https://youtube.com/watch?v=tGgmv1yEI-U windows zip: https://github.com/hydrusnetwork/hydrus/releases/download/v472b/Hydrus.Network.472b.-.Windows.-.Extract.only.zip exe: https://github.com/hydrusnetwork/hydrus/releases/download/v472b/Hydrus.Network.472b.-.Windows.-.Installer.exe macOS app: https://github.com/hydrusnetwork/hydrus/releases/download/v472b/Hydrus.Network.472b.-.macOS.-.App.dmg linux tar.gz: https://github.com/hydrusnetwork/hydrus/releases/download/v472b/Hydrus.Network.472b.-.Linux.-.Executable.tar.gz I had a good week. There are some neat new UI features and several bug fixes. new stuff Although it isn't super useful yet, searching in multiple file domains is now complete. If you click the file domain button on a search page's tag autocomplete dropdown, there is now a 'multiple locations' entry. This lets you select a union of several domains. New users will probably only see 'my files' and 'trash', but advanced users will see more, and deleted domains too. Please feel free to play with this. Autocomplete counts should have accurate ranges and the union file searches, limited though they currently are, do work. We are a step closer to fully supporting multiple 'my files' services. Under options->thumbnails, there is now a dropdown for 'thumbnail scaling'. The default remains 'scale down only', but you can now 'scale to fit' (which scales small things up) or 'scale to fill', which crops and scales up so the thumbnail fills the whole thumbnail space. Give it a go--the animation as one format changes to another is accidentally one of the coolest things I have done. If you right-click on the page tab bar, there is now a 'pages' submenu, listing all the pages at or below that level, that lets you quickly navigate to them. This is a prototype, basically a copy of the same thing you'll see in a web browser, so let me know what you think--I expect to put more time into it. bug fixes I fixed a stupid issue where search pages were refreshing on session load. I regret this slipped through (again!), so I made an explicit test to catch it in future. Sorry for the trouble! I also fixed an issue with 'do work on shutdown' cancelling repository processing and a couple of other jobs as soon as they started. The 'fast shutdown' logic was working a little too well! Images with an alpha channel are now tested on load--if the alpha channel is completely opaque, it is stripped. This saves a little memory and CPU, and it means they will be correctly detected as pixel duplicates of their non-alpha-channel'd equivalents.