null
/
Privacy-Period-Tracker
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
Privacy-Period-Tracker/docs/design/dist/README.md

77 lines
3.9 KiB
Markdown
Raw Normal View History

docs: onboarding illustrations, matched light and dark Seven illustrations in a matched pair, replacing the single light-only concept set added in 9d2a4b2. Named for the composables they replace, and identical in dimensions per card so a theme switch swaps the image without moving the layout. Three corrections happened during review and each is easy to reintroduce by regenerating carelessly, so dist/README.md records them: - No text painted into the pixels. The app renders all seven titles through Heading(...) already; art carrying its own title prints every heading twice, cannot be translated, cannot scale with font size, and is invisible to TalkBack. - No step-number badge. LearningIllustration is also the Insights empty state, where a "4" means nothing, and baked numbers freeze a running order that has already changed once. - 07_Forecast names no fields. One draft listed Fertile window and Ovulation, which the first-forecast screen deliberately withholds — on a fresh install it reports confidence Low and Today reads "Not enough history to estimate". A later draft named the right fields but painted "Confidence — Low" permanently, wrong for anyone with a settled cycle. Naming nothing is the only version that cannot contradict the live forecast. The dark set is a real dark composition rather than the light one dimmed, on the midnight-plum ground BRAND_GUIDE.md §22 specifies. WebP at the delivered resolution: 204 KB for all fourteen against 3.4 MB as PNG. Delivery .zip archives are gitignored; the superseded concept set is removed and remains recoverable at 9d2a4b2. #29 stays open — these are references, and the vectors that ship are drawn from them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:53:23 -05:00
# Onboarding illustrations — the decided direction
docs: add the onboarding concept cards as references, not assets Seven cards, one per onboarding step, supplied by the owner and generated from BRAND_GUIDE.md §22's own master prompt — which is the workflow that section exists to describe. They settle the direction #29 asked for, including all four steps that had no art at all. They are NOT the shipping assets, and docs/design/dist/README.md says so at the point somebody would otherwise drop them into res/drawable: - raster, against a 1.9 MB release APK and §42's ask for vector paths - titles and subtitles painted into the pixels, while the app already renders all seven through Heading(...) — used as-is every title appears twice, and baked text cannot be translated, scaled with font size, or read by TalkBack - light mode only, with dark mode inbound and a preview-pair rule that exists because dark mode was silently broken for all of Batch 01 - card 7 shows a fertile window and ovulation that the first-forecast screen deliberately withholds; on a fresh install it shows confidence Low and Today reads "Not enough history to estimate" Converted from the delivered PNGs to WebP at the original 1086x1448: same resolution, 340 KB instead of 12 MB, and the format docs/data/img already uses. Renamed from the generator's default filenames so the set reads as a sequence. #29 stays open with the assessment recorded as a comment on it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:28:50 -05:00
```
Status: Current
Owner: _null
Last reviewed: 2026-08-18
docs: onboarding illustrations, matched light and dark Seven illustrations in a matched pair, replacing the single light-only concept set added in 9d2a4b2. Named for the composables they replace, and identical in dimensions per card so a theme switch swaps the image without moving the layout. Three corrections happened during review and each is easy to reintroduce by regenerating carelessly, so dist/README.md records them: - No text painted into the pixels. The app renders all seven titles through Heading(...) already; art carrying its own title prints every heading twice, cannot be translated, cannot scale with font size, and is invisible to TalkBack. - No step-number badge. LearningIllustration is also the Insights empty state, where a "4" means nothing, and baked numbers freeze a running order that has already changed once. - 07_Forecast names no fields. One draft listed Fertile window and Ovulation, which the first-forecast screen deliberately withholds — on a fresh install it reports confidence Low and Today reads "Not enough history to estimate". A later draft named the right fields but painted "Confidence — Low" permanently, wrong for anyone with a settled cycle. Naming nothing is the only version that cannot contradict the live forecast. The dark set is a real dark composition rather than the light one dimmed, on the midnight-plum ground BRAND_GUIDE.md §22 specifies. WebP at the delivered resolution: 204 KB for all fourteen against 3.4 MB as PNG. Delivery .zip archives are gitignored; the superseded concept set is removed and remains recoverable at 9d2a4b2. #29 stays open — these are references, and the vectors that ship are drawn from them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:53:23 -05:00
Governs: docs/design/dist/** — the onboarding illustration set and what it is for
Review trigger: Any illustration added or replaced here; the onboarding artwork
