Re: [ft] putting text in a box
As I told you earlier: A negative bbox value descends below the baseline (which has y=0). How could a bounding box represent both the overall size of the glyph and the descent at the same time? Sorry for being imprecise. A bbox consists of four values: The (x,y) values of the lower left (llx,lly) and the upper right (urx,ury) corner. A box of a glyph with a descender looks like this: (urx,ury) +--+ | | | | | | | | --+--+-- baseline (y = 0) | | +--+ (llx,lly) If I say `a negative bbox value' I've meant that lly 0; the descender is thus simply the value of lly. [...] let's say its height overall is 10 pixels (which completely encloses top-to-bottom the glyph) and the descent is 3 pixels. How can both pieces of information be communicated through a box height? Then you should have lly = -3 and ury = 7. Hmm, I see that the description of FT_BBox doesn't mention the term `baseline'. I've improved it now hopefully by adding this: /* Note*/ /*The bounding box is specified with the coordinates of the lower*/ /*left and the upper right corner. In PostScript, those values are */ /*often called (llx,lly) and (urx,ury), respectively.*/ /* */ /*If `yMin' is negative, this value gives the glyph's descender. */ /*Otherwise, the glyph doesn't descend below the baseline. */ /*Similarly, if `ymax' is positive, this value gives the glyph's */ /*ascender. */ Werner ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
RE: [ft] putting text in a box
Werner, Is it also worth stating what a negative llx (xMin) value means, to facilitate understanding. David %^ -Original Message- From: freetype-bounces+dbevan=emtex@nongnu.org [mailto:freetype- bounces+dbevan=emtex@nongnu.org] On Behalf Of Werner LEMBERG Sent: 21 January 2010 07:42 To: d...@kinematics.com Cc: freetype@nongnu.org Subject: Re: [ft] putting text in a box As I told you earlier: A negative bbox value descends below the baseline (which has y=0). How could a bounding box represent both the overall size of the glyph and the descent at the same time? Sorry for being imprecise. A bbox consists of four values: The (x,y) values of the lower left (llx,lly) and the upper right (urx,ury) corner. A box of a glyph with a descender looks like this: (urx,ury) +--+ | | | | | | | | --+--+-- baseline (y = 0) | | +--+ (llx,lly) If I say `a negative bbox value' I've meant that lly 0; the descender is thus simply the value of lly. [...] let's say its height overall is 10 pixels (which completely encloses top-to-bottom the glyph) and the descent is 3 pixels. How can both pieces of information be communicated through a box height? Then you should have lly = -3 and ury = 7. Hmm, I see that the description of FT_BBox doesn't mention the term `baseline'. I've improved it now hopefully by adding this: /* Note */ /*The bounding box is specified with the coordinates of the lower */ /*left and the upper right corner. In PostScript, those values are */ /*often called (llx,lly) and (urx,ury), respectively. */ /* */ /*If `yMin' is negative, this value gives the glyph's descender. */ /*Otherwise, the glyph doesn't descend below the baseline. */ /*Similarly, if `ymax' is positive, this value gives the glyph's */ /*ascender. */ Werner ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
For reference, a good link is also the below, scroll to the bottom which talks about positioning glyphs on a bitmap of specified size and it also mentions subtracting the lower left corner of the bounding box from the pen pos to get the actual glyph aligned with the box. http://freetype.sourceforge.net/freetype2/docs/glyphs/glyphs-7.html On 1/21/2010 10:12 AM, Dave Calkins wrote: Thanks a ton for the help! Sure enough, my code was considering the overall size from the bbox (upper right - lower left) but I hadn't considered the fact that a negative lower left point was exactly the information I needed :) Live and learn! I'm able to now subtract the lower left x and y from my pen position and the result is the text itself is flush against the desired position. On 1/21/2010 2:42 AM, Werner LEMBERG wrote: As I told you earlier: A negative bbox value descends below the baseline (which has y=0). How could a bounding box represent both the overall size of the glyph and the descent at the same time? Sorry for being imprecise. A bbox consists of four values: The (x,y) values of the lower left (llx,lly) and the upper right (urx,ury) corner. A box of a glyph with a descender looks like this: (urx,ury) +--+ | | | | | | | | --+--+-- baseline (y = 0) | | +--+ (llx,lly) If I say `a negative bbox value' I've meant that lly 0; the descender is thus simply the value of lly. ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
It seems like the bounding box calculation is always correct. The problem is that the descender I'm querying is for the font face in general which I assume accounts for every glyph in the font. If you have a negative value in the bbox, this is the amount below the baseline. I'm not getting negative values in the bounding box. I'm getting a positive overall width and height for the text string in question (although I should say I'm calling the FreeTypeGL bounding box method, so I suppose I need to see which FreeType calls thats using). And I can verify that this bounding box is correct. If I render and check the bounding box, I can, in MSPaint for example, line it up with a box that size and it does exactly enclose all the text. So the bounding box looks great. The problem is that I need to know where to render the text. When I set the output position, this positions the baseline of the text. So the only way to put the text exactly in that box (I think) is to know the descender values for the individual chars in the specific string I'm rendering. I can get a descender value for the font face overall, but this ends up being too large (for some fonts) and if I use this to position the string the string is too high since the specific string doesn't actually have descender values that large. Any suggestions on computing the descenders for individual glyphs? Calculate the bbox for individual glyphs. In most cases, a cbox should be sufficient, BTW. I'll look again today, but I don't recall seeing a descender value for a glyph. The example code I looked at computed a bounding box, telling me what box would enclose the char, but I need the distance the char extends below the baseline. I could just have overlooked something though. I took a look at the glyph metrics from FreeType, but these didn't seem to help. For example, I found a case where E seemed to report a descender (comparing height and vertBearingY). It's easily possible that the glyph `E' descends slightly (due to serifs, for example). In that case, the descent was definitely not from a serif. It was a value much larger than you'd expect. I drew the E, then drew a cross-hair at the output positionn and there was really no descent. ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
