Clearly not everyone is a security guard and needs to look at complete sets of video, but sometimes you need the context before/after motion is detected for an arbitrary length of time.
One example is due to the reportedly in research/development of the “Sound Detection” feature. As it stands now, if a dog is in the house crying, but not moving, you won’t know about it. Personally, I’m using the software to help monitor stress levels of foster dogs in a crate to determine how severe their separation anxiety is. Setting up my camcorder, and transferring the video off the SD card is a pain in the neck, and it also can only record up to an hour and a half-ish with of video.
Another example could be if you’re monitoring an intersection visible from your house, and want to provide statistics about usage/congestion/etc to city council to take some action about signage or something like that.
Also in the legal scenario others have mentioned, you need may need arbitrary duration, not 5-90 seconds of video, and hope to get lucky that you captured what is required.
As a software developer, I can understand wanting to avoid constantly uploading and storing large videos on your servers. It chews up bandwidth, and would require a dramatic increase of storage space, for potentially not a lot of use. Clearly this is undesirable.
What if you offered a separate software package that could be installed on Windows/Linux/Mac/etc, that could act as a passive video client to the Alfred recording software that runs on the same network? Given that you seem to support multiple-simultaneous clients already, this would be just one more, that happens to render video to disk, instead of the screen.
A manual labor example of this would be to use the “IP Webcam” App + VideoLan media player setup to record video and store it on disk. Users could specify where they want to drop the videos, be it on a local hard drive, NFS, remote SFTP, etc. Files could be created in 1 hour increments or something configurable like that.
This proposal would give power users who want this functionality the ability to store as much video content as they have space available, with zero impact on your cloud.
Additionally, from the Alfred App, you could potentially have the software make these videos viewable/indexable on the portal if the App is on the same network as the software package, or even expose them via UPNP/HTTP if on a remote network.