docs: add the onboarding concept cards as references, not assets Seven cards, one per onboarding step, supplied by the owner and generated from BRAND_GUIDE.md §22's own master prompt — which is the workflow that section exists to describe. They settle the direction #29 asked for, including all four steps that had no art at all. They are NOT the shipping assets, and docs/design/dist/README.md says so at the point somebody would otherwise drop them into res/drawable: - raster, against a 1.9 MB release APK and §42's ask for vector paths - titles and subtitles painted into the pixels, while the app already renders all seven through Heading(...) — used as-is every title appears twice, and baked text cannot be translated, scaled with font size, or read by TalkBack - light mode only, with dark mode inbound and a preview-pair rule that exists because dark mode was silently broken for all of Batch 01 - card 7 shows a fertile window and ovulation that the first-forecast screen deliberately withholds; on a fresh install it shows confidence Low and Today reads "Not enough history to estimate" Converted from the delivered PNGs to WebP at the original 1086x1448: same resolution, 340 KB instead of 12 MB, and the format docs/data/img already uses. Renamed from the generator's default filenames so the set reads as a sequence. #29 stays open with the assessment recorded as a comment on it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:28:50 -05:00
in issue #29 landing in the app
```
docs: onboarding illustrations, matched light and dark Seven illustrations in a matched pair, replacing the single light-only concept set added in 9d2a4b2. Named for the composables they replace, and identical in dimensions per card so a theme switch swaps the image without moving the layout. Three corrections happened during review and each is easy to reintroduce by regenerating carelessly, so dist/README.md records them: - No text painted into the pixels. The app renders all seven titles through Heading(...) already; art carrying its own title prints every heading twice, cannot be translated, cannot scale with font size, and is invisible to TalkBack. - No step-number badge. LearningIllustration is also the Insights empty state, where a "4" means nothing, and baked numbers freeze a running order that has already changed once. - 07_Forecast names no fields. One draft listed Fertile window and Ovulation, which the first-forecast screen deliberately withholds — on a fresh install it reports confidence Low and Today reads "Not enough history to estimate". A later draft named the right fields but painted "Confidence — Low" permanently, wrong for anyone with a settled cycle. Naming nothing is the only version that cannot contradict the live forecast. The dark set is a real dark composition rather than the light one dimmed, on the midnight-plum ground BRAND_GUIDE.md §22 specifies. WebP at the delivered resolution: 204 KB for all fourteen against 3.4 MB as PNG. Delivery .zip archives are gitignored; the superseded concept set is removed and remains recoverable at 9d2a4b2. #29 stays open — these are references, and the vectors that ship are drawn from them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:53:23 -05:00
> **References, not shipping assets.** Nothing in this folder goes in the APK.
> These are the decided direction for onboarding, drawn from
> [`../BRAND_GUIDE.md`](../BRAND_GUIDE.md) §22's master prompt. The vectors that
> eventually ship are drawn *from* these.
docs: add the onboarding concept cards as references, not assets Seven cards, one per onboarding step, supplied by the owner and generated from BRAND_GUIDE.md §22's own master prompt — which is the workflow that section exists to describe. They settle the direction #29 asked for, including all four steps that had no art at all. They are NOT the shipping assets, and docs/design/dist/README.md says so at the point somebody would otherwise drop them into res/drawable: - raster, against a 1.9 MB release APK and §42's ask for vector paths - titles and subtitles painted into the pixels, while the app already renders all seven through Heading(...) — used as-is every title appears twice, and baked text cannot be translated, scaled with font size, or read by TalkBack - light mode only, with dark mode inbound and a preview-pair rule that exists because dark mode was silently broken for all of Batch 01 - card 7 shows a fertile window and ovulation that the first-forecast screen deliberately withholds; on a fresh install it shows confidence Low and Today reads "Not enough history to estimate" Converted from the delivered PNGs to WebP at the original 1086x1448: same resolution, 340 KB instead of 12 MB, and the format docs/data/img already uses. Renamed from the generator's default filenames so the set reads as a sequence. #29 stays open with the assessment recorded as a comment on it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:28:50 -05:00
