Developer Update - 9 April 2026

[calliemaxine]

I hope that you support our choice to choose our avatars with the avi bodies that we like and feel most comfortable in. even if they arent optimized like other bases.

lol lets not do this
requiring optimization is not an attack on your personal identity or self expression.

There has to be a line, somewhere, because that’s how computers work. Discuss whether the existing line is reasonable all you want! But the line is not an attack on your self expression. In a game made up of user generated content, there is a responsibility to optimize for the sake of other people.

[4oZEF]

I believe the topic of PerformanceRank restrictions is a very sensitive one. People come to this world for a variety of reasons, and self-expression is a key part of that.

To get straight to the point, honestly, the PerformanceRank metric isn’t very useful on PCs at the moment.
(VRAM and GPU performance have improved significantly since it was first introduced.)

With that in mind, here’s an idea: instead of relying solely on PerformanceRank, could we implement restrictions based on an avatar’s download size or uncompressed size?
(Similar to the options in the “Avatar” menu in the settings screen)

The key point is that “we users don’t want to restrict avatars based on a broad category like ‘Very Poor’; rather, we want to restrict avatars with extreme uncompressed sizes or polygon counts.”

I should have mentioned this at the beginning, but I am Japanese.
At least for us Japanese users, we purchase most of our avatars from Booth.
However, under current standards, most of these are rated “VeryPoor,” and adding even a few accessories can push the polygon count or material count into the “VeryPoor” range. In this situation, the ones suffering the most aren’t heavy users like me, but light users who recently got a PC and just started playing VRChat.

As an alternative approach, some have suggested that “we should reevaluate the PerformanceRank itself, including the ‘VeryPoor’ category” or “we should create metrics that measure rendering costs more specifically,” but that wouldn’t be a simple task either.

Compared to factors like polygon count and material count, download size and uncompressed size directly impact internet connection speed and VRAM capacity.
This is a key difference from other PerformanceRank elements.

Furthermore, since download size and uncompressed size are also used as metrics within PerformanceRank limits, if these limits can be implemented, they can indirectly serve as PerformanceRank limits as well.

[frou01]

I strongly agree with the idea of ​​restricting avatars in Group instances based on optimization levels.
However, I cannot fully agree with its implementation.

• The criteria for restricting avatars should be more detailed.
  The performance rank system seems rough and geared towards beginners.
  The number of users and PC-spec in Group instances are much more diverse. You should allow us to build something that suits our needs.

• It would be good if staff could remove them anytime. Group instances sometimes transition to an “after-party” mode after the main event ends, and in such cases, performance restrictions are a obstruction.

[dark]

@knah

For the other side of anecdotal evidence, in crowded instances I’m always CPU-limited. With a 9800X3D.

At this point I’m going to stop debating in this thread because it’s clear that until VRChat offers some way to actually test the actual performance of each individual avatar, we will never actually know what the real cost of avatars are.

I’m tired of every attempt to improve performance of UGC being picked apart and ridiculed by every single person as a way to basically say, “We can’t do anything useful or functional because nothing in VRChat is exactly how I want it to be.”

Can we, for the love of god, just accept this as some change in a good direction and see how it goes? If it turns out to be horrible they can just remove or adjust the feature later. Jesus christ.

@narazaka

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.

Okay, so lets make the system dynamic. How do we tell players the logic behind their avatar being rated as good in one instance and very poor in another? How do we design a system that is both easy to understand at a glance, but also dynamic enough to actually account for all scenarios?

@MondoCat

Let us have the ability to choose for our own machines as well, thx <3

As mentioned before: If individual users are able to override it, then the feature is useless.

I am not sure how you all are not understanding that this feature is meant only for group instances and only if the group admins enforce it. The ENTIRE POINT OF THIS FEATURE is to PREVENT VERY POOR AVATARS FROM BEING IN GROUP INSTANCES.

Anything that would allow people to subvert this system makes it literally useless.

