Talk:Pixel/Archive 1

old junk not in a section

and in english this all means??? ÉÍREman 16:16 May 6, 2003 (UTC)

Indeed. I've had a go at a clearer intro, but I'm not happy with it. I hope someone else can improve it. The problem is that it's very hard to say what a pixel is: screen element; printing; part of a digital image. Though in the last case we're strictly dealing with texels; but only the 3D gaming industry makes that distinction AFAIK -- Tarquin 16:30 May 6, 2003 (UTC)

I'm a historian whole technical skills are such that if I change a light bulb I am likely to black out all of North Dublin, so I am afraid the article went completely over my head, but then anything technical does anyway. But then I suppose people people are are technologically literate might find the stuff I write on historiography or royal naming procedures over their heads too. Each to his or her own! I notice BTW that an earlier version mentioned a mac's 'thousands' and 'millions' in term of screen colours, so I learned something about what that actually means on my eMac. Except that that info was culled from later versions!!! I decided to check just what the hell pixel was because I bought a new digital camera that uses 3.2 million pixels so I was hoping to find out what it was I had bought. The joys of being technologically illiterate! ÉÍREman 19:44 May 6, 2003 (UTC)

Very nice picture, Tarquin! I'm even willing to overlook the fact that the lines connect the wrong corners. ;) -- John Owens 21:41 May 6, 2003 (UTC)

dang! I'll fix it later. -- Tarquin (boy am I glad I always keep my layered photoshop documents!)

I'm a bit put off by: " This can be expressed as a single total,", because it seems to me the single number is a product, not a total... but I can't see how to change it without making it too complicated -- Tarquin 21:48 May 6, 2003 (UTC)

Is this the bit that now reads "For depths larger than 8 bits, the number is the total of the three RGB ..."? If so, I agree that total is the wrong word, but describing it in a way that a non computer-literate person will immediately grasp is difficult. Perhaps something along the lines of "For depths larger than 8 bits, the bits themselves are divided into three seperate values that describe the intensities of the red, green, and blue components of the desired colour.", since the values are not really either a product or total. Lumpbucket 01:08 August 12, 2003 (BST)

I added a section on what a digital camera "megapixel" REALLY is, and also added sections for pels and sub-pixels.

On the digicam megapixel subject, I sent an email to the apparent author of http://megamyth.homestead.com, asking him to contribute to the subject. Scott McNay 07:45, 2004 Feb 8 (UTC)

On another subject, can someone clarify several issues? I'd like to know:

whether a "dot pitch" is the same as the pixel size, as loudly implied by this article, and not clarified at all by the dot pitch article?

how "native resolution" (often seen in relation to LCDs) relates.

whether "native resolution" applies to CRTs or not (which is implied by this article in the sub-pixel section), at least for horizontal resolution.

How CRTs and LCDs display different horizontal resolutions when they have embedded color masks, and ditto for vertical resolution for LCDs.

Scott McNay 07:45, 2004 Feb 8 (UTC)

Ok, looks like everything's been explained, and it all makes sense to me now; just have to clean up a bit now. Scott McNay 01:30, 2004 Feb 14 (UTC)

To do:

Find or make a picture showing how sub-pixel font rendering (SPFR) helps. Put it on appropriate page
Add mention on appropriate page of Apple (?) using SPFR on Apple II (?)
Factor Pixel page so that image pixels are discussed separately from monitor pixels, ec. Right now, kinda mixed up.
Find or make a picture showing pixel geometry for a typical digital camera. Mention that a digicam triad is quite large, due to having more green pixels than red or blue pixels. May want to un-redirect megapixel page.
Clarify that pixels and sub-pixels are logical constructs, and triads and dots are physical constructs.
Can someone clarify that SPFR really requires that the SPFR routine know the LCD monitor's native resolution, and/or works best when monitor is set to native resolution?
Can someone explain how SPFR works on CRTs (see Apple II reference), when the image's location on the screen can vary from moment to moment?

