Choose Jellyfin Preview Images by How You Browse

Chapter images help viewers choose a known section of a video. Trickplay helps viewers recognise a moment while moving through the playback timeline. For a self-hosted media server, the useful choice is the one the household's clients actually display—not automatically both.

This is a comparison for planning a homelab library, not an upgrade procedure. It assumes an existing Jellyfin video library and permission to review its settings. An administrator controls generation; viewers need a compatible client to benefit. No additional public network exposure is needed to make this decision.

Match the preview to the viewing habit

A film with meaningful chapter boundaries benefits from pictures attached to those boundaries: viewers can identify an act, concert movement, or named scene before jumping to it. Jellyfin's chapter-image documentation distinguishes the images from the chapter markers themselves. Turning off image extraction does not inherently remove the markers.

Trickplay instead samples the running video for previewing positions across its duration. It is the better conceptual fit when someone remembers a visual moment but not its chapter. The Jellyfin forum explanation describes these as different image sets, rather than two names for the same navigation feature.

• Prefer chapter images when browsing by defined sections matters most.
• Prefer trickplay when visual seeking across the timeline is the goal.
• Consider both only when the clients expose both and the extra processing is worthwhile.
• Consider neither when a small home server has more urgent playback or backup work.

A chapterless file changes the comparison

Without embedded chapters, the chapter-image decision also becomes a chapter-spacing decision. Jellyfin can create artificial chapter points at a configured interval when the file has none. Its documentation places that control at Dashboard → Libraries → Metadata, with the interval expressed in seconds and zero disabling artificial chapters.

Artificial points are not semantic scene detection. A regular timestamp cannot know where a demonstration, song, or story beat begins. Very short intervals can also burden the web player's timeline rendering, according to the official guide. Therefore, manufacturing dense chapter markers is not a good substitute for deciding whether timeline previews are the desired experience.

Budget extraction work separately from playback

Images have to be generated before they are useful. For chapter images, the per-library option to extract during a library scan moves that work into the scan rather than leaving it to the separate extraction task. Jellyfin warns that this can substantially slow scans of large libraries.

Trickplay has its own cost controls. The official TrickplayOptions reference exposes the sampling interval, output widths, JPEG quality, thread settings, and hardware-related switches. These are different dimensions of a workload, not a promise that one preset suits every server.

More frequent samples mean more candidate previews; larger image dimensions increase the image-processing and storage burden. Actual completion time depends on the media and hardware. A fast playback transcode alone is not evidence that an entire archive will generate previews quickly. Avoid adopting another household's processing-time estimate as a capacity plan.

Decide which storage belongs to the server

Chapter images reside in Jellyfin's metadata directory, normally within its configuration location. A media-only backup is therefore not a backup of those generated images. They may be rebuildable, but rebuilding still costs time and reads from the source library. Keep that distinction in the recovery plan.

Trickplay can be stored beside media, as discussed in the Jellyfin forum reference. That changes the write boundary: a server cannot create sidecar output in a media directory mounted read-only. Do not broaden access to an entire NAS merely because an optional preview feature wants somewhere to write.

For a container deployment, the relevant distinction is between writable application storage and the original media mount. Decide whether preview portability is worth sidecar writes, or whether preserving a read-only archive matters more. This comparison does not require replacing an existing storage layout or deleting any generated files.

Judge the result on the client that matters

Server-side generation and client presentation are separate questions. Before committing to a library-wide job, the acceptance criterion should be a visible, useful preview on the intended television, phone, or browser. The forum discussion notes that chapter navigation varies by client; do not generalise a web-client result to every player.

A useful evaluation includes a file with meaningful chapters and one without them. Success looks different: section browsing should make the chaptered file easier to navigate, while trickplay should help identify positions between those boundaries. If files exist on the server but a client shows no useful navigation aid, generating more of them does not resolve the compatibility question.

Keep version changes out of the image experiment

The Jellyfin 12.0 release notes describe trickplay changes including reuse of existing files during scans and fixes for interlaced-video duplicates. An older report about regeneration or incorrect previews may therefore describe different behaviour from the installed server. Record the server and client versions before comparing results.

An upgrade is nevertheless a separate maintenance operation. The same release notes warn about database changes and require a complete stopped-server backup for a path back to the previous version. Do not combine a major upgrade with a whole-library preview experiment and assume reversing an image setting will reverse the database migration.

The practical decision is modest: choose one navigation benefit, confirm it on the actual client, and budget its processing and storage before extending it to the archive. Preview images should make a media library easier to use, not become the home server's largest unexamined background job.