I think adding the additional platforms now is good, maybe in the future there will be mind control, but for now adding touch controls means that content will be able to be build with all the interfaces in mind.
“Early access” is an interesting term, when to take off? Will VRChat develop fast enough to ever be finished?
I actually don’t know how common phone are with 6gb of ram are. Maybe VRChat needs like a compression gate or something I can turn on as an anti scream feature. Basically a volume limiter.
And there would be no good way to detect if an avatar is lying about its “real performance cost” on your side since it is dependant on countless factors.
So when are we going to get the ability to scale avatars like ScaleGoesBrr? I have been waiting for this since mod support was removed. Is this a thing that will be officially supported in VRChat sometime this year, or is it gonna be pushed further and further down the priority list.
I really do not think this is a valid reason for this feature to not exist. It is likely that at least 95%, probably 99%, of avatar creators are not using a modified SDK or are not modifying the VRCA in any way before upload. The safety settings are already working alright (I think texture memory should be a separate value like download size).
Surely there is some method to check server-side or on runtime clients if the performance statistics do not match runtime performance. Even if not, users can still report malicious avatars.
This is already being done afaik, they’ve also said as much in one of the recent dev updates, but the check is only done once, I believe on avatar load.
About Unity3D skin mesh performance starting from 2020 is much slower than 2019
They use compute shaders, and the utilization of the GPU pipeline continues to decrease as the load increases
In extreme cases, it can be as much as five to six times slower than the original load, with poor CPU and GPU utilization.
Could compare the performance rankings between the asset bundle and running clients. Whenever there is a difference, VRChat the service now has a chance to verify two people, is the client lying or is the bundle lying?
take a quick look at protonDB and count how many users have been complaining of black screens, crashes after 30min, unplayable lag, etc and then come back to me again
I will say that a friend of mine did start to have a better experience with Linux, but they ended up installing a Windows partition just to run VRChat.
Glad this is finally getting addressed, been hearing about this for months now.
Oof, bounded by URP. Spacewrap has hit or miss results so probably shouldn’t be relied on to have more projected frames.
Looking forward to better performance with 2021, I’ve been experimenting with 2019 a lot recently and hope there is some noticeable frame time differences after the update.
VRChat doesn’t have any servers that could check that.
It would also be really expensive if they did.
Doing automated client side checks would be a nightmare to develop for something with, as said, countless factors. Would be a ton of false positives and positive negatives.
User reports have to be tested by moderators, which also costs money. And finding the poorly performing avatar, checking that it is lying and then reporting it would be something not many would bother doing. And of course a portion of these reports would be trolls.
And while the percentage of modded SDKs is low now, it would sharply increase with a change like this.
So in the end we end up with a very gradual performance measuring system, that can be skewed by your computer having a bad day, or updating, or you doing other things in the background. A system that could falsely report your avatar.
And VRChat gets a nightmare in terms of costs and the ratings meaning even less than before. And an increase in modded SDKs.
Then you need to trust that the clients aren’t lying.
100% script kiddies would start using mods that allowed them to report others avatars as lying and if enough of them do to a single avatar then VRChat support would waste their time checking the validity of those false claims.
In my scenario VRChat support asks a virtual machine to check the performance of an avatar and is then given the option to ban a pile of script kiddies or an avatar creator or both
That’s amazing! I appreciate the time and energy the staff have put into helping the game grow. The commitment to creating an exceptional virtual reality experience is REAL
Tupper covered this in the first post, VRC+ members will have a crack at it first. So if you’re paying for it, consider yourself signed up. I think they’ll sell a few memberships to YouTuber alts when it’s aBETAble.
Of course they would automate it as far as they could, but they can’t test every avatar and user reports have to be sifted through, because of the amount of false reports and etc.
And this is how I see it ending up:
It would be a better system if we ignore the drawbacks and it was possible. And you couldn’t even just have one performance stat, you’d at least VRAM, CPU and GPU, but within the latter two you could, and should, probably split them into more since not all components run on the same thread.
And now we are sort of back where we started, but people are far more confused, clear understandable performance metrics help.
keep up the good work <3