Talk:List of monochrome and RGB palettes
- The entire paragraph was a fossil; even it is not mine, but part of the primitive stub article (february 2007, which mixes and misleaded many concepts). As far I can remember, it was intended for something related to YUV to RGB accuracy or so. It was nonsense in the current article (obvious). This was my fault, sorry.
- Well, actually you wasted a right opportunity to delete things! ;-) -Ricardo Cancho Niemietz (talk) 14:24, 29 February 2008 (UTC)
- It means that the human eye is less sensitive to blue light and more sensitive to green light (see File:Luminosity.png). For this reason, a #FF0000 red color will look brighter than a #0000FF blue color but darker than a #00FF00 green color, and green and red have a higher weight in calculating the luminance than blue. (I think it's clear enough, but maybe adding a reference helps understanding it; check candela for possible references. Another option is to put a pic with 32 levels of R, G, and B (don't know if there's already one) to show how green bands appear sharp and blue ones blurry.) —22.214.171.124 (talk) 15:29, 17 July 2013 (UTC)
- The currently tagged sentence "The 3-3-2 bit RGB use 3 bits for each of the red and green color components, and 2 bits for the blue component, due to the lesser[dubious – discuss] sensitivity of the normal human eye to this primary color" is still dubious. The lesser sensitivity to blue is not an obvious or sensible reason for using coarser quantization there. The lower resolution for blue is more likely, but still dubious. A reliable source that says why would be useful there. Dicklyon (talk) 15:37, 17 July 2013 (UTC)
Banding in 24bit samples
The section on 24-bit palettes have 4 sample images which it states "must be seen as continuous" and it suggests that if banding is visible, the display is running at 16-bit or less.
I have looked at the images from 2 different computers, both of them running at 24-bit, and on both of them the two first images (Red=0 and Red=85) have very clear banding, while the third (Red=170) have less obvious banding and the fourth (Red=255) has subtle banding.
I realise that the displays could be badly calibrated, and my knowledge of color and displays is very limited so I haven't made any edits, but for someone with my eyesight (which to the best of my knowledge is average at best) the claim that there is no banding is clearly false. —Preceding unsigned comment added by 126.96.36.199 (talk) 13:55, 26 May 2008 (UTC)
Test image clarification
If the intent of the test images is to represent a collection of possible colors at each bit depth, then, at a minimum, the 15-bit image is incorrect.
It contains over a thousand colors (for example, #CED6FF) which are not representible with 15 bits.
If this is not the intent, could some explanation be added (to the images themselves, perhaps) as to how these images are representative samples of their respective palettes? —Preceding unsigned comment added by Joedavidson (talk • contribs) 14:37, 6 June 2008 (UTC)
4-bit RGBI palette incorrect
The example palette is a 4-bit RGBI palette, but it is not the CGA 4-bit palette. See Color Graphics Adaptor. The bright part of the palette are not pure colors and also instead of light yellow a brown color is used. —Preceding unsigned comment added by 188.8.131.52 (talk) 09:45, 22 September 2008 (UTC)
Atari 2-bit Greyscale
Around 1989 I saw an Atari computer that supported a high resolution monitor in monochrome and a medium resolution monitor in two modes, one of which was a 2-bit greyscale. There was even a Breakout-style game available that supported the high resolution monitor by dithering to obtain an intermediate grey (it only used three of the greyscale colours when played on a medium resolution monitor).
Why non-linear color value increases?
I've noticed that in all low-depth RGB and grayscale cases the increases in values of individual colors were non-linear for some reason. For example, the case of 6-bit RGB palette, the acceptable levels of individual colors when converted to 24-bit RGB are 0, 104, 183 and 255 when they should be 0, 85, 170, 255 (see Enhanced Graphics Adapter). This is also confirmed by the fact that on my own PC the 16-bit RGB test image looks differently in the 16- and 24-bit color modes. This also explains the "Test image clarification" issue above: the extra colors are there because of the wrongly chosen individual color values. —Preceding unsigned comment added by 184.108.40.206 (talk) 09:26, 14 March 2009 (UTC)
- If the values were equally spaced, the intensities would be very nonlinearly spaced. The values shown in the table on the EGA page are equally spaced in 24-bit RGB, so it's not clear why you're asking. It sounds like you've got different gamma corrections in your different color spaces. Dicklyon (talk) 16:15, 14 March 2009 (UTC)
6×6×6 = 216 "web colors" are missing
Another unrealistic palette article with no dithering
I find it hard to believe that a supposedly encyclopedic article discussing paletted colors of older computer systems would completely exclude examples of dithering from the article.
In the old days when display capabilities were limited, dithering was heavily used to overcome the inherent technical limits of the hardware. The early Macintosh used dithering for all photo-realistic images such as in HyperCard.
Is the original article author just trying to exemplify the limits of the technology at the time, and how much more awesome things are now by comparison?
To exclude dithered examples makes this article either a half-truth of how things really were, or mostly irrelevant.
I think it's worth mentioning that 18 bits is the most colours an LCD screen could produce, up until quite recently. It was a limitation of LCD manufacturing. Many laptops had 24-bit graphics with an 18-bit screen, ignoring the 6 least significant bits to keep things simple, though when plugged into a CRT monitor you got the full 24 bits. 220.127.116.11 (talk) 12:53, 3 September 2013 (UTC)