Switch knob: scanline model from 42 measured sizes

Integer radius (h-1)//2 with fixed sub-pixel offsets, 4 sub-rows x 1/8 px spans, 1/32 coverage
in linear light; closer to the simulator than the previous distance model on the knob test page.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
mangavalkandClaude Opus 5.5 committed 2026-11-16 13:52:40 +01:00
1 parent 0138a0ba56
commit 922ca21f0f
3 files changed
+35 -4

No files matched your search

+1 -1
View File
@@ -94,7 +94,7 @@ byte for byte, or a file written by nxhmi was opened in the Editor and showed th
| Item | Notes | | Item | Notes |
|---|---| |---|---|
| Other display series | Discovery / Enhanced / Edge not sampled; only Basic + Intelligent templates | | Other display series | Discovery / Enhanced / Edge not sampled; only Basic + Intelligent templates |
| Display direction | byte 12 decoded and written (`new_project(portrait=...)`); picture order verified for all four (format 4); font glyph order for 90° only | | Display direction | byte 12 decoded and written (`new_project(portrait=...)`); picture order verified for all four (format 4); font glyph order verified for 0° and 90° |
| Some header fields | main.HMI bytes 8, 13, 36, 0x41; page header 21/23; entry bytes 25-27 | | Some header fields | main.HMI bytes 8, 13, 36, 0x41; page header 21/23; entry bytes 25-27 |
| Font name shown in the Editor's Fonts panel | cosmetic, generated fonts show an odd label | | Font name shown in the Editor's Fonts panel | cosmetic, generated fonts show an odd label |
| Media resources | video / audio / Gmov files themselves (not just the components) | | Media resources | video / audio / Gmov files themselves (not just the components) |
+6 -1
View File
@@ -47,7 +47,7 @@ TextSelects, several of each other type):
| FileBrowser | 100 % | frame and fill; the file list only exists at run time | | FileBrowser | 100 % | frame and fill; the file list only exists at run time |
| ExPicture, video, audio, Gmov, FileStream | nothing drawn (same as the simulator) | they show content only once a file is loaded | | ExPicture, video, audio, Gmov, FileStream | nothing drawn (same as the simulator) | they show content only once a file is loaded |
| ComboBox | ~99 % | structure exact; anti-aliased corners and arrow approximate | | ComboBox | ~99 % | structure exact; anti-aliased corners and arrow approximate |
| Switch | ~88-97 % | structure exact; anti-aliased knob and track edges approximate | | Switch | ~88-97 % | structure exact; anti-aliased knob and track edges approximate (see below) |
## Rules learned from the Editor ## Rules learned from the Editor
@@ -70,6 +70,11 @@ TextSelects, several of each other type):
`[h, w)` (off) or `[0, w-h)` (on). ComboBox: text placed by `xcen`/`ycen` over the whole box, `[h, w)` (off) or `[0, w-h)` (on). ComboBox: text placed by `xcen`/`ycen` over the whole box,
`pco3` arrow `wid` px wide 4.5 px from the right edge. Corner radius for both: `dis * h // 200`, `pco3` arrow `wid` px wide 4.5 px from the right edge. Corner radius for both: `dis * h // 200`,
square below 3 px; rounded shapes are inset half a pixel (outer ring 50 %, blended linearly). square below 3 px; rounded shapes are inset half a pixel (outer ring 50 %, blended linearly).
- Switch knob (measured on 42 sizes): radius `r = (h-1)//2` with fixed sub-pixel offsets, edge
coverage in steps of exactly 1/32 blended in linear light; the top row always holds a single
