For those who don't know either company: Unity is a popular 2D/3D commercial game engine, and Parsec is a low-latency remote desktop tool that initially focused on gaming, but pivoted to an enterprise product for remote creative teams during the pandemic.
As someone who used Parsec for ages to game at 1440p 60hz via AWS on my old-as-sin MacBook: thank you for the amazing work you and your team mates did, and I'm stoked to see the company have such success!
I use parsec at home to run a gaming machine in a gaming room... and use the same machine as an art creation machine in another room, parsec is one of the few solutions that plays well with pressure and stylus, custom resolutions and even 120 hz (surprisingly useful when trying to have responsive UI in art apps)
It’s the remote control product, with some extra features and permissions for teams (useful for having a cluster of devices and assigning them out to remote workers who need specific hardware etc).
I hope this doesn't mean the end of Parsec as a standalone service. There's nothing out there that comes close to Parsec's performance and it'd be a huge loss.
> Oh, and one more thing: we love the communities around games as much as anyone, and have no plans to change anything about the free Parsec app. We hope creative professionals get as much joy out of creating the next great fighting game as you do playing them.
On the client side at least a lot of what Parsec was leveraging seemed to be built on webrtc, so there's a lot of potential for similar things to come along I think.
I'm working on something similar, which will be called CloudGamepad. The trick to getting good quality and low latency is forward error correction and UDP. UDP will arrive quickly but is lossy. And the FEC allows you to recover a certain percentage of lost packages.
Does it work similar to rollback netcode in fighting games? Where the game tries to predict what the other player is going to do by looking at what they’ve done in previous frames, and if there’s lag or dropped packets and a destiny happens it rolls back the game state and replays it correctly?
Not exactly, FEC basically means you add in redundant data to your transmission so that if some of the data is lost, you can “rebuild” the missing data from the packets that did arrive. The cost of this is extra bandwidth spent on the redundancy, as well as a little latency.
Rollback is as you say, if the game loses connection with your opponent, rollback 1) uses your opponent’s past actions to predict what they will do next and then presents that to you as if they did it then 2) when the connection is restored, reconcile (if necessary) the difference between what your opponent actually did with what it predicted, rolling back if needed.
FEC in basic form works like raid 5. M parity packets for every N data packets, allowing loss recovery of up to M of the M+N packets without round trip back to host.
You can dynamically tune M and N based on measured loss rates and RTT.
Just because it's a desktop client doesn't mean it's not using webrtc. WebRTC is a much broader standard than "in-browser videoconferencing".
The main thing is that it's an open(ish) standard built around peer-to-peer low latency connections, with a lot of software already built up around using it.
I figured they got better video encode/decode with real hardware and not in the web. But it also makes sense they can do better networking the lower to metal they can get. I was using a 2011 mac air, and web client was worked but the app install was like magic.
Parsec's latency is great, but it's not great for text (like to the point I can barely read text in VSCode). Are my settings off, or is this an acceptable tradeoff for the heavy graphics workloads they are targeting?
Congrats to the Parsec team, I'm a big fan of the service. Perhaps I'm out of touch, but it seems like $320M is a bit low, although I know it was still somewhat niche. Twitch turned out to be a huge bargain at $1B. These server-intensive video streaming services seem to require a lot of engineering and opex, although maybe Parsec was mostly on the client. It looks like they raised 2 rounds last year A at $7M in May and B at $25M in Dec- but I couldn't find a valuation. I guess that's nice cashflow for those investors, but I would think they would rather swing for the fences, no?
Slightly offtopic, but is there any open source/fair-code equivalent of Parsec that doesn't fail blatantly on non-standard aspect ratios such as 32:9 or have lots of other failure modes that they seemed entirely uninterested in fixing (marked as known issues for years, and no public source code available so I can't diagnose or fix them myself)?
I looked and there didn't seem to be much at all that didn't rely on just one vendor's as-buggy encoding stack - such as NVIDIA Experience software instead of just their drivers, and it's sad to see the closest tech to this this be part of some acquisition instead now by a company that seems uninterested in open sourcing anything even if it doesn't endanger any of their commercial market (e.g. Unity and their Burst LLVM compiler thing)... I don't even need all the matchmaking and team sharing stuff, I just want NVENC/etc.-based desktop streaming that doesn't bomb itself over the slightest system anomaly.
I use Nvidia's ShieldTV server in the Experience package with Moonlight as the client, but yes, the Nvidia closed source backend does fault on some things (such as only shows primary monitor in multimonitor setups), but the stream quality is much higher and latency lower than using Steam to stream my desktop.
