The screenshot above is 90+ high res avatars in High Fidelity in a small shared space while also kitted out in VR rigs. As anyone who's tried Second Life (or any other MMO for that matter) can tell you, this would be nigh impossible without hellacious, grotesque lag. Indeed, VR evangelist Robert Scoble argues that's the key reason SL never went mass market. I don't quite agree with that*, but it's definitely true that the first social VR platform to master high quality high concurrency will have a key advantage over others. According to HiFi founder Philip Rosedale, High Fidelity has done just that:
In the image above there are 90 people together, most of them with HMD's and hand controllers (either Vive or Rift + Touch). This is a lot of data moving around: Each person generates 100Kbps of compressed audio from their microphone which is sent to the audio server, and also about 200Kbps of joint motion data, which is sent to the avatar server. That is about 30Mbps received by the servers. The output from the servers was about 270Mbps, or about 3Mbps per receiver... Latency has to stay low for everyone to feel connected (with each other and the performers). In this event, most people were experiencing a latency of between 100 and 150 milliseconds - the time delay between motion or sound from their actual bodies being seen by others.
Very impressive, if the average High Fidelity user can replicate this too. Which Philip is challenging them to do -- which turns out to be a surprisingly cost-effective stress test:
Try it yourself: If you want to host a 100-person event with High Fidelity, you will need to use a server with at least 50Mbps upstream (to your server) and at least 500Mbps downstream. For this test we used an Amazon c4.xlarge instance for the avatar, entity, and message servers, and a c4.4xlarge instance for the audio mixer (which is the most heavily loaded). As a side-note, bandwidth and CPU are getting really cheap: the Amazon hosting cost of a two hour 100-person event run this way was only about $4.50!
I'd say that's more than a side note, because that cost is likely to come down even more in coming years.
* Why I think the lack of high concurrency didn't hurt Second Life: Linden Lab insiders tell me something on the order of 95% of new SL account users quit on the first try well before they even got to a social area with more than a few avatars -- and for those who did get that far, often crowds contributed to their sense of feeling overwhelmed and intimidated, and also discouraged them to stay. Meanwhile, Minecraft went mass market even though (correct me if I'm wrong, 'Crafters) only a couple dozen users typically log into the same server. To Scoble's point, however, I suppose you could make the case that if Second Life could host packed live music concerts and other kinds of big crowd-based content, many more would have been encouraged to check it out.
Quitters quit for all sorts of reasons. Some because there's too many around, others because there is nobody around. Everyone's needs vary. I don't consider 90 to be a particular achievement, we did a test in cooperation with Sin Wave back when the first Class V regions were introduced, and we packed over 100 into a region. There wasn't a lot of content or scripts there, of course, but the build in that image you have looks pretty basic as well. I see half of those avatars have the exact same textures on, and I don't see any flexi hair or dresses waving around.
The architecture of SL and HiFi/Sansar will be a scaling failure so long as they continue to fix server resources to simulated geographies. The entire focus of server resources should be user based, so the more people you have in an area, the more servers you have there sharing the load. Service Based Architecture is the only way to scale properly to limit latency no matter how crowded it is.
Posted by: IntLibber Brautigan | Thursday, February 09, 2017 at 01:06 PM
Good progress though, this is important for drawing crowds who are there to see specific events and may not have ever heard of Hi Fidelity before. CSI did this and maybe some fan art efforts on flickr, youtube and other social media? I don't know how much they draw, CSI TV show ran a sim for more than just a few hours or weeks lol. So, maybe it was drawing attention?
Hardware requirements are a big issue, Steam's page for Hi Fidelity doesn't seem to show a common desktop user sys req. Although for VR headsets, sure. Oculus has similar sys req. But Windows 7 only? What if I don't use a VR headset, would they be less?
IMVU is successful but seems to just have roomish spaces. SL makes a bundle of cash for LL, can't really complain if you are making money right?
HI Fidelity looks like a fun space and a ground floor opportunity to experience something, build things and become a bit of a VR expert in as it is open source so you can dig into the code and scratch your head a bit at least. Maybe you will learn something even! Or just...check out the cows or see a concert. Can't run it, don't have that type of hardware right now. Who knows, maybe one day things will work out a little better for me. I should really waste my time on something other than VR, I know.
Posted by: nameinthisbox | Thursday, February 09, 2017 at 07:54 PM
Looks like the military is closing down MOSES Project out of opensim due to the opensim community rejecting any help while turning down it's foundation offer.
They are closing down the Halcyon testing grid at the end of February.
They are forking a version of High Fidelity - this is big news for HiFi fans while a huge but unappreciated loss to the aging opensim platform.
Posted by: Private Ryan | Saturday, February 11, 2017 at 05:56 AM
High Fidelity hit over 100 avatars last Friday, February 10th. With a small, but dedicated dev team, they are slowly moving towards the vision of VR that will be open and scalable. Can I get a woot?
Posted by: DrFran | Sunday, February 12, 2017 at 07:04 AM
This is an impressive achievement, good job HF. We have found that the network bandwidth in these situations seems to be a rather precious resource.
I need to address comment above:
1) We no longer produce patches or submit updates to the open simulator core, rather the MOSES code can be obtained from our gitHub site if anyone wants it. Additionally, our work with Halcyon continues with some directly from their git repository.
2) MOSES-Halcyon test grid is under development and no plans to shut it down. We recently added Freeswitch voice support and tested during last two recent office hours meetings. Test grid is extremely useful.
3) While we maintain our own HF server for testing purposes, we have no plans to fork High Fidelity. We monitor HF and other projects for suitability in our research.
To monitor our progress or start a conversation, feel free to join the MOSES community mailing list:
- Send an email to [email protected] with "SUBSCRIBE MOSES-LIST" in the body (without the quotes). No subject line is necessary.
- If you successfully subscribed, you will receive a confirmation email within a few minutes. Please note that copying and pasting this text can introduce errors, so please type it out manually.
- Messages to and from the MOSES mailing list will use the email [email protected].
v/r - Doug Maxwell, Director of Project MOSES
Posted by: Douglas Maxwell | Monday, February 13, 2017 at 10:52 AM