Focus-quality (HFD) indicator on the Focus screen#449
Merged
Conversation
Self-contained Half-Flux Diameter detector (PiFinder/focus.py) plus a focus strip in UIPreview: log-axis V-curve over a 10s window, best-focus marker, past-best "BACK UP" cue, and a background-anchored display stretch replacing per-frame autocontrast. The detector runs in the main process on the raw frame and does not depend on plate solving, so it works across the full defocus range. See docs/adr/0005-focus-hfd-self-contained-in-ui.md and the "Focus indicator" section of docs/ax/ui/CONTEXT.md. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Reframe "Setting Focus & First Solve" around the HFD readout and its V-curve rather than zooming in to judge stars by eye. Add a focus-strip walkthrough and HUD example images (focused vs unfocused, with the strip composited over the existing example frames). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…drop back-up cue
- HFD readout is now a large, right-justified number filling the strip height
(the hero element); the V-curve and small labels move to the freed left
region. The no-reading states fall back to a small dim hint ("—" / "keep
going") instead of a giant placeholder glyph.
- V-curve axis is 4 → 20 px (was 1 → 50): ~4 px is about the best a real
camera/lens hits and ~20 px is clearly soft, so the trend spends its range
where it's useful. Out-of-range readings clamp; the numeric readout is exact.
- Display stretch is calmer: lower EMA alpha + larger minimum span cut the
frame-to-frame brightness swings, and a little uniform dither breaks the
8-bit banding a narrow stretch otherwise posterised into.
- Remove the past-best "BACK UP" cue and its supporting machinery.
- Update the Quick Start prose and the UI CONTEXT glossary to match.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The HFD readout and V-curve now carry the focus workflow, so the central concentric-circle reticle no longer adds anything on the Focus screen. - Drop draw_reticle() and its zoom-0 call site, plus the unused reticle_mode attribute and the orphaned camera_reticle config option (no settings UI). - Fix a stale comment: the display stretch is numpy + dither, not a LUT. - Note in the Positioning doc that solve_pixel(screen_space=True) now has no UI consumer (the reticle was its only one); the accessor stays. The Chart/Align Telrad reticle (chart_reticle) is unaffected. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Resolves conflicts from main's #429 solver/integrator dataclass refactor: - preview.py: keep the reticle removal + focus-strip work; port _matched_star_text() to the new PointingEstimate API (solution.solve_source / .estimate_time / .diagnostics.Matches), mirroring main's own refactor of that logic. - docs/ax/positioning.md: take main's solve_pixel -> target_pixel rename, keep the note that the Focus-screen reticle (its only consumer) is gone. - docs/source/quick_start.rst: keep the focus-strip documentation, adopt main's leaner opening/closing prose and its +/- zoom tip. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
brickbots
added a commit
that referenced
this pull request
Jun 5, 2026
The focus-quality (HFD) strip was added in #449 before the resolution-flexible UI work in #453. #453 converted the camera preview's image resize and star selectors to derive from the display resolution but never touched draw_focus_strip(), which kept 128-only literals: strip_top = 90 # bottom-band top plot_top = +9 # V-curve top clearance plot_bottom = -10 # V-curve bottom clearance bottom label = -9 On the 176x176 SSD1333 panel strip_top=90 lands mid-screen (~51% down) instead of as a bottom band, stranding the exposure/matched-star labels high up with a large empty gap above the readout, and the bottom detected-star label clips at the screen edge. Derive the geometry instead (ADR 0009): - strip_top: a fixed fraction of screen height (res_y - res_y*38/128), so the band stays ~30% of the screen on any panel. - label-row clearances: from self.fonts.small.height, so the rows track the (larger) small font on the 176 panel and never collide with the V-curve. The 128 small font is 9 px, so strip_top+9 == strip_top+small_h, res_y-10 == res_y-small_h-1, and res_y-9 == res_y-small_h: the 128 layout is reproduced pixel-for-pixel (zero regression), while the 176 layout becomes a correct proportional bottom band. Verified headless on both DisplayHeadless (128) and DisplayHeadless176 (176). Docs: update the CONTEXT.md focus-strip glossary entry, which described a fixed "~38 px" band, to note it scales with the display. Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Adds a quantitative, graphical focus aid to the Focus screen (
UIPreview). It tells the user how well-focused they are and makes the point of best focus obvious as they sweep through it — independent of plate solving, across the full defocus range.How
PiFinder/focus.py— pure, UI-free module: a self-contained Half-Flux Diameter (HFD) detector (detect_stars,half_flux_diameter,focus_hfd). Runs in the main process on the raw frame; tuned to accept broad/defocused blobs and reject hot pixels / oversized blobs. No dependency on the solver's centroids or SQM (see ADR 0005).UIPreview— a focus strip: fixed log-axis V-curve over a 10 s window, best-focus marker, past-best "BACK UP" cue, HFD/exposure/detected-count/matched-count HUD. Replaces per-frameautocontrastwith a background-anchored, EMA-smoothed display stretch. SQUARE toggles the strip; it persists across zoom (HFD is zoom-independent).python/tests/test_focus.py(11 unit tests: Gaussian HFD vs theory, saturated core, blank frame, oversized→too-defocused, monotonicity, median robustness, hot-pixel rejection).Design references already on
main:docs/adr/0005-focus-hfd-self-contained-in-ui.mdand the "Focus indicator" section ofdocs/ax/ui/CONTEXT.md.Verification
ruff/ruff format/mypyclean;pytest -m unit(190, incl. 11 new) and-m smoke(5) pass.—/det 0, SQUARE toggles the strip, zoom persists it with the relocatedZoom xNlabel, no crashes.Notes
_()-wrapped and fall back to English until the nextnox -s babel; the Babel catalog regeneration is intentionally not included (it swept up unrelated catalog drift).🤖 Generated with Claude Code