Open jvgaultney opened 4 years ago
I don't think users need V at W/F to mathematically flush with the bounding box (whether C or asp), but they did indicate a desire to reduce the gap so that at least visually one could still consider it flush. In the figure above, if we vertically translate V at F upward to S, there'll still be a bit too large a gap between C/asp & V--now this may have to do with this particular C & V. If we can reduce the gap to that between C & asp (ideally set 2 combo 2), that would be ideal, for then even when you vertically translate V back down to F, you won't notice the gap so readily.
I don't think users need V at W/F to mathematically flush with the bounding box (whether C or asp), but they did indicate a desire to reduce the gap so that at least visually one could still consider it flush.
That is reasonable and I'll see if there's anything I can do to improve it, though maybe not for 1.0. What is difficult is that the visual effect is related to the design of the particular V. Context-specific kerning of particular V will definitely have to wait.
In the figure above, if we vertically translate V at F upward to S, there'll still be a bit too large a gap between C/asp & V--now this may have to do with this particular C & V. If we can reduce the gap to that between C & asp (ideally set 2 combo 2), that would be ideal, for then even when you vertically translate V back down to F, you won't notice the gap so readily.
What you're seeing is an illusion. If you look at the following combinations it may seem that the alignment at S and F differ:
However when you overlay one on another you see that S and F are actually perfectly aligned (sorry - these two images are not at the same scale, but do show the proper alignment):
In the following example the spacing of the V seems consistent in all but the final C+asp+V. In that last one the V seems to be too far right. But pulling that in will not just affect that particular combo - it will make everything closer, as combos 2, 4, and 6 in the second example shows:
What I will do is reduce the space to right of all asps overall. That may cause a few tight situations where asp+V will be closer than C+asp, but that may be less noticeable.
I've now reduced the space after asp. Results in build 274:
Comparing the above with this image (also shown in #32): it looks like we do need context-specific kerning (or even mathematical flushing?), but of course I agree that will be post-1.0.
What's in the font is almost identical to your example:
The only difference is that Shimenkan, in general, is not spaced as tightly as your drawing. The asp is never that close to the consonant. The relative proportions of space between C, asp, V are the same as in your drawing.
Maybe this can help:
(Moved from issue #75 )
cheuk879 wrote:
mhosken responded: