Cool, and that makes sense. I must have heard someone say something regarding polycounts about another game and got it mixed up in my head.
This makes absolutely zero sense to me. We can already set performance limits on our own. I don’t see why someone else should be able to override this, it removes control from the user and offers no benefit.
5 Likes
It allows event organizers to enforce a consistent experience for their patrons. 
6 Likes
I don’t view this as “VRChat knows best”, but instead see it as VRChat letting the host of the event/instance decide the rules, and not giving people a way to work around/circumvent the hosts’ decision by demanding everyone else force-show their avatar.
7 Likes
This is a HUGE step in the right direction and going to make hosting our events SO much easier! Looking forward to experimenting as soon as this becomes available. My only concern has already been mentioned above regarding players being able to see their performance rank on both platforms on the UI and also a way to control these independently.
2 Likes
So I can finally charge people who love to use verypoor avatars?
7 Likes
Avatar Performance Gated Instances is a huge step in the right direction and I welcome it. In a time where accessibility to VRAM is difficult for the price, this is much needed.
I do wonder how Group Hosts will manage this. Considering the Furry community, bringing ranks down to Good would be a challenge for most newbies. From my perspective, Medium or better is where I find to be a balanced approach.
2 Likes
If I come into an instance with a very poor avatar, I don’t have to worry about others performance. Plus, lower chance of crashing.
I do like the option of a bypass for select users such as performers. I will still want to see what they show us.
Agreed! I try to recommend at least medium to as many in my communities as possible. Very attainable with a little bit of effort.
Feel like we need a revamp for the PolyBase rankings before having a rank-gating system, the current numbers literally don’t make any sense, like all the data truly shows a performance difference isn’t significant and having a mid-level poly account at poor would make people want to optimise. Also, on the quest side of things, even a small increase to the poly limits would improve the avatar economy,
2 Likes
I also strive myself to always meet that requirement everyday! Hard to convince others to put in the work if we don’t start with ourselves.
1 Like
The reason for this is because…
While there’s certainly some per-polygon cost, it scales linearly with polygon count. Moreover, between the hypothetical 25M polygons and ~18M pixels (3000×3000×2, approximately quest 3 render res) it’s about the same level of things to process, and a typical fragment shader is way heavier than a typical vertex shader, not to mention overdraw and other things that increase per-fragment cost. There’s nothing quadratic scaling in there, it’s fixed cost per vertex. The only potential issue can arise from polygon density resulting in up to 4x effective fragment shader cost, but that’s not related to the count and not accounted for by performance stats anyway.
There are other things that a simple polycount metric ignores, like geometry amplification from outlines or tesselation - applying the same limit to outline-less materials feels simply unfair.
As for anecdotal evidence, sure, blocking all the very poor avatars will help performance. This will also remove everything else from them, be it physbones or animators. You can’t use this method to pinpoint any particular thing as responsible for bad factual performance of those avatars.
For the other side of anecdotal evidence, in crowded instances I’m always CPU-limited. With a 9800X3D. Guess what has near-zero CPU cost? Polygons. The CPU ships them once into VRAM on avatar load, and then does nothing about them. I could throw way more at my GPU before it starts to become a limiting factor, and I’m usually surrounded by those horrible very poors with over 100k polygons. So my anecdata points in the opposite direction /shrug
There are other options one could explore, like individualized performance limits, but those sound as complicated as upgrading the current performance rank system.
That performance-gated group instances addition sounds really cool! I already attend a couple events that ask us to use an avatar that is below a certain rank, and I keep manually showing everyone’s avatar one by one (safety measure so I do not crash from loading too much at once). So for prints and just having more enjoyable group events as a whole, this is great!
If I may add a suggestion though, it would be to define a “host” and “cast” roles for members attending the events as, well, hosts, but also guests, performers, streamers and what not. Because in correlation to this comment here:
That said, you are not allowed to override the instance level restrictions to make other’s appear, even if you force-show their avatar or are friends. This is to prevent the slightly awkward dynamic of players asking others to show their avatar, thus defeating the point of the feature.
