I like the idea, but I wonder about scalability and the value of this individual state information versus the lost recognition value of a consistent icon? It seems good on the individual scale, but I wonder about larger sets of files. Do you think it could work in larger scale?

I’d say it depends on the set of files; if they are part of a series of similarly formatted documents, thumbnails are of little use. If the files are differently formatted, they are a useful visual cue.

While it’s attractive, I simply wonder about variable thumbnails when trying to visually locate one file amongst many others, esp. if the majority themselves have variable, and contextually relevant mini representations of the files, often in semi-recognisable states (I’m not sure I can always pick video footage by a small still of arbitrary content, as opposed to say a title scene, or previously viewed or distinctive poster frame.)

Allowing the user to set a poster frame does help. Both MovieGallery and iVideo provide this feature if you want to check out working implementations. The display of variable thumbnails could be tied to application preferences for state retention such as those in Preview’s PDF preferences section.

Thumbnails are undoubtedly useful for images, OK for video if poster frames are set, but a bit dodgy for text docs, as you know… guess it wouldn’t be an interesting challenge otherwise. : )

Yep! Part of me wants comprehensive thumbnailing for visual consistency’s sake, but they will never be as useful for text documents (probably the majority of most users’ files) as they are for images and can be for movies.

So thumbnails of text docs are still subject to the same problems, unless we could determine and use the most visually recognisable regions of a document. Much easier is to just present a text, or optionally audio, summary of the doc, such as via the column-view preview pane.

e.g. thumbnails of whitepapers are not particularly useful in search results, but it would be rather useful to have easier access to summaries of text documents in these result lists. This is esp. true since the summarising engine is already in place, but probably still too much of a performance hit, to be feasible just yet (although with larger HDs, such summaries could become a part of the cached meta-data stored for files.)

Summaries could be superior as a means of locating a file, but exposing them in icon views poses some problems. Perhaps an audio playback widget displayed beneath the icon that would use text-to-speech to read the summary?

Thanks for the feedback, Marc!

I like the idea, but I wonder about scalability and the value of this individual state information versus the lost recognition value of a consistent icon? It seems good on the individual scale, but I wonder about larger sets of files. Do you think it could work in larger scale?

While it’s attractive, I simply wonder about variable thumbnails when trying to visually locate one file amongst many others, esp. if the majority themselves have variable, and contextually relevant mini representations of the files, often in semi-recognisable states (I’m not sure I can always pick video footage by a small still of arbitrary content, as opposed to say a title scene, or previously viewed or distinctive poster frame.)

Thumbnails are undoubtedly useful for images, OK for video if poster frames are set, but a bit dodgy for text docs, as you know… guess it wouldn’t be an interesting challenge otherwise. : )

So thumbnails of text docs are still subject to the same problems, unless we could determine and use the most visually recognisable regions of a document. Much easier is to just present a text, or optionally audio, summary of the doc, such as via the column-view preview pane.

e.g. thumbnails of whitepapers are not particularly useful in search results, but it would be rather useful to have easier access to summaries of text documents in these result lists. This is esp. true since the summarising engine is already in place, but probably still too much of a performance hit, to be feasible just yet (although with larger HDs, such summaries could become a part of the cached meta-data stored for files.)

Anyway, just some thoughts…