Sky Annotation Studio

Sky Annotation Studio is a PixInsight script that helps you find and present galaxies and other deep-sky objects in your solved astrophotography images.

It queries SIMBAD for objects in your image, draws an annotated overview, creates close-up preview crops, and combines everything into a final presentation image directly inside PixInsight.

Installation

Add this repository URL in PixInsight:

https://raw.githubusercontent.com/Anima1337/SkyAnnotationStudio/main/

In PixInsight, go to:

Resources > Updates > Manage Repositories

Add the URL above, then run:

Resources > Updates > Check for Updates

After installation, you can find the script in PixInsight's script menu as Sky Annotation Studio.

Requirements

You need:

• PixInsight.
• A solved FITS, FIT, or XISF image with valid astrometric/WCS metadata.
• Internet access for SIMBAD queries.

The script does not plate-solve images by itself. If your image is not solved yet, solve it first in PixInsight.

What The Script Creates

After a successful run, Sky Annotation Studio creates three PixInsight image windows:

• SkyAnnotOverview: your image with object boxes and labels.
• SkyAnnotGrid: a grid of cropped object previews.
• SkyAnnotFinal: the overview and preview grid combined into one final image.

You can also export the detected object list as CSV.

Basic Usage

1. Open a solved image in PixInsight.
2. Start Sky Annotation Studio.
3. Leave Source set to Active image, or choose a FITS/XISF file manually.
4. Enter the main object, for example M81.
5. Adjust the settings if needed.
6. Click Query and Annotate.
7. Inspect the generated overview, grid, and final composite.

For most images, the defaults are a good starting point.

Parameters

Parameter	What it does
Source	Chooses whether the script uses the active PixInsight image or opens a file from disk.
FITS/XISF file	File path used when Source is set to File on disk.
Main object	Your primary target, for example M81. This object is sorted first and receives the largest crop.
Title	Text shown on the annotated overview image.
Catalogue filters	Controls which catalogue/type prefixes are kept. Remove SDSS or Gaia if you want fewer very faint candidates.
Preset	Applies or saves reusable processing settings such as Galaxy field, Widefield, Messier target, Deep field, Messier showcase, NGC/IC showcase, Deep survey, and Minimal clean labels. Type a new name before saving to create a custom preset.
Label mode	Chooses whether labels show numbers, object IDs, or both.
Label length	Chooses short, medium, or full object names below preview tiles. Medium keeps the classic compact labels.
Preview label alignment	Chooses whether preview-grid numbers and object names are drawn left-aligned or centred below each crop.
Object type	Controls the SIMBAD object family. Use Galaxies only for the normal workflow, Galaxies, quasars off to suppress QSO/blazar-like entries, Known galaxies priority to avoid survey-only candidates, or Stars only for stellar fields.
Catalogue priority	Chooses which catalogue aliases win during naming and duplicate handling. Prefer Messier, prefer NGC/IC, favour known catalogues, or hide Gaia entries when a stronger duplicate is nearby.
Spatial balancing	Advanced control. When enabled, the preview selection is distributed across the image with a grid-based round-robin pass, so one dense corner cannot consume the whole output. Enabled by default.
Fill validation gaps	Advanced control. When Spatial balancing is enabled and strict crop/duplicate validation leaves unused Max rows capacity, the script refills from remaining candidates with a relaxed exact-overlap guard. Enabled by default.
Experimental SIMBAD size crops	Advanced control. Uses SIMBAD galaxy major/minor axis values, where available, to enlarge crops for physically large or elongated galaxies. Enabled by default, but experimental because catalogue dimensions are incomplete and heterogeneous.
Box size	Base crop size in source pixels. Increase it to zoom out object previews.
Patch scale %	Global crop scale. Lower values zoom in, higher values zoom out.
Preview patch	Rendered tile size in the preview grid. This affects grid resolution, not the selected sky area. Very high object counts may compact the final grid tiles to fit the overview width without downscaling text; long names are auto-truncated and reported in the log.
Max rows	Maximum number of preview objects kept after filtering, up to 25,000. This is a cap, not a target: if fewer objects survive catalogue and visibility filtering, the script keeps the valid set instead of inventing extra entries. With Spatial balancing and Fill validation gaps enabled, unused capacity after strict crop/duplicate validation is refilled from remaining candidates. Higher values go deeper, but can include more faint or ambiguous objects and create very large final images.
Sort by	Controls final preview order. Use Visible size for showcase-style output, Image contrast for the clearest crops first, Distance from centre for centre-first browsing, Name for alphabetical ordering, Redshift for approximate farthest-to-nearest depth ordering when SIMBAD redshift is available, or Object class / colour to group the grid by the dominant colour legend class.
Visibility filter	Controls how aggressively low-contrast, nearly blank preview crops are removed. Balanced is the recommended default.
Output mode	Quick presentation style. Scientific keeps dense numeric+ID annotations, Showcase balances presentation and discovery, Poster favours cleaner output with fewer objects, and Discovery Map keeps many small candidates for inspection.
Top 5 object notes	Adds a compact educational notes footer for the most prominent selected objects. Known showcase targets use built-in human-readable descriptions where available; other objects fall back to catalogue, size, redshift, and visibility context.
Highlight showcase objects	Draws stronger boxes/labels for the first five output objects.
Colour-code object classes	Uses different annotation colours for object families and adds a legend to the final image. Yellow is the main target, orange is Messier/NGC/IC, purple is quasar/AGN-like, green is Gaia/stellar catalogue, and deep blue is other galaxy/survey objects.
Show advanced controls	Reveals spatial balancing, validation fill, experimental SIMBAD size crops, debug probe, manual keep/remove, auto-review scrolling, CSV/HTML report export, redshift, auto-title controls, and the log toggle.
Debug probe	Advanced diagnostic. Enter source-image pixel pairs such as 8284,166; 4740,1675 to log nearby SIMBAD candidates and the pipeline stage where they were kept or removed.
Show log	Advanced control. Shows or hides the detailed run log in the script dialog. The PixInsight Process Console still receives the same messages either way.
Reset to defaults	Restores the default settings.

