|WikiProject Color||(Rated Start-class, Mid-importance)|
A recent edit removed all mention of the Purkinje effect, leaving only a link in the See also section to the article. I think it is relevant and useful to mention the Purkinje effect briefly in this article, rather than giving only a link to the article about it. If there are no objections, I propose to restore the content with an added citation to satisfy the fact tag that was placed on the removed content. - Neparis (talk) 01:28, 6 May 2008 (UTC)
- Mention it, by all means, but back it with a WP:RS. Don't just leave it for some joe to fill it in two years from now. --Adoniscik(t, c) 01:33, 6 May 2008 (UTC)
Kruithof curve chart
I think, if the explanation in the article is right, then the chart is wrong:
The explanation says that, at a given (constant) temperature, the "hue" of the ligt varies with the variation of its luminance. Otherhand, in the chart when you go un and down (luminance axis) the color (hue) remains unchanged. I think the color gradient in the chart must be vertical, not horizontal.
If I'm wrong, please tell me. And please excuse my bad english.
- I think you are using the chart wrongly. The color is just a crude representation of the color temperatures along the X-axis, and given that it's white at about 4,500K, you could say it's taken from the perspective of an observer who is situated at an luminance level where 4,500K feels correct for white, and higher and lower color temperatures feel too red or too blue respectively. However, the coloring is for artistic effect to highlight the area bounded by the lines and you shouldn't be reading off the color directly, which has no direct meaning.
- However, you got me thinking about the chart, and this is an understandable misinterpretation, so I'm thinking how it can be improved. The area bounded by the lines is actually the area where light appears white, so having it colored at all is misleading. The area outside the lines is where light appears to be wrong color, and it's probably this which merits the artistic coloring, i.e. an inversion of the currently colored area. However, a single gradation along the X-axis and the viewpoint at 4,500K is then inappropriate, given that the chart depicts all possible viewpoints. I'm playing with changing the coloring as a result of your comments.
- How's something like this one: ?
- (It's got a problem with the SVG which stops wikipedia resizing it, which is why I didn't leave it in place.)
- Sadg4000 (talk) 11:23, 18 July 2009 (UTC)
- Maybe if the color gradient was confined to a bar along the axis, and the entire graph was uncolored?
- But there's an even worse problem, IMO. The MR-16 entry is portrayed as a single point. while sunlight may be considered as a point, having both a presumed color temperature and illuminance, a lamp type has only a color temperature. The illuminance varies depending on the distance from the lamp, and on the beam type (VNSP, SP, etc.). What's most disturbing to me is that it seems to have acquired its current presumed illuminance value not as a typical value, but simply to force it within the pleasant zone when the CCT was corrected from 3700K.
- I think we'd be better off without the MR-16 point unless we've got some legitimate data on illuminance, but how about ditching points altogether and marking vertical lines (wax candle, halogen, D65) and horizontal lines (noon sun, office lighting, candlelight in 4m x 4m room, moonlight)? I'd take a crack at it myself if we were talking text, but alas, I've no SVG skills.
- 184.108.40.206 (talk) 19:36, 21 February 2010 (UTC)
I notice someone has reverted the image to the original one with the wrong color temperature for a filament lamp, and a comment about the daylight position which shows they don't understand how the chart works. 220.127.116.11 (talk) 12:03, 3 August 2011 (UTC)
- I've changed the SVG (areas perceived as white/bluish/reddish shaded in the corresponding colors) and removed the tungsten lamp point. Regarding other suggestions made above:
- Color gradient was confined to a bar along the axis -- makes sense, but is a bit tedious to implement in RGB colors, might turn out to be confusing, and creates the new problem of dealing with the luminance- and monitor-dependent white point of the monitor of the viewer.
- vertical lines (wax candle, halogen, D65) and horizontal lines (noon sun, office lighting, candlelight in 4m x 4m room, moonlight)? -- makes sense, but we'd have to agree on the positions of those lines. The lux article mentions 50 lux for a living room, which seems to be awfully dark to me (although my wife might like it. ;-) ) Noon sun illuminance (10^5 K) will not fit on the scale, though.