Scott McNay 02:29, 2004 Feb 14 (UTC)

More info about SPFR available at http://www.grc.com/cleartype.htm

Considering that apparently some monitors (CRTs?) have vertical stripes, that would imply that diagonal dot pitch measurement simply doesn't apply to them, since verical resolution would be limited only by the number of scan lines that can be squeezed onto the screen. Comment?
Here is an example of a great pixel website Commerce.co.uk
Update pixel geometry to mention triads instead of pixels
Merge triad and pixel geometry pages? Could edit/replace pictures to show both the geometry and the outline of a triad on the pictures
Update triad to clarify that sub-pixel is not the same as a triad

Scott McNay 03:01, 2004 Feb 14 (UTC)

As i have always known and was taught, dot pitch IS a diagonal measurement. It is the measurement of the shortest distance between 2 like-colored phosphor dots (what someone started terming sub-pixel). Since a triad is made up of the 3 color dots in a triangle formation, as they are placed on the screen the measurement of dot pitch is always diagonal.

Dot pitch simply doesnt apply to an LCD screen, since LCD does not use phosphor dots, but rather a square made up of 3 tall rectangles in the primary colors. Sorry i do not know the term for that element in an LCD screen.

Im sure someone can clarify it more in the article.

Enos Shenk 22:37, 14 Feb 2004 (UTC)

Does dot pich apply to Trinitron tubes? They have stripes also, I believe, and I don't have one to look at, but I would guess that the stripes are solid, not broken, meaning no vertical component.

So now the question is, are some companies actually reporting incorrect values for dot pitch? The first link below has some discussion on the subject.

http://www.pctechguide.com/06crtmon.htm (starting with the dot pitch section)
http://www.pctechguide.com/07panels.htm
http://entertainment.howstuffworks.com/tv.htm
http://computer.howstuffworks.com/monitor.htm (see page monitor4)
http://computer.howstuffworks.com/question401.htm (none of the links at the end of the article are good)

Add note that SPFR is sometimes called "pixel borrowing".

After reading Wiki, HSW seems, well, mildly biased and incredulous. :)

(Forgot to put my sig above.)

Ok, I will edit (unless someone else does it first; I have some cleaning up to do around here, and headed out for a few hours with a friend this afternoon) the various articles to indicate:

CRTs are analog and LCDs are digital, in the sense that on an LCD, a specific pixel can be guaranteed to correspond to a triad, whereas on CRTs, a pixel is most likely to land on varying parts of two triads (a pic showing this would be nice). Would it be accurate to say "Because of this vagueness, "native resolution" does not apply well to CRTs"?
"sub-pixel" refers to digital displays
"phosphor dot" refers to analog displays
stripe pitch (sometimes called dot pitch) is a horizontal measurement for Trinitron CRTs
dot pitch is a diagonal measurement for non-Trinitron CRT screens. For non-digital displays, the term is rather vague.
SPFR only applies to digital displays
SPFR seems to help on analog displays because it is an anti-alising method but standard methods give beter results for non-digital displays.
Add an image for slotted mask. [1] shows the dot pitch being measured horizonally in this case, which I think is likely part of the cause of the "some companies measure wrong" issue.
Incorporate other dot pitch caveats from [2].
Etc.

Is someone else here better than I am with making pictures? I just have MS Paint, here... Scott McNay 17:47, 2004 Feb 15 (UTC)

New pixel article

I'm starting the new article at talk:Pixel/new. The idea about tiles in mosaic came from Fuzheado. Scott McNay 06:16, 2004 Feb 18 (UTC)

Difference between px, pt, em

Is there an article which covers the difference between px, pt, and em measurements? These are used in drawing programs, css, etc. A single article that ties them all together would be nice. (also pica, ex, etc.)

css definitions: http://css.weblogsinc.com/entry/1234000933032102/

SVG definitions: http://www.xml.com/pub/a/2001/03/21/svg.html (what i was specifically looking for)

how it relates to font size in pt, how it relates to DPI, etc., etc.