Output Modes

The Output mode control applies a small bundle of presentation defaults before each run:

Mode	Best for	Behaviour
Scientific	Dense reference plates and object identification.	Uses numeric+ID labels, balanced visibility filtering, and default catalogue-style ordering unless Redshift sorting is selected.
Showcase	General-purpose final images.	Keeps your chosen sort/filter choices and is the safest default for attractive annotated outputs.
Poster	Cleaner presentation images.	Uses stricter visibility filtering, fewer objects, numeric labels, and highlighted showcase objects.
Discovery Map	Exploring as many candidates as possible.	Keeps more objects, uses permissive filtering, and favours image-contrast ordering unless Redshift sorting is selected.

If Sort by > Redshift is selected, output modes no longer overwrite it. Objects with usable SIMBAD redshift are sorted approximately farthest-to-nearest; objects without redshift are kept after those.

Colour Legend

When Colour-code object classes is enabled, annotations and preview-grid numbers use these colours:

Colour	Meaning
Yellow	Configured main target, for example M81.
Orange	Classic catalogue objects such as Messier, NGC, and IC entries.
Purple	Quasar, AGN, blazar, or similar active-galaxy candidates from SIMBAD.
Green	Gaia or stellar-catalogue style entries.
Deep blue	Other accepted galaxy/survey objects, such as LEDA, SDSS, 2MASX, UGC, MCG, and similar catalogues.

The same legend is added to the bottom of the final composite image when colour coding is active. If an object belongs to more than one useful class, for example a Messier object that is also AGN-like, its crop box can use a split border: the upper half shows the catalogue class colour and the lower half shows the secondary active/stellar class colour. Colour-group sorting follows the legend order: main target, Messier/NGC/IC, quasar/AGN-like, Gaia/stellar catalogue, then other survey objects.

Result Table And Preview Inspector

After a successful run, the result table lists the objects that were actually used for the final outputs.

The table includes:

• Display ID.
• Main SIMBAD ID.
• Catalogue/type.
• RA/Dec.
• Crop patch size.
• Measured visibility score.

Click a row in the result table to inspect that object in the embedded preview panel below the table. The panel stays hidden until a result grid has been built, then shows a two-part inspector: a zoomed overview context on the left and the generated crop on the right.

The rendered PixInsight image windows are static outputs, so the result table is the interactive selector for jumping between preview objects. With advanced controls enabled, selected rows can be marked as kept or removed, then re-rendered with Render kept.

Double-click the Keep cell in the result table to toggle a row between Yes and No. Start auto-review steps through the table automatically at the selected delay, which makes blank or weak crops easier to spot without manually pressing the arrow keys. The default/minimum delay is 150 ms; faster values can starve PixInsight's UI event loop and make Stop unreliable. During auto-review the inspector uses a lightweight crop-only preview so the dedicated Stop button remains responsive. Click a row manually when you want the full context-plus-crop inspector.

The run log can be hidden with Show log to keep the dialog compact. Each run still reports the elapsed execution time in the log and in the PixInsight Process Console.

Tips

• If the output has too many faint or nearly blank previews, lower Max rows.
• If you still get too many faint catalogue-only objects, remove SDSS or Gaia from Catalogue filters.
• If galaxies are cropped too tightly, increase Box size or Patch scale %.
• If previews show too much empty space, reduce Patch scale %.
• If the preview grid looks too low-resolution, increase Preview patch.
• If the final image becomes very large, reduce Preview patch or Max rows.
• If labels clutter the overview, use Numbers label mode.
• Save your favourite processing setup into one of the preset slots when a workflow feels right.
• Type a new preset name before clicking Save preset to create a custom setup instead of overwriting a built-in slot.
• Leave advanced Spatial balancing enabled for wide fields or crowded survey frames where one dense area would otherwise dominate the output.
• Use advanced Experimental SIMBAD size crops when large or elongated galaxies are being cropped too tightly.
• Use Top 5 object notes for educational/shareable outputs, but leave it off for cleaner pure annotation plates.
• Use advanced manual curation when the automatic filters leave a few false positives that only a human eye can judge.
• Use Label length > Full when preparing outputs where complete object names matter more than compact spacing.
• Use Object type > Galaxies, quasars off if quasar-like catalogue entries are not useful for your presentation.
• Use Catalogue priority > Prefer NGC/IC when you want classic NGC/IC names to win over Messier aliases where possible.
• Use Catalogue priority > Hide Gaia if duplicate exists if Gaia survey entries are causing repeated near-identical previews.
• Use Sort by > Visible size when you want the most prominent galaxies first.
• Use Sort by > Image contrast when you want the clearest detected objects first.
• Use Sort by > Redshift for a depth-style presentation; objects without SIMBAD redshift are kept after the redshift-ranked objects.
• Use Sort by > Object class / colour when you want the preview grid grouped in the same order as the colour legend: main target, Messier/NGC/IC, quasar/AGN-like entries, Gaia/stellar entries, and other survey objects. Multi-class objects can show split crop borders while staying in their legend-order group.
• Use Preview label alignment > Centred for a cleaner poster-like grid where the tile number and object name sit centred with the crop.
• Use Visibility filter > Strict if too many blank-looking previews remain.
• Use Visibility filter > Permissive or Off if the script removes faint objects you still want to inspect.
• Use Start auto-review in the advanced controls to step through previews hands-free while manually toggling obvious false positives. The minimum delay is 150 ms; if the UI feels strained, raise the delay to 300-800 ms.
• Disable advanced Show log if you want a smaller script window, or enable it when you want detailed diagnostics after a run.
• For very crowded cluster fields such as Abell regions, keep Spatial balancing enabled and prefer Sort by > Visible size or Image contrast if you want the most obvious galaxies to dominate the output instead of catalogue order.