docs: onboarding illustrations, matched light and dark Seven illustrations in a matched pair, replacing the single light-only concept set added in 9d2a4b2. Named for the composables they replace, and identical in dimensions per card so a theme switch swaps the image without moving the layout. Three corrections happened during review and each is easy to reintroduce by regenerating carelessly, so dist/README.md records them: - No text painted into the pixels. The app renders all seven titles through Heading(...) already; art carrying its own title prints every heading twice, cannot be translated, cannot scale with font size, and is invisible to TalkBack. - No step-number badge. LearningIllustration is also the Insights empty state, where a "4" means nothing, and baked numbers freeze a running order that has already changed once. - 07_Forecast names no fields. One draft listed Fertile window and Ovulation, which the first-forecast screen deliberately withholds — on a fresh install it reports confidence Low and Today reads "Not enough history to estimate". A later draft named the right fields but painted "Confidence — Low" permanently, wrong for anyone with a settled cycle. Naming nothing is the only version that cannot contradict the live forecast. The dark set is a real dark composition rather than the light one dimmed, on the midnight-plum ground BRAND_GUIDE.md §22 specifies. WebP at the delivered resolution: 204 KB for all fourteen against 3.4 MB as PNG. Delivery .zip archives are gitignored; the superseded concept set is removed and remains recoverable at 9d2a4b2. #29 stays open — these are references, and the vectors that ship are drawn from them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:53:23 -05:00
Seven illustrations in a matched light/dark pair, named for the composables they
replace in
[`../../../core/designsystem/src/main/kotlin/dev/privacyllc/period/designsystem/art/Illustrations.kt`](../../../core/designsystem/src/main/kotlin/dev/privacyllc/period/designsystem/art/Illustrations.kt):
docs: add the onboarding concept cards as references, not assets Seven cards, one per onboarding step, supplied by the owner and generated from BRAND_GUIDE.md §22's own master prompt — which is the workflow that section exists to describe. They settle the direction #29 asked for, including all four steps that had no art at all. They are NOT the shipping assets, and docs/design/dist/README.md says so at the point somebody would otherwise drop them into res/drawable: - raster, against a 1.9 MB release APK and §42's ask for vector paths - titles and subtitles painted into the pixels, while the app already renders all seven through Heading(...) — used as-is every title appears twice, and baked text cannot be translated, scaled with font size, or read by TalkBack - light mode only, with dark mode inbound and a preview-pair rule that exists because dark mode was silently broken for all of Batch 01 - card 7 shows a fertile window and ovulation that the first-forecast screen deliberately withholds; on a fresh install it shows confidence Low and Today reads "Not enough history to estimate" Converted from the delivered PNGs to WebP at the original 1086x1448: same resolution, 340 KB instead of 12 MB, and the format docs/data/img already uses. Renamed from the generator's default filenames so the set reads as a sequence. #29 stays open with the assessment recorded as a comment on it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:28:50 -05:00
docs: onboarding illustrations, matched light and dark Seven illustrations in a matched pair, replacing the single light-only concept set added in 9d2a4b2. Named for the composables they replace, and identical in dimensions per card so a theme switch swaps the image without moving the layout. Three corrections happened during review and each is easy to reintroduce by regenerating carelessly, so dist/README.md records them: - No text painted into the pixels. The app renders all seven titles through Heading(...) already; art carrying its own title prints every heading twice, cannot be translated, cannot scale with font size, and is invisible to TalkBack. - No step-number badge. LearningIllustration is also the Insights empty state, where a "4" means nothing, and baked numbers freeze a running order that has already changed once. - 07_Forecast names no fields. One draft listed Fertile window and Ovulation, which the first-forecast screen deliberately withholds — on a fresh install it reports confidence Low and Today reads "Not enough history to estimate". A later draft named the right fields but painted "Confidence — Low" permanently, wrong for anyone with a settled cycle. Naming nothing is the only version that cannot contradict the live forecast. The dark set is a real dark composition rather than the light one dimmed, on the midnight-plum ground BRAND_GUIDE.md §22 specifies. WebP at the delivered resolution: 204 KB for all fourteen against 3.4 MB as PNG. Delivery .zip archives are gitignored; the superseded concept set is removed and remains recoverable at 9d2a4b2. #29 stays open — these are references, and the vectors that ship are drawn from them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:53:23 -05:00
| File stem | Onboarding step | Composable |