- Omegatron 16:26, Apr 6, 2005 (UTC)

Rollback

I meant to edit a local version of this page, but instead edited the Wikipedia version. Is there a way to rollback to the previous version?

This article needs a lot of work

This article contains a large amount of pseudo-erudition caused by layer on layer of pedantic corrections. Some of the pedants knew something about image processing. Unfortunately, most didn't. It now desperately needs editing by actual image-processing experts, followed by extensive copyediting to remove redundancy and simplify flow. -- Karada 20:18, 22 Jun 2005 (UTC)

On a quick read, I found a lot of good stuff here. I hope we don't get carried away with style and whatever and reduce the meat in the article. I appreciated what I read. May give it a critcal read later.

Phil ^talk 14:47, 14 September 2005 (UTC)

I've worked it over some, from a position of knowledge. I hope my changes don't fall into the overly pendantic category, but that's always a danger when trying to make sloppy stuff be correct. Dicklyon 03:10, 25 May 2006 (UTC)

picture element versus picture cell

which is correct? the article states element, but i have always heard cell - mastodon 22:56, 22 April 2006 (UTC)

Most early uses of pixel (1965-1980) specifically said it meant picture element. The notable exceptions that I am aware of were in publications and patent applications out of Xerox Palo Alto Research Center, 1976 to 1982, which used picture cell. For example, see US4034343: Optical character recognition system, by Michael Wilmer, filed 1976-10-01. But since you weren't born yet, maybe that's not where you heard it. Your pixel historian, Dicklyon 03:07, 25 May 2006 (UTC)

Coinage

Who coined the term? Does anyone know? I know a guy who thinks he may have coined it in a naval document in 1973 or so. Jfing ers88 01:37, 15 June 2006 (UTC)

The earliest known publications of pixel are by Frederic C. Billingsley in 1965; but he did not coin it, nor did Keith McFarland, subcontractor from whom Fred leaned the term. We don't know who coined it. See the last external link in the pixel article. Dicklyon 02:06, 15 June 2006 (UTC)

Subpixels -- new images

What's the story with the new images? I notice one is alternating black and white pixels, while the other is alternating green and magenta. Both look gray. So what does this have to do with subpixels, and why is the first one repeated, and then the pair are shown combined, and there are not words about the figure or its point. Can someone please explain or fix? Dicklyon 05:20, 1 July 2006 (UTC)

At least on my display the first two look different from each other. Maybe you have got different equipment/drivers. The pixels on the image file, show up differently on display as the software tries to fit 100*100px on the image to 98*98px or 97*97px on display. Doesn't the 100/98 and 100/97 -ratios force the display driver handle subpixels? I'm not sure if this is the right place for these. Some of the software I have, show the combination as gray with a magenta line in the middle. Also, Image:Resolution illustration.png on image resolution -article might be a better example. Dreg743 06:20, 2 July 2006 (UTC)

The drivers probably have nothing to do with it. The web browser scales the image, and it's usually (at least on Mozilla/Firefox) a quick and simple nearest-neighbour algorithm. Safari and/or Opera might do a higher-quality resampling of the image. So relying on the thumbnail width to force "subpixels" is not going to work. Different web browsers do things differently, it's how the web works. Just what exactly are you trying to show with these images? Imroy 11:16, 2 July 2006 (UTC)

Well, the difference of one pixel changed the appearance of the image (on my (browser) display. On my system, they have one light grey circle (98px) and 4 light grey circles (97px) on dark grey, even though the 'image' is what it is. Just thought this was an universal fenomenon. At least I know now that that those images aren't always like that. Thanks for the response. removing images.... Notgray.GIF and Stillnotgray.png ... Dreg743 12:24, 2 July 2006 (UTC)

pixel size.

can pixels be different sizes? or are they a physical unit of measurement (distance).