I haven't tried it yet, as I do run Windows /w Nvidia GPU on my desktop, but it apparently uses nvenc correctly (I'm on a Turing, Turing and Ampere have that fancy low latency mode for their encoder; something Steam still doesn't engage for in home streaming, yet uses the nvenc API directly), but also can support multimonitor (has a shortcut key to cycle through monitors).
They got quite a few business clients that needed high fidelity and low latency Remote Desktop client including ironically enough many game dev studios that needed a better solution than to ship heavy and expensive dev workstation to their employees homes, they also got quite a few proviz users onboard that also needed to be able to leverage their existing hardware remotely.
I’m guessing that Unity can offer now a remote development workstation solution for game development and visualization.
Don't know why Unity is pursuing cloud, you're still going to need a good network speed. They'd be better served investing further in WebGL support, specifically in reducing the file binary size so we can have lightning fast load times for console-quality games running at near native performance in the browser.
Shameless plug, my startup is working on exactly this at the moment but for Unreal Engine 4, but we have plans to bring this tooling to Unity as well. The web has the opportunity to become the #1 distribution platform of choice for real time 3D developers, and it's the path to the decentralized metaverse.
Ten years ago I thought that people would stop installing random indie games onto their computers and just play these small games in the browser. It never happened, people trust Steam, Gog, Epic and Itch.io more than they trust the web. Instead the web became increasing bloated and the ads and tracking became unbearable.
When Unity do streaming, it wont be through a browser, it will be a dedicated app that users trust. (like Netflix).
The value proposition of streaming is not that its on the web, it's that you don't need expensive hardware.
It is hard for WebGL to pick up, when the tooling is even worse than OpenGL for 3D debugging (use the native ones they say), and you are at the mercy of the browser to choose between software and hardware rendering depending on the client's hardware setup.
Unity is pursuing cloud because there is such a large opportunity. Even in the world you imagine where WebGL games take off multiplayer net code, matchmaking, server hosting, live ops, etc. could remain large markets.
Today, most large game companies do many of these things in-house — by offering services to solve some of these challenges Unity is hoping to lower the barrier to entry for smaller developers to compete and convince the larger ones that they would be better off using managed services instead.
WebGL is only good for shader toy, ecommerce and Flash like games from 10 years ago.
I wouldn't bet on a technology stack frozen in 2012 hardware capabilities, where I cannot be certain if the users can use my 3D renderings at all if they are misfortune enough to be on a mix of hardware, OS or GPGPU whose browser decided the best thing to do for user's safety was to switch to 3D software rendering.
The problem with the browser is that it is too bloated. Even if you manage to get performance with wasm you are still high up on a huge dependency tree that you'll never be able to climb down from.
As electricity prices go up we need to get closer to the metal, OpenGL might not be good but it's the best tool we have and probably will ever have. Vulkan might replace the middleware but the OpenGL API will survive.
As for the metaverse it does not depend on one implementation, but a bunch of file formats and network protocols, I'm pretty sure this protocol will prevail: http://github.com/tinspin/fuse
I've been playing around with similar ideas in the space. For instance, instead of delivering full frame buffers to your clients, send them instructions on what to draw based on their current viewport dimensions (e.g. webgl). This is more constrained and requires the application to target a specific UI framework - although I can support arbitrary 3d scenes of modest mesh and texture complexity now. The headline advantage here being the amount of data going across the wire is almost trivial in comparison for most use cases.
I think there is still a lot more to go in the area of server-side application delivery tech. Our networks are getting faster and our compute is getting denser.
The next generation of shardless MMO would depend on something like this. Getting to 1 million concurrent players will probably look a lot like a stock exchange.
Congrats to the Parsec team! I remember using it several years ago when it was still fairly early on and it was impressive then. I wasn't sure if it was a model that could really take off at the time but glad to see it has.
Really wish there were opensource alternatives to Parsec and Rainway that weren't tied down to hardware (Moonlight). As/if we move away from x86 I think it would be a great way to archive old games that haven't found their way to Webassembly.
It would definitely be great to see archive.org use something like it stream old software/games/media without having to distribute it.
“remote desktop with low-latency, sound, and gamepad support” _feels_ like the sort of well-scoped highly-technical problem that open-source should excel at... I remember wanting that back in the days when VNC was considered state of the art and being told it was impossible, and now we’re once again in the situation of closed-source innovation and open-source trying to copy it :(
I wonder what the odds are of parsec opening their code / protocol, making their money from renting out servers...
In addition to using Parsec's tech for cloud-based gaming, I also foresee Unity using it to stream shared applications into multi-user mixed reality apps (for example, multiple users in AR or VR sharing a Windows desktop).
Isn't VR extremely latency sensitive? I remember reading Carmack's blog posts about needing to reduce input and screen lag even on local machines to make it bearable. It just seems to me like the least likely place for streaming to actually work.
