freenode/#clim - IRC Chatlog
Search
13:01:44
beach
Gray-scale anti-aliasing is fine. But I also think I need to align the left and right side of noteheads with the pixel grid so that I can fit stems that are not blurred with those noteheads.
13:03:31
beach
Also, I remember reading that the bytecodes used for TTF hinting have some copyright or patent problems.
13:08:17
beach
When Fontforge renders a glyph, two parameters are used, the resolution and the point size.
13:08:52
beach
But when I create a text style from a font, can I also give the resolution? Or is that some global information?
13:21:19
jackdaniel
scymtym: I think that it would be better to have inspect and source commands on parameters with dedicated panes for both source and inspector
13:40:15
ck_
I think their response would probably be "you can pry these from my cold dead furlongs (or fourty rods)"
13:49:04
scymtym
jackdaniel: ok. in any case, to make any of this work, i think we would need something like pane-local command tables
13:55:10
beach
I'm going to give up these Clovetree experiments for now, because they are extremely frustrating. However, when I pick up Clovetree one day, I will not make the same mistake that I made for Gsharp, i.e. try to handle all the rendering details from the start.
13:55:21
beach
I will concentrate on the mental model of the score and on the user interactions. Initially, I will just use this font as it is, and try to fit stems on noteheads as well as possible.
13:55:22
beach
Improved rendering can come later, and may not even be required on a 4K monitor, or something with the same resolution.
13:57:46
beach
The font software I defined for Gsharp was in many respects superior to what TTF can do, but I neither have the time, the skill, the patience, nor the right software tools to create my own font using that technique.
15:13:00
loke
beach: resolution and point size are different. If you double the resolution as well as the point size, the result is different.
15:15:50
loke
beach: Here's an example shoing how doubling the point size gives you a different resolut compared to scaling the text: https://photos.app.goo.gl/SSPv7JiLr7dZ1QUo6
15:22:31
beach
I know that they are different. I am just saying that they both influence the result.
15:24:05
loke
Yes, but relying on it to manually duge it to the correct dimensions feels quite hackish
15:25:22
beach
Not that I know why those two results are different, but I assume it has to do with hinting or something like that.
15:25:53
beach
I know that Knuth in the Metafont book spent a lot of time explaining why scaling and changing the point size is different.
15:27:17
beach
I was talking about the fact that two parameters are required in Fontforge in order to render a glyph, namely device resolution (in dpi) and point size.
15:27:46
beach
Presumably, if the point size is the same, the same shape should be rendered, except with a different number of pixels.
16:04:46
loke
beach: There are cases where you'd want the low-resolution shape, scaled up. But I'd say they're pretty rare, an din almost all cases you want to have the scale at 1.0 and just use point size to adjust size, so that the you get th ebest looking shape.
18:26:19
scymtym
if already had the input focus thing, the command processor would look in the pane command table of the pane that has the input focus. each pane command table would have to be treated as if it had :inherit-from <frame-command-table>
18:27:59
scymtym
the debugger illustrate the problem: if, for example, you add an interactor, you can't type an "e" in the interactor because that gesture is bound to a command. it would make more sense if the gesture would only behave this way if the debugger pane has the input focus
18:33:18
jackdaniel
I wonder if a similar effect couldn't be achieved with conditional-commands (here is such module in Extensions/ directory)
18:34:01
jackdaniel
I don't know the module well, but afair it was made to make commands enabled only under certain conditions (I can imagine checking input-focus being one such condition)
18:34:56
jackdaniel
that would probably apply only to translators, because frames are responsible for reading commands (i.e there is no read-command loop associated with a particular stream)