They can be arbitrary sizes, but within a given system are usually of a constant size. When you change the display resolution on a CRT, you change the pixel size. Camera pixel sizes can sometimes be changed by an integer factor, by aggregating several pixels into one. In an image file, pixels have no definite size, but a size can be attributed to them via metadata (for example, a tag that says how many pixels per inch); in that sense, the image pixel size is easy to change. Dicklyon 22:41, 4 July 2006 (UTC)

thanks... so, for example, on a standard TV, all else being the same.. would a 40" TV have 4 times the number pixels of a 20" TV?

No, not at all. With a TV, the number of pixels is determined by the video source. The screen doesn't have pixels of its own, and even if it did, bigger wouldn't mean more of them. With LCD and plasma screens, you can get pixels associated with the screen, but again the screen size has little to do with number. There are various sizes, with names like EDTV that correlate with numbers of pixels, but I'm no expert on that. Dicklyon 18:51, 6 July 2006 (UTC)

The resolution of the image indicates the number of pixels. TV is generally only one resolution, no matter how large or small the TV is; a small TV which will fit in your hand is the same resolution, and therefore the sme number of pixels, as a large console or projection TV. You only change the resolution when you change to a different type of TV, such as from NTSC to HDTV. --Scott McNay 04:26, 28 August 2006 (UTC)

New image removed

I don't understand this image. Can someone explain, maybe fix it or make a better caption? What is the relationship between the top row images and the bottom row, which are labeled only by different X values? Dicklyon 18:09, 27 August 2006 (UTC)

The bottom set isn't labelled well and/or properly. The top two look like what you'd see logically, and the bottom two show what you see physically, when looking closely at a flatpanel monitor or a Trinitron-style CRT. --Scott McNay 03:10, 28 August 2006 (UTC)

Opinions, Sigma/Foveon Hype Removed

Most of the section on "Megapixel" had almost nothing to do with the topic at hand, and can be found in the sections on CCDs and digital cameras. In addition, an encyclopedia entry isn't really the place to continue the Foveon vs. Bayer Pattern sensor debate. As most of the section was extraneous, I removed everything but the simple definition. ElGordo 23:32, 27 September 2006 (UTC)

I restored most of it, but took out some of the informal stuff like "marketing ploit". I don't think the statements about Foveon here are either hype or debatable; it just explains how they count pixels in both types of sensors, which is easily verifiable. Dicklyon 23:40, 27 September 2006 (UTC)

I think my issue with it has more to do with the question that was on my mind... in an encyclopedia entry for pixel, which is a simple term, do I expect to find a full explanation of pattern array interpolation vs. stacked sensor, or just a general definition of pixel with perhaps a few "see also's" to topics briefly covered. The topic as written now is verifiable and true, but it isn't why someone typed the word pixel into wikipedia. ElGordo

Not everyone is there for the same reason, I suspect. It doesn't seem out of place to explain the derived term "megapixel" as commonly applied to cameras, without having to read articles on sensors, etc. Similarly, the application of "pixel" and "subpixel" to displays, etc., is explained, and rightly so, in this article. Dicklyon 23:58, 27 September 2006 (UTC)

I have also shifted the order of terms in the section below this, as the VGA/SVGA etc. names were created to define the resolutions listed and by standard format should be listed first on the line, not the other way around. All information retained and the topics were linked to the appropriate sections about them.ElGordo

I have to say...

I'm a real technophile, but I read this article several times to no avail. I always thought that the pixel was just that little dot that is multiple colors... Now it's trying to say that it's something positively different? Hmm... I think this could be revised so the people who haven't majored in English, Latin, and Arabic can understand it :) ~ Theta :D 06:57, 14 November 2006 (UTC)

If that's all you thought it was, then you learned something, yes? I missed the Arabic part. Dicklyon 06:37, 19 November 2006 (UTC)

Megapixel section

The information I added has been removed (Dicklyon : revert long-winded not-quite-right stuff specific to 4:3 cameras with non-real numbers). OK it might have been a bit long winded, but could it not have been edited instead of completely removed. I think the section could do with some more information. I'm not sure which bit was not-quite-right.