How It Works

Sky Annotation Studio reads the astrometric solution from your image, queries SIMBAD for objects in the field, filters for galaxy-like objects, removes many duplicate aliases, and projects the catalogue coordinates back onto your image.

It then creates crop boxes, removes very low-contrast preview crops, and renders the final annotated overview and preview grid.

Before expensive preview rendering, candidates are now ranked by likely practical visibility. The main object stays first, then objects with larger SIMBAD angular dimensions and stronger catalogue classes are favoured. This prevents high-row fields from spending most of their output quota on the first faint catalogue entries returned by SIMBAD.

When spatial balancing is enabled, the script also divides the image into a grid and selects candidates from occupied cells. Existing catalogue priority is preserved inside each cell, but dense regions cannot take every slot before other parts of the frame are represented. Spatial balancing is non-destructive: if Max rows is higher than the number of valid objects, all valid objects are kept and balancing only affects ordering. With advanced Fill validation gaps enabled, a second crop-validation pass refills unused capacity from remaining candidates after the first strict duplicate/overlap pass. If Max rows is lower, the script uses adaptive per-region quotas and refill logic after visibility filtering.

The filtering is intentionally conservative. The script tries to remove obvious blank or duplicate previews while keeping scientifically interesting objects.

Known Limitations

• SIMBAD can list objects that are technically in the field but too faint to be visible in your exposure.
• Very high Max rows values can still include faint or ambiguous objects.
• The visibility filter is based on image contrast, so it cannot perfectly know whether an object is meaningful to you.
• SIMBAD object dimensions are incomplete and heterogeneous, so experimental size-based crops are helpful but not guaranteed for every object.
• The script needs a solved image. It does not run plate solving itself.
• Internet access is required for SIMBAD queries.

Changelog

Versioning note: Patch releases stop at .9. The next release after 1.0.9 is 1.1.0, not 1.0.10, so update repositories and file listings sort naturally.

1.3.6

• Reissued the current debug-cache, crop-blocker, and SIMBAD-size-crop fixes under a fresh package filename to avoid raw GitHub/PixInsight cache checksum conflicts from rebuilding 1.3.5 in place.
• Prioritised galaxy-looking SIMBAD and known-galaxy candidates over star-like Gaia/SAI/FIRST dots during candidate ranking and duplicate crop validation.
• Reduced Gaia/SAI default preview crop size so stellar-looking catalogue points do not reserve oversized boxes or suppress nearby visible galaxies.

1.3.5

• Changed the advanced debug probe into a post-run workflow: a debug run now writes a reusable cache, and Probe saved cache can inspect entered source-image pixel coordinates after the script has been closed and reopened.
• Added debug-cache records for projected candidates, filtering/deduplication, crop validation, validation-fill recovery, visibility filtering, final spatial selection, and rendered output, so missed visible objects can be diagnosed without rerunning SIMBAD.
• Expanded crop-validation probe diagnostics to name the accepted object that blocked a rejected candidate, including duplicate distance and overlap ratio, instead of reporting only a generic duplicate/overlap rejection.
• Guarded experimental SIMBAD-size crops against aggregate/group-like objects such as pairs, groups, clusters, HDC/LDC/CGB entries, and GP catalogue objects so one catalogue aggregate cannot create an oversized overview box or preview tile.
• Added caps for oversized SIMBAD-derived crops by catalogue class, preserving generous padding for real large galaxies and the main target while preventing generic/fallback catalogue entries from dominating the final composite.
• Added log summaries when SIMBAD-size crops are capped or ignored by these guards.