[Victoria]

@strasz

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.

It’s my computer and I’ll show whoever I damn well please. If I lag a little, so be it. But do not touch my freedom of seeing who I want in whatever instances I go to.

[dark]

@Victoria

It’s my computer and I’ll show whoever I damn well please. If I lag a little, so be it. But do not touch my freedom of seeing who I want in whatever instances I go to.

You can simply not attend events that enforce it. End of story.

Group owners have the right to determine how their instances are run. Don’t like it? Don’t attend.

[Victoria]

@dark

You can simply not attend events that enforce it. End of story.

Group owners have the right to determine how their instances are run. Don’t like it? Don’t attend.

They can have their “minimum requirements”, sure. But if I choose to show an avatar because I want to, that’s on me. I should not be restricted for hitting Show Avatar because I’m making the active choice to see it completely and not an imposter.

[dark]

@AleynaSand

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.

I think we need to start educating people on how to make production-ready game assets.

For example, one Tracer model I grabbed from Overwatch? Less than 25,000 polygons. Yet, it looked more detailed and complex than 90% of VRChat avatars.

This stuff is possible and the VRChat limits up to Very Poor are already very generous.

[dark]

@Victoria

They can have their “minimum requirements”, sure. But if I choose to show an avatar because I want to, that’s on me. I should not be restricted for hitting Show Avatar because I’m making the active choice to see it completely and not an imposter.

So then should Group Owners simply not be able to enforce any avatar standards at all? Where do you draw the line on what is or is not allowed to be shown? How about groups that ban “meme” avatars? What about huge avatars that visually block a stage? How about avatars with aggressive emissions? What about avatars that can induce seizures?

You either allow group owners to define their experiences, or you don’t. There’s no middle ground because of personal preference.

You are joining THEIR instance in THEIR group. If they wish to enforce an avatar perf rank in order to allow more people to see actual avatars and not fallbacks/impostors all the time (especially in instances that rely on certain shader overrides and lighting effects), then they should be able to do that.

If you want to curate an experience that allows someone to do anything with Very Poors in a group instance, there is not much stopping you from making your own group and running your own events.

At the end of the day, this is currently just a Group feature. People can still host their own Friends+, Public, etc. instances as they please.

[Victoria]

@dark

So then should Group Owners simply not be able to enforce any avatar standards at all? Where do you draw the line on what is or is not allowed to be shown? How about groups that ban “meme” avatars?

You either allow group owners to define their experiences, or you don’t. There’s no middle ground because of personal preference.

You are joining THEIR instance in THEIR group. If they wish to enforce an avatar perf rank in order to allow more people to see actual avatars and not fallbacks/impostors all the time (especially in instances that rely on certain shader overrides and lighting effects), then they should be able to do that.

Ok, let me type this very slowly for you.

The avatars load on MY PC.
I am the one seeing them.
The group owner can set whatever restriction they want so they don’t see them.
I want to see them.
I should be able to press Show Avatar to see the avatar because it is MY PC taking the load of rendering it. Not theirs.

[dark]

@kaii-chan

The reason why everybody uses very poor

In most of the club instances I go to, Very Poor avatars are the minority. Most people are in optimized avatars.

[dark]

@Victoria

Ok, let me type this very slowly for you.

The avatars load on MY PC.
I am the one seeing them.
The group owner can set whatever restriction they want so they don’t see them.
I want to see them.
I should be able to press Show Avatar to see the avatar because it is MY PC taking the load of rendering it. Not theirs.

YOU are joining THEIR GROUP INSTANCE.

THEY are the ones that define what happens in THEIR instance.

End of story.

[Victoria]

@dark

YOU are joining THEIR GROUP INSTANCE.

THEY are the ones that define what happens in THEIR instance.

End of story.

Then they can use SHIELD MODE like everyone else who has a subpar PC that lags when they get near two other people.

[dark]

@Victoria

Then they can use SHIELD MODE like everyone else who has a subpar PC that lags when they get near two other people.