It was specific to 4:3 cameras, but this includes the majority of consumer digital cameras and video camera sensors. Maybe the calculations were not needed, but I find it useful to be able to calculate the horizontal and vertical pixels counts given any Megapixel value. The numbers in the table gave an exact number of megapixels (unless I am mistaken 1MP is 10^6 and not 2^20 as in the computer Mega). I agree that these mathematical figures do not correspond to actual image sensor horizontal and vertical pixel counts, but it does not make it incorrect. The image sensors used have different pixel counts depending on the manufacturer.

I thought that this comment was useful, as I believe it is a common misconception : 'Note that a 6 megapixel image does not have double the number of horizontal pixels of a 3 megapixel image. It is not until 12 megapixels that the number doubles.'

Andy, 19 November 2006

OK, maybe I overdid it, but it looked too hard to fix. Of course, it can be recovered and reworked any time you or someone is motivated to do so. Being specific to 4:3 bothered me, since DSLRs are usually 3:2 and there are other aspect ratios as well; if you're going to put equations, try to be more general. The main thing that was not quite right was the list of sizes, which didn't correspond to any actual camera or sensor sizes of the stated megapixel ratings, and would therefore just confuse people who were trying to compare with known facts. The stated relationships were also not applicable to a 3D pixel arrangement (rows x cols x layers) like in the Foveon sensors, nor to the pixel arrangements in SuperCCD sensors. And finally, it was not really obvious why one would generally need to go from megapixels to numbers of rows and columns; what for? It's hard to be sufficiently general there, since aspect ratios and organization and roundoff error are so variable; and those numbers are usually provided explicitly with any camera. Using two sentences to point out that twice as many rows and columns means four times as many pixels also seemed like overkill, though it's a good point. But is any of this needed to explain what megapixel means? Why it is in the pixel article if it's about particular digicam organizations? Seems like it would go better in a digicam article. Dicklyon 06:36, 19 November 2006 (UTC)

More etymology

I removed this interesting contribution:

The concept of digital still photography was conceived by Eugene F. Lally of the Jet Propulsion Laboratory/California Institute of Technology. Lally propopsed an on board all-optical guidance system for manned planetary missions. His paper "Mosaic Guidance for Interplanetary Travel" was published in 1961 at the annual American Rocket Society convention. The concept used an array of tiny light sensors in a mosaic pattern with each element refered to as a mosaic element that formed an image sensor. The sensor would record star and planet positions during transit for navigation purposes and when near an approaching planet would provide additional stadiametric information for guidance purposes of the astronauts onboard.

This is fascinating, and I'd love to hear more about it. If there's a verfiable source, that would be good to know. But even if there is, how does it relate to pixel etymology? The concept of digital still photography predates that in a lot of image coding work in the 1950s, doesn't it? Dicklyon 03:09, 2 December 2006 (UTC)

Mr. Dicklyon: The "mosaic element", later to be called "pixel" was historically first used to describe the concept of recording light for still image capture. It was a small light sensor, one of many that were grouped together in a large number to form a mosaic image plane for navigational use on manned spacecraft. It was described in the paper of 1961, "Mosaic Guidance for Interplanetary Travel" from JPL. I have a copy of that paper. It is part of the history of how digital photography began and was presented at an annual meeting and published in their magazine "Astronautics". In the trade, this paper has been accepted as the initial disclosure of still digital photography as it first described the components needed. I would think it made sense to preserve this interesting piece of the Space program and the history of photography. E. Lally

I think it makes sense, too. I'd love to have a copy of it. Can you email me dicklyon at acm dot org please? Dicklyon 04:07, 2 December 2006 (UTC)

Merge from picture element

I just found this little-editted article on picture element that should really be a redirect to here. It might have some good content worth merging in, but I haven't looked for it yet. Support or oppose? Dicklyon 04:35, 14 December 2006 (UTC)

