Repository navigation
Visual capture and assets catalogue
On a Linux development PC with LÖVE 11.5 installed:
python3 tools/capture_love_screenshots.pyThe script starts the Python companion service, runs the LÖVE frontend, switches through its scene catalogue and captures each rendered screen. When no display is available it uses Xvfb. The capture summary is saved as data/visual-captures/love-batch-manifest.json.
This pass verifies that scenes can be opened and rendered in the desktop runtime. GitHub Actions uploads a voiddesk-love-visual-captures artifact from each run (screenshots, JSON sidecars, manifest and logs; 30-day retention). Open the workflow run and download it from Artifacts. Synthetic system/network status is used for this test, not live host data. It is not hardware validation for muOS or a claim that all LÖVE actions are complete.
Both frontends support F12. PNG screenshots and JSON metadata sidecars are saved under data/visual-captures/.
python3 tools/visual_capture.py
python3 tools/visual_capture.py --checkThe standard-library script scans images and fonts in assets/, resources/, love/assets/ and glyph/. It records size, dimensions where available and SHA-256, verifies active LÖVE NEXUS node textures and indexes screenshots.
Generated reports:
- docs/visual/ASSET_CATALOG.md
- docs/visual/SCREENSHOT_CHECKLIST.md
- docs/visual/visual-inventory.json
- docs/visual/ASSET_GALLERY.html
- docs/visual/SCREENSHOT_GALLERY.html
Captures are ignored by Git by default. They may reveal SSIDs, IP addresses, hostnames, personal media names or other private data. Inspect every screenshot before publishing.
python3 tools/visual_capture.py --export-publicThis copies reviewed screenshots into docs/visual/captures/. Do not publish raw captures until they have been reviewed.