Wikimedia Commons recently announced their 10 millionth hosted image, and is seen as the best and most stable way to get image collections out into the public domain remaining free to use and with no advertising. It has rapidly become a default resource for journalists and publishers to find accurately described images with unambiguous and reviewed copyright status. Images must be free for any use (including commercial reuse) but can be legally obliged to have any attribution you require, such as a name and link back to an institution website. You can even have a special license template with the logo of your institution on it. If you click on the Commons link to the official Obama photograph below you can see a specially designed license with the logo of the US Congress built in.
|The first known Japanese Buddhist map of the World dated 1710.|
Publicly available thanks to Geographicus donation
There is an active community of experienced tech-savvy wizards who will investigate the best way to mass upload collections from other sites or databases giving a great deal of thought into how to preserve metadata (such as dates or geolocation), optimize the resulting page layouts and then create a script (bot) to fully automate the task. They take requests for new jobs at Commons:Batch uploading and don't charge a bean for their intricate and tricky work.
Lastly, there is also the option of converting from other online hosting systems. For example many public bodies have loaded image collections to Flickr (owned by Yahoo) as their staff find it easy to understand, categorize and use. Commons has tools such as Flickrripper to automate the process of importing them over and there is no need to stop using Flickr if that is what you are comfortable with.