The thing you are fundamentally not understanding here is that the group is enforcing a vibe/experience for EVERYONE.

YOU are not the only person at the event. The point of event instances (like music venues, the ones most likely to use this feature) is to curate an overall experience and everyone participates in that experience.

Allowing you to override reduces the effectiveness of this feature. It means people won’t have an incentive to use better avatars and make the experience better for everyone.

If you have a problem with this concept, then you are exactly the person this feature was designed for. Get a better avatar.

[calliemaxine]

You either allow group owners to define their experiences, or you don’t. There’s no middle ground because of personal preference.

yeah basically this. wear whatever avatar you want, but if a group owner doesn’t want that there… well, sorry? It’s not your group.

that being said, I’ve already made my opinion known that the current performance ranking system is creating the wrong incentives and not really accomplishing what it set out to do.

If anything I’m hoping the discussion here regarding how effective the current ranking system is might get VRC to prioritize revising the ranking system a bit sooner? If they go digging and find that there just isn’t a better alternative I’d be cool with that, but I don’t think the current system is actually useful for a typical player in large events.

[dark]

@calliemaxine

yeah basically this. wear whatever avatar you want, but if a group owner doesn’t want that there… well, sorry? It’s not your group.

that being said, I’ve already made my opinion known that the current performance ranking system is creating the wrong incentives and not really accomplishing what it set out to do.

If anything I’m hoping the discussion here regarding how effective the current ranking system is might get VRC to prioritize revising the ranking system a bit sooner? If they go digging and find that there just isn’t a better alternative I’d be cool with that, but I don’t think the current system is actually useful for a typical player in large events.

Yeah! I honestly agree.

The problem here is people are assuming that this feature is bad because of the current meta or norm of how things work on VRChat.

If we use this feature and the meta shifts, then the problem is adjusted/disappears. This can include and incentive for VRChat to create a more comprehensive/accurate performance ranking system for avatars.

I think this feature is just universally good, regardless of what the specifics are of the ranking system currently or how people use VRChat.

[Plaster]

@Rekka

And this idea of “influencing the community to optimise”? Why, when the imposter system already exists? There’s already a built-in way for people to see a version of your avatar without requiring everyone to become optimisation experts.

Imposters are the solution to not having to look at a grey robot, it’s not the solution to avatars not being optimized. Would be insane to consider imposters as the solution to poor performance, like surely you’d also want to see the facial expressions of the avatar move, audiolink if it got that, etc. Imposters don’t have any of that.

I get not wanting to learn unity or blender, and tbh I’m sure nobody wanted to learn unity lol, but that’s why I said avatar creators, the ones making bases, are also the ones who should not get away with this either. An avatar base should leave enough headroom over for the person buying the base to also be able to add stuff to it, have fun with it, without immediately falling into the very poor rating.

It is very easy to avoid having a very poor rated avatar, which is why this outrage is kinda baffling to me.

But even then, if you don’t got the know-how on how to improve the avatar rating, just ask around! Look at your avatar rating, see what could be the cause of the rating, what can be done to fix it, ask others for help on finding what could be causing the rating to get that bad.

There’s tons of creators who make their own avatar, there’s a ton of tutorials as well. Yes, it involves getting your hands a little dirty with blender, but it doesn’t have to get complicated. I mentioned earlier, if poly count is the issue, deleting edge loops and polys that are hidden underneath clothing is a very easy way to improve it. Don’t know what an “edge loop” is? There’s resources for that too.

Very important to remember: You can have your fully featured, all-in-one very poor rated avatar, but you can also create an optimized version that you can use for events that enforce the rating.

I think this is what some seem to forget, you can upload multiple versions of your avatar, you’re not forced to overwrite your existing one and delete things from it forever. Have two versions, use the optimized one when an instance enforces it.

