I am currently chewing on the following feature request for a customer project: “As an editor I want to upload videos in any format and Neos should convert the video in one (or several) suitable web formats.”
For images we already have variants and thumbnails, that for example take care to convert TIFFs on the fly.
But we don’t have that for videos.
Why? Wasn’t there a request yet, or are there any other thoughts that stand against it? Would that be a sensible core feature?
I always told customers not to upload and host videos on their website but rather use video platforms that do all the hard work (optimisation, different formats, adjustments to slow networks…). I know that other devs and agencies have the same opinion.
And converting and working with videos has more demands for the server setup than converting images.
Therefore I think this would be a super rarely used feature. Of course it would be nice, but I see more issues than gains as a core feature.
I can only agree with Sebastian. I usually discourage customers from hosting it themselves. I also have customers who have a lot of videos and use services like Kaltura to host their videos.
Since it then only needs a small package that makes it easier for the editors to maintain and integrate the data.
Converting videos is a really heavy operation, so uploading a 1GB video and converting it directly on the webserver would drasticly slow down your server for the next 10 minutes.
There are services though which convert the videos for you and directly give you a uri to point the video player to.
thanks for your feedback! And sorry for not replying for such a long time.
After I did some research myself i have to admit, that its not a trivial task. While hosting and processing larger video files is not really the issue in our project setup. Maintaining the video codecs and formats and keeping them up to date with the device and browser progress, as well as doing adaptive streaming right seems not an easy task.
I evaluated Kaltura on premise, mentioned by @markusguenther, which looks promising.
In the end we decided for Cloudflare stream, as the whole project is delivered by Cloudflare.
For this, I developed and open-sourced a small adapter package, which takes care to hand uploaded videos over to cloudflare and provides the needed playlist- and thumbnail urls for the use in the frontend web player: https://github.com/punktDe/cloudflare-stream