docs: add the onboarding concept cards as references, not assets Seven cards, one per onboarding step, supplied by the owner and generated from BRAND_GUIDE.md §22's own master prompt — which is the workflow that section exists to describe. They settle the direction #29 asked for, including all four steps that had no art at all. They are NOT the shipping assets, and docs/design/dist/README.md says so at the point somebody would otherwise drop them into res/drawable: - raster, against a 1.9 MB release APK and §42's ask for vector paths - titles and subtitles painted into the pixels, while the app already renders all seven through Heading(...) — used as-is every title appears twice, and baked text cannot be translated, scaled with font size, or read by TalkBack - light mode only, with dark mode inbound and a preview-pair rule that exists because dark mode was silently broken for all of Batch 01 - card 7 shows a fertile window and ovulation that the first-forecast screen deliberately withholds; on a fresh install it shows confidence Low and Today reads "Not enough history to estimate" Converted from the delivered PNGs to WebP at the original 1086x1448: same resolution, 340 KB instead of 12 MB, and the format docs/data/img already uses. Renamed from the generator's default filenames so the set reads as a sequence. #29 stays open with the assessment recorded as a comment on it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:28:50 -05:00
| --- | --- | --- |
docs: onboarding illustrations, matched light and dark Seven illustrations in a matched pair, replacing the single light-only concept set added in 9d2a4b2. Named for the composables they replace, and identical in dimensions per card so a theme switch swaps the image without moving the layout. Three corrections happened during review and each is easy to reintroduce by regenerating carelessly, so dist/README.md records them: - No text painted into the pixels. The app renders all seven titles through Heading(...) already; art carrying its own title prints every heading twice, cannot be translated, cannot scale with font size, and is invisible to TalkBack. - No step-number badge. LearningIllustration is also the Insights empty state, where a "4" means nothing, and baked numbers freeze a running order that has already changed once. - 07_Forecast names no fields. One draft listed Fertile window and Ovulation, which the first-forecast screen deliberately withholds — on a fresh install it reports confidence Low and Today reads "Not enough history to estimate". A later draft named the right fields but painted "Confidence — Low" permanently, wrong for anyone with a settled cycle. Naming nothing is the only version that cannot contradict the live forecast. The dark set is a real dark composition rather than the light one dimmed, on the midnight-plum ground BRAND_GUIDE.md §22 specifies. WebP at the delivered resolution: 204 KB for all fourteen against 3.4 MB as PNG. Delivery .zip archives are gitignored; the superseded concept set is removed and remains recoverable at 9d2a4b2. #29 stays open — these are references, and the vectors that ship are drawn from them. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-08-18 19:53:23 -05:00
| `01_WelcomeIllustration` | 1 Welcome | `WelcomeIllustration` |
| `02_LastPeriodIllustration` | 2 Last period start | *new* |
| `03_PeriodEndIllustration` | 3 Period end | *new* |
| `04_LearningIllustration` | 4 Previous history | `LearningIllustration` |
| `05_PrivacyIllustration` | 5 Privacy promise | `PrivacyIllustration` |
| `06_NotificationPrivacyIllustration` | 6 Reminder privacy | *new* |
| `07_ForecastIllustration` | 7 First forecast | *new* |
Each stem exists as `Light/<stem>_light.webp` and `Dark/<stem>_dark.webp`, at
**identical dimensions per card**, so a theme switch swaps the image without
moving the layout.
## What these got right, and why it is worth not undoing
Each of these was a correction during review, and every one of them would be easy
to reintroduce by regenerating carelessly:
- **No text of any kind is painted into the pixels.** The app renders all seven
titles and subtitles itself through `Heading(...)` in `OnboardingScreen.kt`.
Art carrying its own title prints every heading twice, cannot be translated,
cannot scale with the user's font size, and is invisible to TalkBack.
- **No step-number badge.** `LearningIllustration` is not only onboarding step 4
— it is also the Insights empty state, where a "4" means nothing. Baked numbers
also freeze the running order, and this flow has already changed shape once. A
step indicator, if wanted, belongs in live UI where TalkBack can read it.
- **`07_Forecast` shows a calendar and a cycle arc, and names no fields.** An
earlier draft drew a forecast panel listing *Fertile window* and *Ovulation*,
which the real first-forecast screen deliberately does not show — on a fresh
install it reports confidence Low and Today reads "Not enough history to
estimate". A later draft named the correct fields but painted *Confidence —
Low* permanently, which is wrong for anyone with a settled cycle. Naming
nothing is the only version that cannot contradict the live forecast.
- **The dark set is a real dark composition**, not the light one dimmed — the
midnight-plum ground `BRAND_GUIDE.md` §22 specifies.
## Still to do before anything ships
Tracked as **#29**, which stays open until the app renders these:
1. Redraw as Compose vector paths behind the existing replaceable names, taking
colour from theme tokens rather than shipping two raster sets.
2. Add the light/dark preview pair per step, per [`../README.md`](../README.md).
3. Reconcile step copy — `PRODUCT_PLAN.md` owns the screen wording.
`EmptyStateIllustration`, the Today "nothing logged yet" state, has no card here.
Onboarding would otherwise ship real art while Today keeps a placeholder visible
in the same session.
## Format
WebP at the delivered resolution — 204 KB for all fourteen, against 3.4 MB as
PNG, and the same format [`../../data/img/`](../../data/img/) uses for the marks
privacyllc.dev renders. The delivery `.zip` archives are deliberately not
tracked; see [`.gitignore`](.gitignore).