Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[BryanC]

I'm finding that over time there are more and more "crufty" cover arts to sort through when using Cover Art>Get from Internet. I think that a little automated cleanup of the database could lead to exponential time savings among end users in both acquiring and parsing available cover art (and save JRiver some bandwidth). For instance, removing artwork under 300x300 if it exists in higher quality or removing duplicates with "non-squarish" dimensions. I'm sure there are some more sophisticated heuristics that could be implemented, but even starting simple would have big benefits.

Another suggestion is to sort retrieved cover art by dimensions (from largest to smallest) in the Get Cover Art window. Increasingly often I have to scroll down to the bottom of the list of retrieved cover art to find the highest offered resolution. Or even worse, it's somewhere in the middle of the list that I overlook when scrolling.

Lastly, it might be worth curation to only provide the best quality available, as cover art has a negligible impact on modern disk space. If users really want smaller cover art it could be handled by the thumbnail generator or a user-defined downscaled target resolution in the Get Cover Art tool.

[comox]

+1!

[JimH]

You may have found some weak examples, but the system is self cleaning.  If users don't choose the art, it drops down the list and eventually falls off.  If they do choose the art, it climbs up the list.  Eventually the most frequently chosen art reaches the top.

It's been this way since time out of mind.

[BryanC]

You may have found some weak examples, but the system is self cleaning.  If users don't choose the art, it drops down the list and eventually falls off.  If they do choose the art, it climbs up the list.  Eventually the most frequently chosen art reaches the top.

It's been this way since time out of mind.

Thanks Jim, I figured that something had been implemented like this for purposes of selecting the "Best" artwork, it's good to know it's also being used for cleanup as well. With that said, I don't feel like it is performing sufficiently well (and perhaps getting worse over time), but maybe the blame lies on us. You can imagine that as newer/better artwork is added to the database (for instance, when an album is remastered or reissued they'll often provide higher resolution digital art than what was available previously) it's going into the bottom of the heap. Meanwhile, the old scan with moire effects that has been sitting there for years likely has garnered a significant number of "upvotes" by the benefit of time. I'm wondering if this algorithm couldn't be improved somewhat, perhaps by increasing the heuristic weight of the image dimensions, file size, or newness. Retrieving good cover art is one of the rate-limiting bottlenecks of MC's music organization flow, so even small improvements could have a dramatic effect.