[dark]

I would like to add a final thing before I get off for the night so I can actually rest my brain from this insane thread:

I’m gonna be 100 that optimizing/making avatars is not actually that hard. The fundamentals are simple and getting some level of optimization is possible without that much work.

What is hard is making an avatar from scratch, primarily because of facial features and visemes/expressions. That is hard. The rest? Not to hard.

I know it may not seem like that, but I can guarantee that if someone were to find the right combination of tutorials/guides or just make a really good guide/pipeline system to help people achieve this, the problem would be solved very quickly.

Granted, I say that while not spending any time changing my current avatar, but it’s also because I already did a lot of the work to get my avatar how I want it and it just works great in any scenario pretty much. There’s no reason for me to update it because it already features what I need from it and it is inherently optimized.

If you are interested in some basic level of optimization, learn:

• Retopology
• Baking materials (simple bake in blender)
• Backing up/duplicating your project/collection/scenes

These are all you really need to get pretty far with optimization. There’s also a really, really good video in improving UVs and texture density: https://www.youtube.com/watch?v=VnmYM3vkYBk

In the mean time, I’ll look into finally making that VRC creator resource website I’ve been meaning to make for two years. I hope to make the knowledge as easily accessible as possible, because I’m tired of people complaining that Blender and Unity are hard. They aren’t hard, you just don’t have the right resources to learn.

VRCFury? People think it’s easy, but it’s because it’s presented as easy. Ultimately, it’s just an entirely new concept/system on top of another system, to make certain functions a little faster/more reliable. Otherwise, if you’re able to learn VRCFury, you can absolutely learn Blender and base Unity.

I’ve tried babysitting people to help them learn how to do these things, and they put no effort in to try to learn the fundamentals of the things they’re doing. They just want me to guide them step by step and basically micromanage their mouse with voice commands so they can feel like they achieved something. It’s really frustrating as someone who put in the effort and actually tried.

I am someone with crippling ADHD who usually can only output about 10% of what I actually want to do in a given week because I don’t have the attention span to do so. I started learning Blender in mid-2022 and have been very, very casually learning it on and off in that entire time. If I can do that, then it’s really an insult to me that people are complaining about having to even manage to get above Very Poor.

Just try to be curious. Try to understand why these things exist. Try to have fun with it. Actually try.

People assume it’s hard and feel like it’s a gargantuan effort to do anything. They don’t have the energy to try to learn. They give up before they try. These people? I frankly am tired of listening to.

Just at least try to learn. Try to do something. Don’t just say “but Blender is hard”. Actually give it a shot.

[Plaster]

@dark

What is hard is making an avatar from scratch, primarily because of facial features and visemes/expressions. That is hard. The rest? Not to hard.

I hope people understand that nobody is forced to make an avatar from scratch, and learning the basic of blender and unity don’t require that level of knowledge. It makes editing avatars in blender seem daunting when it really isn’t, I know a lot of people that are simply modifing their favorite avatar base and are making some incredible things with them.

Does making an avatar from scratch teach you the ins and outs, including what impacts the avatar performance and how to improve it? I mean yeah, sure it does, but so does learning the very basics in modifying an existing avatar base. It doesn’t have to be that complicated, and it is best to not make it get to that point if it doesn’t need to.

And I say that having made my avatar fully from scratch.

Find a base you enjoy, modify it, have fun with it, don’t be intimidated about learning blender or using unity to upload an avatar, and if something seems unclear or confusing, ask people that do know or look up tutorials.

[calliemaxine]

Yeah, starting with retopology is a great way to be turned off from blender forever haha. I feel like the XKCD comic is relevant here.

In my opinion, if you wanna start learning to optimize? Start with the simple stuff you can do in unity! Theres some amazing community tools that make the stuff that can be fixed in unity easy.

You may never even have to open blender if you’re lucky. If that’s not enough, deleting edge loops in blender and decimating a few polygon dense areas (hands and legs are often a good place to start) is often enough! Getting the model back into unity can be… a bit tougher, haha.