1.3.4

• Changed Spatially Balanced mode to query SIMBAD in spatial image tiles, reducing early server row-order bias where dense catalogue regions could consume the candidate pool before local balancing.
• Spatially Balanced mode now preserves projected SIMBAD galaxy-type candidates that do not match the default catalogue-prefix list as low-priority SIMBAD objects, reducing missed visible galaxies in sparse frame regions.
• Added recognition for PGC identifiers and uses them as normal low-priority galaxy catalogue candidates when present in SIMBAD aliases.
• Added advanced Fill validation gaps, enabled by default, so Spatially Balanced mode can refill unused Max rows capacity from candidates rejected only by the strict duplicate/overlap crop pass.
• Raised the Max rows control ceiling from 5,000 to 25,000 and expanded the SIMBAD working-set cap for high-row Spatially Balanced runs.
• Rendered large preview grids directly at final composite width instead of scaling the completed grid bitmap, reducing pixelated high-row labels and avoiding oversized intermediate grid bitmaps.
• Medium-sized preview grids now choose as many full-size columns as fit in the available overview width instead of staying near a square sqrt(N) layout with unused side space.
• Auto-truncated preview-grid object names to the measured tile width in compact high-row layouts and added a log message explaining when labels were shortened.
• Added baseline-safe preview-grid label bands so object names do not intrude into the coloured crop boxes at compact high-row tile sizes.
• Added advanced Debug probe diagnostics for entered source-image pixel coordinates, reporting nearby candidates and whether each was filtered, deduplicated, crop-rejected, visibility-rejected, spatially removed, or rendered.
• Reworked final spatial selection to guarantee occupied-region coverage first, then fill remaining Max rows capacity with deterministic round-robin selection across image cells.
• Kept Classic mode on the original single-cone/catalogue-priority path while applying the new tiled query and coverage-first selection only when Spatial balancing is enabled.
• Improved crop validation in Spatially Balanced mode by clamping edge-crossing crops safely inside the image instead of rejecting useful edge-near galaxies.
• Added per-cell spatial diagnostics after true coordinate conversion, catalogue/type filtering, filtering/deduplication, crop validation, visibility filtering, final spatial selection, and actual rendering.
• Added catalogue/type filtering diagnostics and catalogue-type summaries to show whether candidates were lost because of object type, catalogue/name filtering, deduplication, crop validation, visibility filtering, or final rendering.
• Confirmed overview annotations and preview-grid tiles are generated from the same final selected object list, so output sorting affects display order only and cannot change which objects are rendered.
• Kept the update package filename aligned with the script name as SkyAnnotationStudio-1.3.4.zip while preserving the install path src/scripts/SkyAnnotationStudio/SkyAnnotationStudio.js.

1.3.3

• Fixed spatial balancing so it no longer discards valid objects when Max rows is higher than the valid candidate count. In that case every valid object is kept and balancing only changes ordering.
• Reworked balanced selection to use adaptive region quotas only when the valid pool is larger than Max rows, with refill after visibility filtering so rejected crops do not leave unused output capacity.
• Added clearer spatial-selection diagnostics: valid objects before spatial selection, Max rows, occupied cells, per-cell quota range, selected candidates, visibility removals, refill count, and final object count.
• Kept catalogue priority and main-target priority inside each spatial region, preserving deterministic naming and ordering behaviour.
• Rebuilt the repository ZIP writer to store POSIX / paths plus Unix permissions: directories are 0755 and script files are 0644, fixing macOS PixInsight updater extraction failures caused by non-executable directory entries.
• Removed SIMBAD server-side major-axis sorting from dense TAP cone queries. Local ranking is used instead because SIMBAD can reject large sorted queries with HTTP 400.

1.3.2

• First pass at repository ZIP path normalisation for cross-platform PixInsight extraction.
• Expanded dense-field SIMBAD candidate retrieval and ordered the working set by available SIMBAD major-axis data before local filtering, reducing early server-row-order bias in crowded galaxy clusters.
• Increased SIMBAD alias query chunk size to reduce many small network requests on dense fields.
• Reworked spatial balancing to use a deliberately coarse coverage grid. Dense cluster regions can span many fine cells, so the coarser grid prevents one physical region from consuming most of the Max rows quota.
• Added diagnostic spatial distribution logs after coordinate conversion, after filtering/deduplication, after crop validation, and after final selection.
• Preserved Classic behaviour: disabling Spatial balancing keeps the older pure-priority ordering path.

