They demonstrated Nvidia VRS to be working with eye tracking back in 2022 in VRChat.
That was more of an experimental build of course, but I don’t actually know what happened to the proper implementation of Nvidia VRS. They did mention that it would take a lot of work though, so it may not have been worth it in the end considering nothing has come of it since then.
I love them, one of the best things every month. Appreciate them a lot and that we can comment here.
That… secret random changes thing is both exciting and scary. That’s my autism and anxiety not liking it. orz
The best in this update is the use of the word “user” instead of “player”! Love to see it. Really wish VRC would universally move away from using words like player and game, since VRC isn’t a game! Sometimes you use “game” and that’s like reeeeee. But in your TOS these words aren’t used so I’m happy about that.
I think you need to be reminded to know what changes have taken place, and you can open a separate detailed column to inform you.
In addition, regarding the animation changes in the 2025.1.1 Open Beta, are we trying to improve the performance issues of animations this year?
Can the maximum number of animation layers of the Avatar be displayed, or even prohibit uploading more than 100 or 128 layers?
I think having a certain hard limit is the most effective method, at least the performance will not drop too much and be improved through forced update requirements.
then what is the issue with having transparency here?
This is severely depressing for me, because my disability makes it debilitatingly difficult to make the transition.
Knowing how and ability to do are very different things. You guys treat Av3 and udon like anyone can just walk into them prancing and pick it up in no time. No. Both of these are unnecessarily overcomplicated just for simple things, and band-aided by tools that create ungodly nightmares and introduce bugs and complications of their own. The old sdk’s were infinitely easier, simpler, quicker, and less tedious and stressful. I’d have to completely set aside everything i’ve built up since 2018, and reset to start from scratch, and this time with magnitudes more effort required.
Vrchat’s creation pipeline DESPERATELY needs the accessibility and simplicity that sdk2 provided, and without the complications and issues (bugs, performance, headaches) that are introduced by 3rd party automation ‘solutions’.
Can’t be done in unity Built-in Render Pipeline. Same with FSR/DLSS. Needs scripted pipeline. Render pipeline shift would make all prior uploaded content in vrchat incompatible - ie total content apocalypse. BIRP has been better and easier for VR, and backwards compatible versatility. There were a number of reasons that picking BIRP and forward render were the ideal choices.
Citation needed.
As far as i know, modders only managed to hack in a static radial bulk render target resolution falloff, and postprocess FSR. Both of which were botched and not useful.
Sure, here
Also I was slightly wrong, apparently they figured out how to deal with the huge amount of work that would’ve been needed in 2023. But still not really sure why nothing has come of it.
It’s a waste of time if it doesn’t work on AMD or any GPU’s prior to the last couple years anyway. A great many users are still on GTX 1070s/1080s and RX580s and similar gpus. They are the ones that need help, not the latest RTX 40xx and RX 7xxx gpu’s etc.
as a ps: vrchat is most severely plagued by cpu problems due to udon and animators etc, rather than gpu shading costs, unless you are in a particularly bad world shader-wise. people have a habit of blaming the gpu’s power instead of memory abuse and cpu bottlenecking which is usually what is actually hurting their frames (as evidenced by X3D being a monumental perf jump for people). taming cpu time and improving threading needs to be their priority.
Nvidia VRS works on Turing GPUs, so GTX 16 series and up, you have this weird narrative spin for some reason that it’s only useful for the latest and greatest high-end GPUs.
So it could be used to help performance on 60 series GPUs starting from GTX 16xx, or run higher resolutions for better clarity.
It’s still nVidia exclusive though. AMD’s variable shading requires motion vectors the same as FSR, which BIRP cannot support, so spending time and effort (money) to develop something only a portion of users can even take advantage of, is wasteful (and unfair).
Significant systems like these ought to be hardware agnostic.
The best in this update is the use of the word “user” instead of “player”!
The word player is still valid in some cases, it’s commonly used in the context of networking where a player is an instance of a user in the world instance. So in-app the wording does prefer user but in world creation and networking the term player is still used in appropriate cases.
A good example where it’s useful to have the distinction is local test worlds where you can have multiple copies of the same user in an instance, they are the same user but they are different players.
nani? orz