Thanks, I'll try both of those. Do you know, though, of a way to get the descent value for an individual glyph reliably for any font? On 1/20/2010 9:51 AM, Werner LEMBERG wrote: If you have a negative value in the bbox, this is the amount below the baseline. I'm not getting negative values in the bounding box. Hmm. You should indeed write a small FreeType test program which uses the FreeType API directly. Then you can compare the results. It's easily possible that the glyph `E' descends slightly (due to serifs, for example). In that case, the descent was definitely not from a serif. It was a value much larger than you'd expect. I drew the E, then drew a cross-hair at the output positionn and there was really no descent. Which font? You might use, say, FontForge to check the glyph's dimensions. Werner ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
Do you know, though, of a way to get the descent value for an individual glyph reliably for any font? Normally, computing the cbox (or the bbox, to be absolutely sure for the weird cases which shouldn't happen with well designed fonts) should do the right thing. Werner ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
Do you know, though, of a way to get the descent value for an individual glyph reliably for any font? Normally, computing the cbox (or the bbox, to be absolutely sure for the weird cases which shouldn't happen with well designed fonts) should do the right thing. I guess this is what's confusing me. I looked at the tutorial code for getting the cbox, for example, and the result of the call is just that, a box. So that would be a box which completely enclosed the glyph right? How does that yield the information I need, namely the distance below the baseline which the glyph extends? If I have a box which encloses the glyph, then I could use that information to figure out how much space I need. However, when I render the text at position X, Y that position sets the baseline of the text. So I'm still stuck figuring out how far below that baseline my glyphs will go. Without knowing that, I can't put text into a box. ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
Do you know, though, of a way to get the descent value for an individual glyph reliably for any font? Normally, computing the cbox (or the bbox, to be absolutely sure for the weird cases which shouldn't happen with well designed fonts) should do the right thing. I guess this is what's confusing me. I looked at the tutorial code for getting the cbox, for example, and the result of the call is just that, a box. So that would be a box which completely enclosed the glyph right? Yes. How does that yield the information I need, namely the distance below the baseline which the glyph extends? As I told you earlier: A negative bbox value descends below the baseline (which has y=0). I have no idea what's going on with GL. There might be even a bug in FreeType, thus my request to test it with native FreeType API calls. A possible reason for GL to disallow negative bbox values is the PostScript rendering model which by default cuts off all coordinates with non-positive values (since they are drawn outside of the `drawing paper' with its physical origin (0,0) at the lower left corner). Werner ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
Re: [ft] putting text in a box
I guess this is what's confusing me. I looked at the tutorial code for getting the cbox, for example, and the result of the call is just that, a box. So that would be a box which completely enclosed the glyph right? Yes. How does that yield the information I need, namely the distance below the baseline which the glyph extends? As I told you earlier: A negative bbox value descends below the baseline (which has y=0). How could a bounding box represent both the overall size of the glyph and the descent at the same time? For example, consider a lower case y. A single value wouldn't be able to communicate both the overall height of that glyph and the descent distance below the baseline would it? In that case let's say its height overall is 10 pixels (which completely encloses top-to-bottom the glyph) and the decent is 3 pixels. How can both pieces of information be communicated through a box height? I have no idea what's going on with GL. There might be even a bug in FreeType, thus my request to test it with native FreeType API calls. A possible reason for GL to disallow negative bbox values is the PostScript rendering model which by default cuts off all coordinates with non-positive values (since they are drawn outside of the `drawing paper' with its physical origin (0,0) at the lower left corner). Even ignoring the rendering piece of it for now, I'm just confused with the metrics I guess. If I have a glyph, lowercase y for example which descends below the baseline, I know I can determine the bounding box for it, but I also need another piece of information. I need the distance from the baseline to the bottom of the glyph, which I think will be less than the bounding box size. As far as GL, when you render it with GL at a given X,Y coordinate, for example, the baseline of the text goes at that X,Y coordinate. So if you want the text string to be flush with a horizontal line, you have to know how far each glyph in the string goes below the baseline so you can shift it up by that amount. Shifting it up by the largest descent in the string would then have the string just touching the horizontal line. I've been able to reliably (so far) get the bounding box, which tells me the overall size of the text, but I can't seem to figure out how to get those descent values for individual glyphs. They would be smaller than the heights of the bounding boxes so I'm thinking there must be another value somewhere. ___ Freetype mailing list Freetype@nongnu.org http://lists.nongnu.org/mailman/listinfo/freetype
