including the background).

Here's an image I made to graphically illustrate what's going on. The naive approximation just finds the object at the center of the pixel and renders the entire pixel with that value, while the blended method properly averages the different colors according to how much of the pixel they cover. In practice, most anti-aliasing methods only approximate proper blending, with various degrees of success, but that's the idea. And it's what happens automagically in a digital camera, since each pixel receives light from everything in the pixel's field of view proportionally to the solid angle each object takes up.

From there, nothing is really correct. While it's true a very high resolution monitor (or a standard monitor viewed from far away) could render the jaggies at a small enough angular size that a human viewing it wouldn't notice the jaggies, it doesn't actually remove the jaggies. I.e., the jaggies aren't caused by a low-resolution output device; they are merely more noticeable as resolution decreases.

Next up, jaggies don't occur because of resolution conversions. That's called Pixelation. Jaggies are the stairstepped lines caused by a lack of proper blending, which is what this page is about. Pixelation is caused by zooming in on an image until each pixel is noticeably distinct from its neighbors, and is often seen when stretching low-resolution images across large monitors or printing them on large posters. Pixelation can make jaggies more noticeable (especially if you don't use some kind of filtering to smooth out the edges), but they aren't the same thing.

Finally, vector graphics most certainly don't look the same regardless of resolution. A low resolution device will look bad whether you're viewing bitmaps or vector graphics, and a high resolution device will look good as long as the source is good. The advantage to vector graphics is just that simple shapes can easily be re-rendered to the resolution of a given target device to minimize both jaggies (on edges) and pixelation (both on anti-aliased edges, and on gradients and such). Once you get to complex shapes, like photographs, the quality of the image will only substantially increase if you add more data, whether that's more pixels or more vector elements. It's like rendering a ten-triangle 3D model of a banana in 4k. Sure, the edges look a little better, but it still looks like garbage.

Of note, the lines about low-resolution output devices and bitmap conversions are duplicated in the Solutions section (and are still wrong there) without any kind of solutions attached to them.

It's probably worth noting that jaggies are primarily seen on monitors (or TVs) with square pixel elements. On a halftoned newspaper or similar, there are still jaggies if not properly anti-aliased, but they are less jarring because of the randomized, curvy edges of the dithering process. Old CRT monitors tend to blur the edges of pixels into each other, causing a slight anti-aliasing effect that helps diminish the jaggies. Any monitor using non-square elements (I don't actually know of any) would cause different shapes to be jagged. For example, a hex-grid would render lines at 60° from horizontal just fine, but would have jaggies on vertical lines (or be fine at 60° from vertical and have jaggies on horizontal, depending on the grid's orientation). A screen with randomized pixel elements (like the human retina, but for displaying images rather than capturing them), would have less noticeable jaggies, but everything would tend to be jagged regardless of orientation. Etc. 199.127.114.114 (talk) 11:39, 1 January 2019 (UTC)

This is a duplicate of Pixelation

It looks like this article should be merged in Pixelation

Kmkaplan (talk) 10:11, 19 February 2022 (UTC)