1.3.1

• Changed pre-render candidate selection so Max rows is filled from the most likely visible/prominent objects first instead of simply following early catalogue order.
• Strengthened spatial balancing with a finer image grid and stricter per-cell selection pass, while relaxing the cap when Max rows is larger than the valid candidate set so high limits can keep all valid objects.
• Reworked preview crop generation to render the source image once and crop directly from that bitmap, avoiding thousands of temporary PixInsight previews on dense fields and improving high-row runtime behaviour.
• Tightened Visibility filter > Strict so even strong catalogue objects must show image evidence, reducing cases where bright corner stars keep otherwise empty crops.
• Added clearer README guidance explaining that Max rows is a cap, not a guaranteed target, and how spatial balancing behaves in crowded cluster fields.

1.3.0

• Added spatially balanced final object selection before the Max rows limit is applied, so one dense region cannot consume the entire output quota while the rest of the image is left under-represented.
• Made spatial balancing an advanced option, enabled by default, so users can switch back to the older pure-priority selection behaviour when desired.
• Strengthened spatial balancing with a per-cell cap that is applied even when the visible candidate count is below Max rows, preventing dense corners from dominating under-limit result sets.
• Moved Experimental SIMBAD size crops and Show log into the advanced controls section; SIMBAD size crops are now enabled by default.
• Preserved catalogue priority and main-target priority inside each spatial region, keeping existing naming and selection behaviour deterministic.
• Extended spatial balancing into crop validation, preview-candidate selection, and the post-visibility-filter final selection so dense regions are not able to re-dominate later pipeline stages.
• Improved SIMBAD-size crop handling for large or elongated galaxies by protecting size-aware crops from being over-shrunk by overlap handling and by clamping edge crops safely inside the image.
• Tightened Visibility filter > Strict so bright stars near crop edges are less likely to validate an otherwise empty catalogue hit.
• Added concise diagnostics for spatial balancing and notable SIMBAD-size crop decisions.

1.2.9

• Added an explicit #engine v8 directive so PixInsight 1.9.4+ builds that no longer ship the legacy SpiderMonkey (sm) JavaScript engine launch the script with the V8 runtime.
• Removed the legacy pjsr/Sizer.jsh include and switched the dialog layout to PixInsight's native Sizer constructor to avoid V8 redeclaration errors such as Identifier 'HorizontalSizer' has already been declared when reopening the script.
• Ported the main dialog to V8 class-style inheritance so PixInsight no longer raises The 'Dialog' class constructor cannot be invoked without 'new'.
• Replaced VectorGraphics usage with PixInsight's available Graphics drawing class to fix V8 builds where VectorGraphics is not exposed as a global constructor.
• Added a central null-safety layer for PixInsight object lookups so V8 builds that return real null values instead of legacy null objects do not fail with .isNull property errors during final output creation.

1.2.8

• Fixed Sort by > Object class / colour so the object order follows the same class order shown in the final-image colour legend.
• Added split crop borders for multi-class objects, such as Messier/NGC/IC objects that are also AGN-like, so secondary class membership remains visible without breaking colour-group sorting.

1.2.7

• Added a PixInsight 1.9.4+/V8 compatibility hardening pass.
• Removed Sky Annotation Studio's own configuration dependency on ObjectWithSettings, reducing early reliance on PixInsight's bundled AdP/WCSmetadata.jsh helper code.
• Wrapped persistent settings reads/writes so corrupt settings or runtime-specific Settings API issues do not prevent the dialog from opening.
• Fixed legacy catalogue-filter settings saved with | separators being loaded as one invalid catalogue name.
• Re-enabled repository compatibility metadata for PixInsight 1.9.4+.