- Han-Kwang (t) 01:58, 1 January 2012 (UTC)
Kruithof was labeled an "engineer", but he actually was a physicist. After his time at Philips, where he developed fluorescent lighting etc., he was appointed professor of applied physics at Eindhoven University of Technology, researching atomic physics, particularly gas discharges.
What does "pleasing" mean in this context? Which body and light colours are preferred should be subject to personal taste rather than to measurable quantities. Unfortunately, there seems to be no original publication available to ordinary users, so it is not possible to check what Kruithof really meant with this. Could someone with access to Kruithof's PhD thesis or subsequent publication and sufficient Dutch language skills retrieve more information about this?
One could try to understand it in the context of the Purkinje effect, i.e. the shift of the white point for different illuminance levels. However, if whiteness would be the key criterion, 3500-4000 K would be ideal for low light levels since light from type 840 fluorescent lamps appears white, even at levels below 100 lux. Similarly, the Moon has a correlated color temperature of about 4000 K (due to the slightly brownish tone of its surface; otherwise it would have the same CCT as the Sun), and it's light appears dim white, not bluish (however, maybe below the response level of the cones). Incandescent lamps and "warm white" CFLs, however, always have a bias towards a yellowish tone if the the light is cast on a white surface. The Kruithof region of "pleasing" light, however, ranges from about 2500 to 3000 (estimated from the SVG figure), which is well in the yellowish region. At high light levels, however, Kruithof roughly fits, but covers a range from slightly yellowish light to blue skylight. Thus, the Purkinje effect alone cannot explain the Kruithof relation.--SiriusB (talk) 14:22, 9 February 2012 (UTC)
- Good points. I edited the article a while ago to make it internally more consistent, but I don't have access to the original sources either. I do speak Dutch, though, so if you have a link to a document, I can read it. I'm not sure that I agree with your "4000 K at 100 lux appears white" statement. A "cool white" fluorescent light appears rather blueish to me in the living room. But that could be because it is surrounded by 2900 K light sources. My wife definitely thinks that "cool white" looks horrible. Han-Kwang (t) 23:56, 10 February 2012 (UTC)
- Sorry, I don't have any link (otherwise I would either have tried the Google translator or asked a Dutch colleague). I guess that the bluish appearance of your 4000 K lamp might either be due to the contrast to the 2900 K ones or due to a product-specific color characteristic. The link given in the article about museum lighting states that the whit point at 200 lux (20 foot candles) is about 3700 K at average, and my personal feeling is that below a few 100 lux there isn't much more color drift (so the according Purkinje white-point curve in the Kelvin-log(Lux), with log Lux on the x axis, graph between 0 and 100000 lx would look a bit like negative cosine curve between 0 and +π, i.e. flat at the extrema and monotonic rising at intermediate lux levels. BTW to me and some of my colleagues, "warm white" (even from an incandescent lamp) looks horrible and uncomfortable (mainly due to the unpleasant yellowish tint that makes everything look dim and melancholic), but not remotely warm, unless it is powerful enough to feel its heat from a distance. The yellow-reddish color of low-temp lamps reminds of a faint mid-winter Sun near the horizon, struggling hard to penetrate the haze. In contrast, a light bluish tint does not look unpleasant to me, therefore I sometimes even prefer light at 6000 K. According to different sources (including Wikipedia) people in southern Europe or in Asia prefer white to daylight-colored light, while "warm" white is only preferred in the north-western hemisphere. However, there is no consistent explanation to this phenomenon. So, the Kruithof relation, if only based on whether light appeas "pleasing", is obviously far from being a strong and reliable relation like the Purkinje shift, but rather highly dependent on culture and personal preferences. It would be interesting if any critical reviews of Kruithof's work could be retrieved.--SiriusB (talk) 08:33, 13 February 2012 (UTC)
- I could certainly believe that preferences in lighting colors are based on experience. That’s certainly true for many other preferences. I would also expect preferences to vary somewhat depending on activity: reading a book, looking at paintings on a wall, eating dinner, browsing the internet on a computer, or playing a ping pong game all involve different amounts of engagement with foreground and background, and different kinds of visual processing. Much of the literature I can see on Google Scholar relates to museum lighting, which is a rather particular example, and probably depends on the details of art materials. Here are some various links:    . –jacobolus (t) 01:48, 14 February 2012 (UTC)