You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
There is no single optimal source-art size: rendered cover dimensions vary by display resolution, grid shape, list/detail layout, and device pixel density. Users should retain good-quality source artwork rather than manually pre-scaling every image.
Implemented pipeline
Frontend calculates cover requirements from actual grid and detail viewport geometry.
Grid and detail views request independent image sizes.
Frontend sends bounded media.image requests through maxSize.
Core snaps requests to bounded thumbnail tiers: 32, 64, 128, 256, 512, and 768.
Core resizes once and caches lossy WebP thumbnails.
MiSTer can consume Core-owned local thumbnail paths instead of transferring base64 payloads.
Frontend caches decoded images in process memory with a strict 128 MiB LRU cap.
This keeps transferred and decoded images close to their rendered size without sacrificing detail-pane quality or requiring users to maintain device-specific artwork copies.
changed the title [-]Determine optimal image sizes for boxes to improve loading performance without losing quality.[/-][+]perf(media): request display-sized cover thumbnails[/+]on Aug 17, 2026
Activity
[-]Determine optimal image sizes for boxes to improve loading performance without losing quality.[/-][+]perf(media): request display-sized cover thumbnails[/+]wizzomafizzo commentedon Aug 17, 2026
Completed through Core’s tiered WebP thumbnail cache and frontend’s display-sized grid/detail requests.