Talk:XO colors: Difference between revisions

From OLPC
Jump to navigation Jump to search
No edit summary
No edit summary
Line 1: Line 1:
How were these colors arrived at? I'm curious because there are some that are ''very'' close together (EG, BCCEFF and BCCDFF). When you look at just the luminance (for evaluating distinguishability for colorblind people or on a b/w screen), there is even more clustering, with most adjacent colors (ordered by luminance) differing by <2%.
How were these colors arrived at? I'm curious because there are some that are ''very'' close together (EG, BCCEFF and BCCDFF). When you look at just the luminance (for evaluating distinguishability for colorblind people or on a b/w screen), there is even more clustering, with most adjacent colors (ordered by luminance) differing by <2%.


Evenly spacing 20 colors along the luminance spectrum would provide a set of much more distinguishable colors (5% difference in luminance between adjacent colors). Using a 8/21 rotation around the hues (all with 100% saturation) can give fairly distinguishable colors. That could be further improved by either varying saturation, increasing the hue rotation at the extremes of luminance (and decreasing toward the middle), or any of a number of other means.
Evenly spacing 20 colors along the luminance spectrum would provide a set of much more distinguishable colors (5% difference in luminance between adjacent colors). Using a 8/21 rotation around the hues (all with 100% saturation) can give fairly distinguishable colors. That could be further improved by either varying saturation, increasing the hue rotation at the extremes of luminance (and decreasing toward the middle), or any of a number of other means. &mdash;[[User:Pfeilgm|Greg]]


: I also noticed the BCCEFF/BCCDFF similarity... I assume it is a typo in the code. I believe the pairs were arrived at by hand-matching color swatches for pairs that "looked good" together (based on a recollection of a photograph of a wall with color pairs on it). Note that these colors are only used to form the color pairs for user logos, and that while these logos are used to distinguish users, there are additional ways to distinguish them (by name, at the least). Which is a good thing, because with only several hundred pairs, it is [http://en.wikipedia.org/wiki/Birthday_paradox statistically probable] that a class of around 20 students will have a duplicate color pair. Also note that while some colors are close together in luminance, it's possible that the ''other'' color in their respective pairs may be sufficiently different to distinguish them. The OLPC people are aware of the value (ahem) of using [[OLPC_Human_Interface_Guidelines/The_Sugar_Interface/Colors | luminance-spaced colors]] in their interface, especially given the XO's additional monochromatic screen mode. &mdash;[[User:Leejc|Joe]] 00:22, 7 December 2007 (EST)
: I also noticed the BCCEFF/BCCDFF similarity... I assume it is a typo in the code. I believe the pairs were arrived at by hand-matching color swatches for pairs that "looked good" together (based on a recollection of a photograph of a wall with color pairs on it). Note that these colors are only used to form the color pairs for user logos, and that while these logos are used to distinguish users, there are additional ways to distinguish them (by name, at the least). Which is a good thing, because with only several hundred pairs, it is [http://en.wikipedia.org/wiki/Birthday_paradox statistically probable] that a class of around 20 students will have a duplicate color pair. Also note that while some colors are close together in luminance, it's possible that the ''other'' color in their respective pairs may be sufficiently different to distinguish them. The OLPC people are aware of the value (ahem) of using [[OLPC_Human_Interface_Guidelines/The_Sugar_Interface/Colors | luminance-spaced colors]] in their interface, especially given the XO's additional monochromatic screen mode. &mdash;[[User:Leejc|Joe]] 00:22, 7 December 2007 (EST)

I thought that OLPC was making a point to not deliver duplicate color pairs to any single school ([http://laptopgiving.org/en/faq.php Why are the Xs and Os different colors on each XO laptop?] – the random selection (and hence the birthday paradox) only applies to the laptops sent to the US and Canada). And while their 27 colors can probably make 400 easily recognizable pairs, I don't think they have any chance of being recognized in the b/w mode. Mousing over everyone in the neighborhood to find a particular friend's name seems a bit tedious (but I don't have my XO yet, so I can't be sure). The need is not to distinguish between the two colors in the pair (when both colors are the same, that's probably ''more'' recognizable, sinc those would be relatively rare), but to make sure that no two people share the same combination of stroke and fill colors (or luminance, which is more likely to be a problem). &mdash;[[User:Pfeilgm|Greg]] 03:34, 7 December 2007 (EST)

Revision as of 08:34, 7 December 2007

How were these colors arrived at? I'm curious because there are some that are very close together (EG, BCCEFF and BCCDFF). When you look at just the luminance (for evaluating distinguishability for colorblind people or on a b/w screen), there is even more clustering, with most adjacent colors (ordered by luminance) differing by <2%.

Evenly spacing 20 colors along the luminance spectrum would provide a set of much more distinguishable colors (5% difference in luminance between adjacent colors). Using a 8/21 rotation around the hues (all with 100% saturation) can give fairly distinguishable colors. That could be further improved by either varying saturation, increasing the hue rotation at the extremes of luminance (and decreasing toward the middle), or any of a number of other means. —Greg

I also noticed the BCCEFF/BCCDFF similarity... I assume it is a typo in the code. I believe the pairs were arrived at by hand-matching color swatches for pairs that "looked good" together (based on a recollection of a photograph of a wall with color pairs on it). Note that these colors are only used to form the color pairs for user logos, and that while these logos are used to distinguish users, there are additional ways to distinguish them (by name, at the least). Which is a good thing, because with only several hundred pairs, it is statistically probable that a class of around 20 students will have a duplicate color pair. Also note that while some colors are close together in luminance, it's possible that the other color in their respective pairs may be sufficiently different to distinguish them. The OLPC people are aware of the value (ahem) of using luminance-spaced colors in their interface, especially given the XO's additional monochromatic screen mode. —Joe 00:22, 7 December 2007 (EST)

I thought that OLPC was making a point to not deliver duplicate color pairs to any single school (Why are the Xs and Os different colors on each XO laptop? – the random selection (and hence the birthday paradox) only applies to the laptops sent to the US and Canada). And while their 27 colors can probably make 400 easily recognizable pairs, I don't think they have any chance of being recognized in the b/w mode. Mousing over everyone in the neighborhood to find a particular friend's name seems a bit tedious (but I don't have my XO yet, so I can't be sure). The need is not to distinguish between the two colors in the pair (when both colors are the same, that's probably more recognizable, sinc those would be relatively rare), but to make sure that no two people share the same combination of stroke and fill colors (or luminance, which is more likely to be a problem). —Greg 03:34, 7 December 2007 (EST)