1/4-row span, so it is a scanline rasteriser (4 sub-rows x 1/8 px). The best model found
(centre `r + 0.625, r + 0.25`, radius `r - 0.25`, span ends rounded to 1/8) gets most edge pixels
exact but the span widths curve more sharply than a circle, so the real routine is still open.
- Intelligent pictures with alpha (format 4, 7-bit alpha `a`): 8-bit alpha is `2*a` (127 = 255), - Intelligent pictures with alpha (format 4, 7-bit alpha `a`): 8-bit alpha is `2*a` (127 = 255),
blended as `bg + round((src - bg) * alpha / 255)` per channel. Exact against the Intelligent blended as `bg + round((src - bg) * alpha / 255)` per channel. Exact against the Intelligent
simulator (342 samples). The Editor *canvas* blends slightly differently (up to 3 levels off at simulator (342 samples). The Editor *canvas* blends slightly differently (up to 3 levels off at
+28 -2
View File
@@ -624,6 +624,31 @@ class Renderer:
self._aa_fill(img, (int(cx - r) - 1, int(cy - r) - 1, int(cx + r) + 2, int(cy + r) + 2), col, self._aa_fill(img, (int(cx - r) - 1, int(cy - r) - 1, int(cx + r) + 2, int(cy + r) + 2), col,
lambda px, py: math.hypot(px - cx, py - cy) - r, gamma=True) lambda px, py: math.hypot(px - cx, py - cy) - r, gamma=True)
def _scan_disc(self, img, cx, cy, R, col):
"""Filled circle the way the Intelligent simulator's switch knob looks: 4 sub-rows per pixel
row, each a span with ends rounded to 1/8 px, so coverage comes in steps of 1/32, blended
in linear light. Closest model found (most edge pixels exact, the rest a few levels off)."""
px = img.load()
W, H = img.size
for yy in range(max(0, int(cy - R) - 1), min(H, int(cy + R) + 2)):
cov = {}
for j in range(4):
d = R * R - (yy + j / 4 - cy) ** 2
if d <= 0:
continue
hw = math.sqrt(d)
l, r = round((cx - hw) * 8) / 8, round((cx + hw) * 8) / 8
for xx in range(int(math.floor(l)), int(math.ceil(r))):
o = min(xx + 1, r) - max(xx, l)
if o > 0:
cov[xx] = cov.get(xx, 0) + round(o * 8)
for xx, k in cov.items():
if 0 <= xx < W:
if k >= 32:
px[xx, yy] = col
else:
px[xx, yy] = tuple(_enc(_lin(f) * k / 32 + _lin(b) * (1 - k / 32)) for f, b in zip(col, px[xx, yy][:3]))
def switch(self, img, c): def switch(self, img, c):
"""Switch: rounded track (corner radius dis % of h/2) in bco / bco2 (on), knob pco / pco2 """Switch: rounded track (corner radius dis % of h/2) in bco / bco2 (on), knob pco / pco2
on the left (off) or right (on), text (txt "off/on") in pco1 centred on the free side.""" on the left (off) or right (on), text (txt "off/on") in pco1 centred on the free side."""
@@ -631,8 +656,9 @@ class Renderer:
on = bool(c.get("val")) on = bool(c.get("val"))
r = _radius(c, h) r = _radius(c, h)
self._round_rect(img, x, y, w, h, r, self.rgb((c.get("bco2") if on else c.get("bco")) or 0)) self._round_rect(img, x, y, w, h, r, self.rgb((c.get("bco2") if on else c.get("bco")) or 0))
kx = x + (w - h / 2 + 0.5625 if on else h / 2 - 0.4375) r = (h - 1) // 2 # knob: integer radius, fixed sub-pixel offsets
self._disc(img, kx, y + h / 2 - 0.8125, h / 2 - 1.3125, self.rgb((c.get("pco2") if on else c.get("pco")) or 0)) kx = x + (w - r - 0.625 if on else r + 0.625)
self._scan_disc(img, kx, y + r + 0.25, r - 0.25, self.rgb((c.get("pco2") if on else c.get("pco")) or 0))
parts = (c.get("txt") or "").split("/") parts = (c.get("txt") or "").split("/")
label = (parts[1] if len(parts) > 1 else "") if on else parts[0] label = (parts[1] if len(parts) > 1 else "") if on else parts[0]
tx, tw = (x, w - h) if on else (x + h, w - h) tx, tw = (x, w - h) if on else (x + h, w - h)