- From: Steve Zilles <szilles@adobe.com>
- Date: Thu, 23 Aug 2007 22:45:56 -0700
- To: Christoph Päper <christoph.paeper@crissov.de>, www-style CSS <www-style@w3.org>
- Message-Id: <6.2.1.2.2.20070823224104.0ac02560@namailhost.corp.adobe.com>
At 03:16 PM 8/23/2007, Christoph Päper wrote: >Paul Nelson: >>The 100 - 900 come from the TrueType font specification. > >Yeah, I thought so, but what happened to "Don't mention the mechanics!"? >Unlike |font-weight|, |font-stretch| already features keywords, >although it lacks the same detail of mapping instructions should a >face not be available. When we were designing the font-weight property in August 1996, we looked at using named weights. The problem was that the usage of named weights was inconsistent. The experience at the time said that the TT value usage was more consistent than that of named weights. From a message sent at that time: As part of an effort to rationalize the handling of font weights with in HTML, I have run a program that we have at Adobe that computes the "density" of the lower case letters in a font face. This is basically the fraction of pixels in a character cell that are black (vs being white) averaged across the whole lowercase alphabet. This value is an estimate of what is called the "color" of a font; that is, it relative intensity on the page. Fonts that have similar densities will look like the have the same blackness (really grayness) when viewing the page at a distance. Fonts with a greater density will stand out as being blacker (thus bolder). The attached PDF file has a table of the major weight values that occur in the Adobe font library. This table has four numeric columns which, for each weight value, has the (1) number of occurances of that value in the library, (2) the maximum density value for (normally) a roman font with that weight, (3) the minimum density value for (normally) a roman font with that weight, (4) the median density value for (normally) a roman font with that weight (the median is slightly fudged to find a more common font that is near the median value). The following three columns (5-7) have the names of the fonts that have the corresponding density values: the lightest font listed first, the typical/median font second and the darkest face last. Below the table is a chart which orders the fonts by increasing median density (more or less) and shows the range of densities (from low to high) as a vertical bar with the median density shown as a mark on the the bar. Below each density bar is a bar whose height reflects the number of faces with the given weight designation. I believe this chart shows that for any given midrange density there are between four and eight weight values that for some font will give that density value. This makes mapping weight names to a numerical scale rather impossible. Steve ===================================== Steve Zilles 115 Lansberry Court, Los Gatos, CA 95032-4710 steve@zilles.org
Attachments
- application/octet-stream attachment: AdobeCatalog.wgtrangeGraph.pdf
Received on Friday, 24 August 2007 05:46:31 UTC