1.2.6

• Added Sort by > Object class / colour to group the preview grid by the same classes shown in the colour legend.
• Added Preview label alignment with left-aligned and centred grid-label options.
• Preserved object-class sorting when applying output modes, matching the existing redshift-sort behaviour.

1.2.5

• Changed Messier/NGC/IC annotation colour to orange and survey-object colour to deep blue for stronger visual separation.
• Restored a safer 150 ms auto-review minimum because 30 ms and 100 ms can starve PixInsight's UI event loop and make the Stop button unreliable.

1.2.4

• Changed the Other galaxy / survey object class colour from blue to coral so it is clearly distinct from the cyan Messier/NGC/IC class.
• Lowered the auto-review delay minimum from 100 ms to 30 ms while keeping 100 ms as the safe default.

1.2.3

• Reworked top object notes to use more natural educational descriptions for known showcase objects such as M81, M82, NGC 3077, NGC 2976, NGC 2959, NGC 2961, UGC 5302, and MCG+12-09-064.
• Improved fallback object notes so unknown catalogue objects read as explanatory sentences instead of raw metadata snippets.
• Changed Gaia/stellar catalogue class colouring from gold to green so it no longer looks too similar to the yellow main-target colour.

1.2.2

• Added a dedicated Stop button for auto-review instead of relying on a relabelled Start/Stop toggle during timer activity.
• Added extra event-loop checkpoints before and after timed auto-review steps so pending Stop clicks can be processed before the timer restarts.
• Made the final composite always include an object colour legend footer explaining the class colours.

1.2.1

• Made auto-review stop responsiveness more reliable by adding an explicit stop-request flag and event-processing checkpoints before restarting the timer.
• Changed auto-review inspection to a lightweight crop-only preview mode instead of rebuilding the full context crop on every timed step.
• Kept manual row selection unchanged: clicking a table row still shows the full context-plus-preview inspector.

1.2.0

• Stabilised auto-review with a one-shot timer loop and re-entry guard to reduce PixInsight/V8 UI crashes.
• Changed the auto-review default/minimum delay to 100 ms.
• Added a Show log toggle so the dialog can stay compact while still logging to the PixInsight Process Console.
• Added elapsed execution time messages after normal and curated renders.
• Added a colour legend footer to the final composite when object-class colour coding is enabled.
• Expanded tooltips and README explanations for output modes, colour classes, visibility filtering, auto-review, and log behaviour.
• Increased the remembered/default dialog size floor to avoid the compressed-button layout seen in smaller windows.

1.1.9

• Added built-in priority presets: Messier showcase, NGC/IC showcase, Deep survey, and Minimal clean labels.
• Improved Redshift sorting to use approximate farthest-to-nearest cosmological depth ordering, with missing-redshift objects sorted afterwards.
• Added double-click toggling on the result table Keep column.
• Kept result-table keyboard focus after keep/remove actions so arrow-key curation can continue immediately.
• Added advanced auto-review scrolling with user-configurable preview delay.
• Colour-coded preview-grid numbers now match the object class colour when class colouring is enabled.
• Reduced default dialog height and added persistent remembered dialog size between launches.

1.1.8

• Added a visible Output mode control for scientific, showcase, poster, and discovery-map workflows.
• Added optional top-five object notes in the final composite output.
• Added showcase highlighting and optional colour-coded annotation classes.
• Added an advanced controls section with manual keep/remove, curated re-rendering, HTML report export, redshift inclusion, and optional auto-title.
• Added custom preset names so users can save new presets instead of overwriting built-in presets.
• Added redshift metadata to SIMBAD parsing, CSV output, HTML reports, and optional notes.

1.1.7

• Tightened Strict visibility filtering so edge stars cannot keep otherwise empty preview crops by themselves.
• Added centre-evidence requirements for Balanced and Strict visibility modes.
• Reduced the influence of global crop contrast in visibility scoring to make false positives less common.

