[FEAT] Vaapi encode

[thelamer]

Is this a new feature request?

• I have searched the existing issues

Wanted change

Now that DRI3 is in it only makes sense to use the GPU for video encoding if available. Need to use vaapi just like we use nvenc now if it is detected.

Reason for change

na

Proposed code change

na

[aptalca]

Just fyi, for device perm fixing, we now have this functionality in all the bases: https://github.com/linuxserver/docker-jellyfin/blob/nightly/Dockerfile#L17

It plugs the value into this line as a find argument: https://github.com/linuxserver/docker-baseimage-ubuntu/blob/noble/root/etc/s6-overlay/s6-rc.d/init-device-perms/run#L5

[thelamer]

Ok so init-video needs to be removed then

[aptalca]

yup, no longer needed

[thelamer]

If anyone is watching this it can be tested with -e DRI_NODE=/dev/renderD123 it is a flag that will enable it.

The problem I have found that unlike DRI3 or the nvidia support the host to container needs to be in sync or near sync with drivers, like on a bookworm host only noble and bookworm containers work.
On an Arch host only Arch works etc.

I am unsure if this is solvable, and likely why AMD and Intel both have their own alternative Cuda like encoders now.

[thelamer]

I have come to the conclusion that I will never be able to advertise this feature, if you find this issue it can be activated with -e DRI_NODE=/dev/renderD128 but your host needs to be the exact same as the container. It is not something I am willing to support.