That also means performers and cast members will have to play with a limited toolset to benefit from performance-gating their group events. As an event goer, I can imagine a lot of situations where this feature could not be used without causing extra restrictions on the organizers: music concerts (with piano gimmicks for example creating an extra visual appeal), bartenders (some hosts complement their performance with extra gimmicks, so until Item SDK is released for UGC…), fashion shows (even if they are split across multiple avatars, some artists showcasing their outfits will have to further optimize their outfits below the 70k polygon count where sometimes, it is not forcibly the point), streamer events (if the event is centered around them, it should be natural they can showcase some extra stuff while also saving on their PC performance, their PC already runs OBS and VRChat and extra stuff)…
So yes, I feel like these extra roles could really make sure that event organizers can benefit from this huge helper of a functionality without having to compromise. Because otherwise, I worry that this would not get used too much…
EDIT: I reread the post after a much needed night of sleep and missed the part where this functionality is actually added as a group role feature! I thought it was going to be a thing to manually add to members of a group instance, but this does pretty much address the issue I had above! My bad… ^w^’ Again, excellent update! Event organizers will thank you!
This is not possible as everyone is running different hardware
3 Likes
Finally, that is very good. I have been waited for this for a long time. I hope they will do the same thing for public, friends(+), and invite(+) instances too. Pretty please. Thank you for the avatar performance gated group instances!
The 70k polygon limit is often discussed under the assumption of 80-player instances, but people do not spend most of their time in 80-player instances.
Rather, there is a demand to allow users within 280k polygons in, say, 20-player instances, while blocking those who exceed that—but currently there is no way to accommodate this. This should also be taken into consideration.
In practice, once you go beyond 70k, there is effectively “no limit,” so polygon counts tend to increase without bound. Therefore, I believe it is meaningful to introduce new constraints beyond that point (not necessarily limited to the current all-in-one performance rank system). With the introduction of the Avatar Performance Gate, such moderately strict constraints could be effective not only for large-scale music events where accommodating 80 players is critical, but also for everyday instances.
Well, setting aside features that may or may not ever be implemented, I think it would be nice if the Avatar Performance Gate allowed adjusting its limits even after an instance has been created.
1 Like
I recall VRChat said in a previous dev update they are working on this, some tags will be disabled by default and to enable them you’ll need to be age verified.
https://feedback.vrchat.com/feature-requests/p/avatar-performance-restriction-overrides
Let us have the ability to choose for our own machines as well, thx <3
4 Likes
The groups that have enforced poor/good or better for events would most likely be implementing this as in you would of joined them before this change and a staff member would come up to you and ask you to change to a more optimized avatar.
I’ve been saying this for a long time and the responses I’ve gotten just continue to basically boil down to “if you can’t meet these thresholds, we don’t care.”
I have friends who panic just opening Unity and adding a simple drag-and-drop Fury asset to an avatar. Meanwhile I’ve been addicted to editing avatars since I bought my first and even then, I would only do what could be done in Unity. It was 2 years before I dared open Blender to learn to refit outfits and now I’m addicted to it 
All this to say, from my experience as an enthusiast, vs my observations of “most” players: even just getting people away from endless toggles and instead uploading separate outfit versions instead has been an uphill battle. The biggest deterrent I’ve found after having this conversation dozens upon dozens of times: the minute most people discover that they are still going to be ranked as Very Poor despite removing everything but one outfit set that they love?Completely kills any interest they had. Why bother?
I feel like a weighted system would be way more accurate to performance but I am not a dev 
one that assigns x amount of “points” per material slot, texture, light, y number of polys, etc (I don’t know what these numbers would be, but there are smarter people than me who would know how much each hurts performance to weight them) and assigns a rank based on your total number of points. It makes more sense than one that punishes you just the same for having 70,001 polys as it does the person with 500k, 40 materials and 1k Physbones.
At the least, adding another performance rank for the truly horrific would be awesome.
5 Likes