2024-08-06 05:14:57 +00:00
|
|
|
# Theme constraints
|
2024-08-06 05:36:26 +00:00
|
|
|
The following general constraints are followed as palette options are mapped onto
|
2024-08-06 05:14:57 +00:00
|
|
|
concrete themes:
|
|
|
|
|
|
|
|
+ Harshness levels have monotone differences of a single shade.
|
|
|
|
+ "Hard" themes anchor their background to the most extreme shade appropriate for the
|
|
|
|
scheme (i.e., lightest shade for "light," darkest shade for "dark"), ensuring the
|
|
|
|
palette's "monotone width" is fully spanned by the theme options.
|
|
|
|
+ App-specific monotone settings have differences of a single shade compared to the
|
|
|
|
system monotone settings.
|
|
|
|
+ Shade differences between corresponding background/foreground settings should be
|
|
|
|
constant (e.g., between `bg0` and `fg3`, `bg1` and `fg2`, etc)
|
|
|
|
|
|
|
|
The primary goal of these constraints is to ensure each theme in a collection defined
|
|
|
|
around a single palette is 1) sufficiently _distinct_, 2) attains sufficient _breadth_
|
|
|
|
under the palette, and 3) upholds _relative invariance_ under key properties (e.g.,
|
|
|
|
lightness differences).
|
|
|
|
|
2024-08-04 11:41:53 +00:00
|
|
|
![How themes are created](images/theme_creation.png)
|
2024-08-06 05:14:57 +00:00
|
|
|
|
|
|
|
## Example
|
|
|
|
The following is a natural solution to these constraints, demonstrated on a general
|
|
|
|
example setting: a possible useful analogy is a sliding window that, on its own spans a
|
|
|
|
given theme's `bg0-fg0` settings, while globally sliding across all available values
|
|
|
|
in the palette. If associating integers `0-10` to indices in a list of monotone shades, and
|
|
|
|
`bg-fg` is the syntax used to indicate that theme's shade range, we might have the
|
|
|
|
following for dark mode themes across harshness levels:
|
|
|
|
|
|
|
|
```
|
|
|
|
Dark (system) 0-7 ; 1-8 ; 2-9
|
|
|
|
Dark (app) 1-8 ; 2-9 ; 3-10
|
|
|
|
```
|
|
|
|
|
|
|
|
There are sliding windows at both the system-app level *and* the harshness-level, in a
|
|
|
|
sense. Constraints are followed:
|
|
|
|
|
|
|
|
+ Harshness levels, separated by semicolon, differ by a single shade from hard to soft.
|
|
|
|
+ The hard theme anchors its background to the darkest available shade.
|
|
|
|
+ Monotones between system and app differ by a single shade.
|
|
|
|
+ Differences between bg/fg (value of 7) remains constant across all themes.
|
|
|
|
|
|
|
|
Mapping this onto the common values used in my theme definition files:
|
|
|
|
|
|
|
|
```
|
|
|
|
System, dark
|
|
|
|
| Hard | Regular | Soft
|
|
|
|
| bg0 <- l15 | bg0 <- l20 | bg0 <- l25
|
|
|
|
| fg0 <- l80 | fg0 <- l85 | fg0 <- l90
|
|
|
|
|
|
|
|
App, dark
|
|
|
|
| Hard | Regular | Soft
|
|
|
|
| bg0 <- l20 | bg0 <- l25 | bg0 <- l30
|
|
|
|
| fg0 <- l85 | fg0 <- l90 | fg0 <- l95
|
|
|
|
|
|
|
|
System, light
|
|
|
|
| Hard | Regular | Soft
|
|
|
|
| bg0 <- l95 | bg0 <- l90 | bg0 <- l85
|
|
|
|
| fg0 <- l30 | fg0 <- l25 | fg0 <- l20
|
|
|
|
|
|
|
|
App, light
|
|
|
|
| Hard | Regular | Soft
|
|
|
|
| bg0 <- l90 | bg0 <- l85 | bg0 <- l80
|
|
|
|
| fg0 <- l25 | fg0 <- l20 | fg0 <- l15
|
|
|
|
```
|
|
|
|
|
|
|
|
# Accent contrast
|
|
|
|
Each group of biome monotones have nearly identical (WCAG 2) contrast ratios against
|
|
|
|
white/black for all lightness levels (ratios identical between biomes). These are
|
|
|
|
selected in a heavily constrained OKLCH context, and given the perceptual uniformity
|
|
|
|
attached to lightness, we can expect very similar contrast ratios for each accent under
|
|
|
|
a given biome lightness (e.g., the `l65` red tone will have the same ratio under the
|
|
|
|
grassland, tundra, and savanna monotones).
|
|
|
|
|
|
|
|
In terms of selecting accents for themes (by harshness and scheme), what matters is
|
|
|
|
at what lightness level all accent colors meet/exceed a particular contrast threshold.
|
|
|
|
Again, the ratios themselves are effectively constant across biome monotones, and thus
|
|
|
|
dependent entirely on the monotone lightness being used. This of course is determined
|
|
|
|
primarily by whether the theme is a light or dark one, and what level of harshness is
|
|
|
|
being used. The following are the relevant values for making a decision. We want to
|
|
|
|
ensure all accents can reach >4.5 WCAG 2 contrast ratio (the standard requirement for
|
|
|
|
small text on the web) against all biome monotones for each theme:
|
|
|
|
|
|
|
|
+ For BG l20 (harsh, dark) -> l65 is min lightness where all accents have CR >=4.5
|
|
|
|
+ For BG l25 (regular, dark) -> l65 is min lightness where all accents have CR >=4.5
|
|
|
|
+ For BG l30 (soft, dark) -> l70 is min lightness where all accents have CR >=4.5
|
|
|
|
|
|
|
|
+ For BG l90 (harsh, dark) -> l45 is max lightness where all accents have CR >=4.5
|
|
|
|
+ For BG l85 (regular, dark) -> l45 is min lightness where all accents have CR >=4.5
|
|
|
|
+ For BG l80 (soft, dark) -> l40 is min lightness where all accents have CR >=4.5
|
|
|
|
|
|
|
|
For the monotone boundaries (l15 and l95, neither of which are possible backgrounds for
|
|
|
|
terminal or nvim in the current theme definitions), the relevant lightness levels are
|
|
|
|
l60 and l50, respectively.
|
|
|
|
|
|
|
|
While not necessary, it feels intuitive for us to shift the accent colors up/down by the
|
|
|
|
relative change in monotones across harshness levels. This has led to the choice of l60
|
|
|
|
accents for the harsh-dark theme, l65 for regular-dark, and l70 for soft-dark. This
|
|
|
|
technically breaks the 4.5 ratio requirement, though, for the harsh theme, so you
|
|
|
|
ultimately need to pick one: either soften the contrast constraint, or allow different
|
|
|
|
harshness levels to use the same accent lightness. I think either is acceptable, but for
|
|
|
|
now I've gone with the former, loosening the contrast to a ratio of >4.0 with respect to
|
|
|
|
the background. This allows for the slightly tighter group of accent lightnesses:
|
|
|
|
l45-l50-l55 for light, l60-l65-l70 for dark. Note that the "center shade" of the l15-l95
|
|
|
|
shade group is l55, meaning these groups are very central (the light triplet could move
|
|
|
|
down by one shade step, but we want these accents to be as bright as we can get away
|
|
|
|
with; otherwise, they are extremely dull in the light modes, and we thus don't mind bias
|
|
|
|
toward a brighter lightness).
|
|
|
|
|