1.1.6

• Added persistent processing presets with Galaxy field, Widefield, Messier target, and Deep field slots.
• Added an experimental SIMBAD-size crop option that uses major/minor axis metadata where available.
• Improved visibility scoring so bright stars near crop edges are less likely to keep otherwise empty preview patches.
• Extended SIMBAD queries and result handling with galaxy dimension metadata for future crop tuning.

1.1.5

• Added Label length control with Short, Medium, and Full preview-grid label options.
• Added Object type filtering modes for galaxies-only, galaxies without quasar-like entries, known-galaxy-priority output, and stars-only queries.
• Added Catalogue priority control for Messier-first, NGC/IC-first, known-catalogue-first, and Gaia-duplicate suppression workflows.
• Included the selected object type and catalogue-priority mode in the run log for easier reproducibility.

1.1.4

• Fixed selected-preview inspection after a completed run.
• Prevented legacy helper previews from remaining open after run, reset, or close.
• Improved cleanup for temporary preview state without adding extra previews to generated output images.

1.1.3

• Fixed a selected-preview crash caused by dialog-side context crops calling an engine-only helper.
• Kept the source-image context crop behaviour from 1.1.2, but made the clamp calculation local to the inspector.

1.1.2

• Fixed selected-preview inspector scaling so large context crops are fitted into the panel instead of being clipped.
• Built the inspector context crop on demand from the source image window for more reliable object positioning.
• Avoided storing hundreds of large context bitmaps in memory.

1.1.1

• Kept the embedded preview inspector hidden until a result grid exists.
• Changed the inspector to show both a zoomed overview context and the selected preview crop.
• Hid the PixInsight Process Console after a successful run to reduce workspace obstruction.
• Clarified that the result table is the interactive selector because generated PixInsight image windows are static outputs.

1.1.0

• Replaced external selected-preview image windows with an embedded preview inspector in the script dialog.
• Clicking a result table row now updates the enlarged preview panel directly.
• Automatically shows the first preview after a successful run.
• Improved usability when the Process Console or modal script dialog blocks access to separate PixInsight windows.

1.0.9

• Added initial result-table click handling.
• Stored generated preview patches with the result data.
• Added selected-object preview support for table rows.

1.0.8

• Added Sort by output ordering modes: default, visible size, image contrast, distance from centre, and name.
• Added Visibility filter modes: off, permissive, balanced, and strict.
• Added patch size and visibility score columns to the result table.
• Added patch size and visibility score fields to CSV export.
• Logged the selected sort and visibility modes for easier debugging.

1.0.7

• Added stronger duplicate suppression for high Max rows values.
• Reduced duplicate low-priority crops from catalogues such as SDSS, Gaia, 2MASX, and LEDA.
• Added tooltips for all relevant controls.
• Clarified parameter behaviour in the interface.

1.0.6

• Made Box size actually affect generated crop sizes.
• Kept catalogue-specific crop behaviour while making it scale from Box size.
• Kept Patch scale % as the global crop zoom control.

1.0.5

• Added a visibility/contrast filter for preview crops.
• Removed many low-contrast, nearly blank low-priority previews.
• Made overview, grid, and final composite use the same filtered object list.

1.0.4

• Improved catalogue priority to better match the original Python version.
• Made the configured main object sort first.
• Suppressed broad group aliases such as M81 Group and M82 Group as preview objects.
• Prevented some subcomponent aliases from appearing before canonical Messier objects.

1.0.3

• Improved crop sizing for elongated galaxies.
• Reduced cases where long edge-on galaxies were cut off.
• Improved final grid spacing around labels and boxes.

1.0.2

• Improved object coverage across the image.
• Broadened SIMBAD query and filtering behaviour.
• Improved handling of solved active PixInsight images.

1.0.1

• Improved compatibility with future PixInsight JavaScript changes.
• Removed older JavaScript patterns that were less future-safe.
• Improved direct active-image workflows.

1.0.0

• Initial PixInsight release.
• Added SIMBAD querying from solved image metadata.
• Added annotated overview output.
• Added preview grid output.
• Added final composite output.
• Added CSV export.