To clarify, I don't mean streaming the VR or AR app itself. I mean streaming content that users can interact with within a VR or AR app. For example, imagine that you're in a productivity VR or AR environment collaborating on a Word document with one or more other people.
Yes but since VR seems to move towards standalone headsets there might be more potential in doing the heavy lifting off the device and deliver it half baked? Then the device can do the latency sensitive work?
With the quest I can stream from PC over wi-fi today. But yeah, doing it over the internet seems far fetched.
Only mostly so. It's latency sensitive in that your eye and head movements need very fast responses to avoid nausea in lots of people and produce a good experience. It's not latency sensitive in that a video stream needs to react to you at 90-120hz kinds of speeds. As long as the position and rotation of the video reacts and moves with the rest of the environment, the application itself won't need to respond any quicker than it already does outside vr.
Now I think you're thinking of running the vr application that's doing all the rendering remotely and strraming thst to the headset, it would definitely run into issues because of the latency.
Is there any way you ever make $320m back by selling a better performance version of teamviewer/remote desktop? Or is there some other play here - e.g. a Stadia/GeForce Now competitor?
Microsoft's Game Pass xcloud service for iOS leverages Rainway, a competitor to Parsec. As a user of both I can say that Parsec outperforms Rainway although Rainway works in the browser.
Seems like Unity is looking to build a similar stack with Parsec at the helm, but I guess we'll see...
Let me rephrase your question: can they sell three thousand more enterprise license seats by bundling this in? Because they’ll have made it back in a couple of years if they do.
Seems like Parsec is being sold to game developers as a remote work/collaboration tool. That's on obvious fit for Unity, which already sells commercial collaboration products/services (https://unity.com/products/unity-teams).
Idk if they'll get into game streaming, but for their existing business it makes a lot of sense already.
Yea I heard TeamViewer is insecure. And average internet users don't care about privacy. More than 2 billion people using Facebook confirms my hypothesis.
Would like to know more about its security breaches. Really hard to find anything. I suspect that they've scrubbed a lot of the bed press off of google and the internet in general.
"Parsec has become a go-to solution for hybrid work models among gaming companies such as Electronic Arts, Ubisoft and Square Enix and is integrated in many of the industry verticals where Unity sells solutions, including media and entertainment, architecture and design, and more. Together, Unity sees an opportunity to drive shared momentum with these customers through targeted cross-selling and bundled solutions."
They don't do cloud hosted machines anymore, but the client is still free. You pay to get additional features (full 4:4:4 color, multi-monitor, drawing tablet support, and other enterprisey features). https://parsec.app/warp has the comparison between free and the other paid plans.
What's special about Parsec compared to competitors like
Shadow.tech that seems just as good? (if not actually a bit better - i think personally after trying both)
Both are impressive but the tech must be somewhat standard no?
Also Shadow runs its own datacenters which seems far more impressive team wise.
Can anyone shed some light on why this technology is worth so much in that context?
Shadow has you uses their servers to put your stuff on.
Parsec is more like TeamViewer, VNC or AnyDesk in that you uses your own hardware to both connect to and from, albeit with a latency that's low enough to play games, and color fidelity that's good enough to do graphics. It is much valuable to businesses that wants to levrage their own powerful hardware but no be limited by teleworking employee.
Thanks, but i've used both of for an extended amount of time and Shadow is just as good, and honestly is more stable in games in my opinion - of course it's a bonus that you can host yourself in Parsec but the tech seems like it's the same since Shadow does this in its own datacenters and it works exactly as parsec with remote work.
I'm in a group for educators that use unity in the classroom and the most requested feature I've seen is having a way to utilize the tool from low power/cost devices like chromebooks. This seems right in line with that.
Our startup is working on Unity and Unreal Engine games and applications exporting to WebGL, targeting Chromebooks. Join our Discord if you're interested in learning more:
Congratulations to parsec team!. The remote desktop experience on Parsec is smooth and close to 60fps. I can even play games from my PC over 4G network. This is a killer solution for remote desktop.
Desktop as a Service seems to be taking off in a lot of areas. I suspect the cost of cloud vs giving employees laptops is the cause. It isn't that devs don't have access, it's that companies find footing the bill an increasing liability.
It much easier to manage Unity licenses when we can't actually install it on our machines :)
Unity has this high level plan to be in the business of hosting things, assets in source control, cloud builds, metrics. I'm sure they will have workstations you can rent, same as MS is doing.
I wouldn't be surprised to see a service where developers can provide streaming test builds to end users. (Much safer than sending end users builds in a world where DRM is mostly useless)
For those who don't know either company: Unity is a popular 2D/3D commercial game engine, and Parsec is a low-latency remote desktop tool that initially focused on gaming, but pivoted to an enterprise product for remote creative teams during the pandemic.