Audio/Video conferencing

I received this question recently by email and thought that my answer deserved to be copied here as well for others who might benefit from it.

The question was about how Foundry’s Audio/Video conferencing compares to Discord’s, and with Discord Nitro. As well as whether hosting through The Forge would give any better results than self hosting. (Edit: When I talk here about Discord, the same applies to Jitsi in that case as they work similarly)

First, when it comes to the difference between self hosting and The Forge, with regards to Audio/Video, most people wouldn’t see much of a difference. This is because you will likely connect to each other, so the limiting factor is each player’s CPU and bandwidth limits (more on that later). In the case however where one player is behind a more restrictive NAT, such as behind a mobile network hotspot, a corporate firewall, or an IPv6 bridge (where another player has no IPv6 address) for example, then The Forge does provide a more robust relay service that is pre-configured and ensures that you will always manage to connect to each other.

With regards to the comparison with Discord, the integrated A/V of Foundry works differently from how Discord works. More specifically, with Foundry, you will connect to each other player and send/receive with each of them individually.

If you have for example 5 players total, then you would send your audio/video to the other 4 and you will receive from them as well, this increases the bandwidth usage you would use. It’s not a lot though, I’d say with both audio and video, probably less than 500Kbps per player. With Discord, you connect to the Discord server itself and send only once, but still receive from all the others. This has a small bandwidth usage overall.

Here’s a graphic showing how the connections could be made. In Foundry’s case, it’s a “Mesh” connection, while Discord uses an “SFU” connection (note that the downlink/uplink bandwidths are probably for HD video, whereas foundry streams 320x240 videos, which is why it would be lower than that) :

I personally don’t notice much of a difference in quality between Discord and Foundry’s integrated A/V though I’ve seen most people praising Foundry’s A/V because it’s a much better quality than Discord. I looked up Discord Nitro’s advantages, but I don’t see an actual quality increase for Nitro users (I might be wrong), instead, I see that it’s related to the server itself, You would need to use a server that is Nitro boosted to level 1 or higher for the audio quality to increase, see the Discord FAQ here : https://support.discord.com/hc/en-us/articles/360028038352-Server-Boosting-

  • Regular server : 96 Kbps (This is radio quality)
  • Level 1 boost : 128 Kbps (This is CD quality)
  • Level 2 boost : 256 Kbps (This is iTunes high quality music)
  • Level 3 boost : 384 Kbps (Probably more of a marketing thing at this point)

Foundry itself does not limit the bitrate that your audio/video will use, instead, it will use adaptive-bitrate and will increase/lower it based on the available bandwidth (yours and the person you are sending to). So if you have a limited upload speed, then the bitrate will decrease, and the more people that join your game, the lower it will be decreased. If you have a good upload speed, usually, you wouldn’t notice any degraded quality as you’d be able to use the full bitrate allowed by the audio and video codecs.

Now to answer the question itself: using The Forge shouldn’t really have much of an impact on your audio/video experience because you would be connecting to each other directly (see the Mesh diagram above). I do have on the Forge a dedicated relay server which would only be used in case you aren’t able to connect directly to another user, in which case, yes, it should make a difference to use The Forge vs a self hosted Foundry.

Note that you can also use the Jitsi module which, once enabled, will use jitsi servers (https://meet.jit.si) for the audio/video calls, which I’ve heard gives better results overall, possibly because it’s also an SFU connection, so uses less bandwidth and CPU for encoding on your own machine. Note that the jitsi connection also doesn’t suffer from one bug in Foundry where the connection between two players may sometimes fail and requires the player to refresh the page.
I know that a lot of videos and tutorials out there explain the long and complicated process of how to set up a jitsi server, but note that it is not necessary and simply enabling the jitsirtc module is enough for it to use the public Jitsi servers with zero configuration from your part, so don’t be intimated in trying the module.

What I personally used in my games (mostly because of some push to talk limitation back then) was integrated video in Foundry as it’s nice to see the video in the VTT itself, but I had kept using Discord for audio, that had worked well for my group.

If you need more clarifications or have more questions with regards to the Audio/Video conferencing feature of Foundry, feel free to comment below.