Finding no oppostion or help, I converted it to a redirect. Here's the entire text of it (with headings demoted) in case there are bits anyone wants to incorporate into pixel:

Picture element

This is the original and proper name for the unit of resolution in visual display units (VDUs). They are now referred to as 'pixels', however this is merely a contraction of the phrase 'picture element'. A computer monitor with a resolution of 1280x1024 has 1280 picture elements (or pixels) horizontally, and 1024 vertically. A higher resolution means more detailed images can be displayed on the screen due to the fact there are more pixels, therefore monitors supporting higher resolutions are usually more expensive.

Pixel colours

A pixel is further defined by the display device colour-depth. This is a definition of the ability for a pixel to display a range of colours, or grey shades or simply to be on or off (white or black). Usually expressed in digital bits, displays operate in a specific 'Bits-Per-Pixel' (BPP) mode indicating the amount of computer bits used in all pixels on the display. In early graphical VDU displays, there was simply one bit for every pixel. The display was said to be monochrome, or black and white. Over time, with the advent of more capable display electronics, we now have 16, 24 or even 32 bits per pixel. The more information (BPP) we use for each pixel, the better a reproduction of colour in individual pixels we have.

The easiest example of this can be described with a 24 bit pixel. The primary colours of light - red, green and blue, can easily be divided into 24 bits, giving eight bits for red, eight for green and eight for blue. Eight digital bits, each being a 0 or 1 gives a total of 256 (2 to the power of 8) possible variations of each primary colour. Three primary colours times 256 (256 x 256 x 256) gives you a total 'colour palette' of 16,777,216 individual colours that any pixel can be at any one time on the display.

Digital storage of picture elements

Display devices represent the colour of a pixel using exactly that - values from 0 to 255 (256 in all) for red, green and blue 'components' of the pixel (if the display is operating in 24bpp mode). A complete 1280 x 1024 pixel screen, being made up of three bytes for every pixel, would mean we need to store 3,932,160 bytes, or just under 4 Megabytes, just for one screen.

For older display hardware that was not as capable, we did not have the luxury of storing 4Mb for the screen, so instead we used a palettised display. Palletising a screen involves not simply granting each pixel an ablility to display its own colour, but instead to represent each pixel with index within a separate table of colours, called a palette.

A common palettised arrangement was an 8-bit palettised display. Each pixel is not an individual colour, but instead is an 8-bit number (256). This number is used as a 'lookup' to the palette table. The table will also be 256 'rows', each row would contain a full representation of a particular colour. So if row one contains light purple, row two contains dark brown, row three contains bright orange, etc, a particular pixel would be able to point to one of these colours, and hence BE that colour.

This was usually fully handled by the display hardware in the computer system and so would operate quickly. The upside - less memory storage for each display, since each pixel would only be one byte, a 1280 x 1024 screen would take up 1,310,720, or 1.3Mb plus a very small palette table (24 bpp = 3 bytes times 256 rows = 768 bytes for the palette table). The downside - The whole display would only be able to show colours out of 256 different pre-defined colours at one time. This usually resulted in less than perfect representations of photographs and other high quality graphics, but was nonetheless effective and worthwhile. In practise though, older display hardware wouldn't be able to show 1280 x 1024 pixels, but more likely at 640 x 480 or 800 x 600 (480,768, or 480KB). A far more practical proposition given the high costs of display memory in the 80s and early 90s.

Palettised picture files

.GIF (Graphics Interchange Format) files are stored in a similar palettised fashion. If a GIF file were displayed occupying the whole screen of a modern full colour computer display, you would see similar poorer colour reproduction than that offered in other full colour image formats, such as .JPG.

This computer graphics–related article is a stub. You can help Wikipedia by expanding it.

a pixel is not a little square?

this is in the caption for a screen showing pixels as circles and other shapes.

Isn't it more appropriate to say "a pixel is not always a little square"?