Changing Performance Limit for Medium and Poor (PC Only)
Frajs
As of right now, it is very unfortunate that we are seeing increasing amounts of avatars that are both popular but also are made in not the best interests in regards to performance.
Myself I always attempt to make sure whatever I make is not a big offender while still trying to keep up with things. I know how to make a Good rated model but have unfortunately ended up realizing that the current set of limitations we have are not quite as fairly adjusted for the creators.
For someone starting out as a creator, the performance limit sheet could very much be treated as a ladder for anyone to climb and get better, but sadly the limits for Triangles/Polygons is not doing this any favors.
It has become what is for many people, a brick wall that shuts the door on any newcomer and instantly labels it as a Very Poor creation despite the fact that in many cases where some beautifully created models that slightly dance over 70k gets treated exactly the same as a model with 700k which is really sad to see. Cause 71k is really not the same performance impact as 700k.
And at this point I see it more and more often that people are just simply giving up on this alltogether and creates "monstrosities" cause why bother when its all gonna be labeled as Very Poor anyway?
I think this is a very big downside and I would really hope we could instead get a change for limits to Medium and Poor to make them viable options for creators to reach for.
If Medium and Poor are realistic options for creators, we will without a doubt see an increase in usage for these ranks by the creators as they would in many cases try to avoid Very Poor rating and over time stretch to both Medium and in some cases also Good rating.
This would in turn over time have a positive effect on VRChat as a whole as the standards would be shifted towards more performant avatars with admittedly less "bloat" and a more sustainable gameplay experience for a lot of users as people would gravitate to avatars with better performance ranking than Very Poor who appears to be the standard today.
A tradeoff could even be limiting the texture memory limits as todays limits are pretty generous and I think we would still see people get by without issues.
Implementing these changes would also not negatively affect previous creations and in some cases would give some unfairly ranked creations better ranking than before, leading to a happier playerbase that would also be more capable of telling "is this avatar good or not?" compared to today where it is quite dubious and obscure information on this topic floating around.
There are a lot of nuances to it all but I believe for inclusivity and positive long term goals I think giving Medium and Poor a better spotlight than it has today would be a very big gain for a lot of the community.
I would be really excited if you gave this some thought cause I love VRChat and would hope to see it go well in the future too, and this would be a really big stepping stone in a good direction.
Look at the proposed limits on the attached picture below.
Log In
Chdata
I've been making avatars for 6+ years and I don't think it's hard at all to get below 70k.
I feel like people must be putting way too much stuff on one avatar to even get close to that.
Especially if using a lot of the nice booth models out there.
Chiimera
I think the issues you mentioned are relevant and challenging, but I don't think that these are appropriate changes for 2 reasons-
1) Unfortunately, avatars don't exist in isolation most of the time. A tricount increase for one avatar means up to 80 avatars in the instance seeing the same increase (compounding).
2) I fear people will max out at the next highest threshold immediately. We will see the same "performance rating bottlenecking" almost immediately.
That being said, I agree that the system DOES have issues and punishes people for going over slightly in one area when all other areas are highly performant. I'd love to see a sliding scale/numerical rating IN ADDITION TO the current system.
The other option, which you briefly touched in with the material slot reduction, and one I'd be in support of, is making other, stricter metrics to allow for increases in other areas (like poly count). For example, an avatar with lots of screenspace and grabpass effects can still slide under the radar as "good" despite being potentially less performant than avatars with many more, non grabpass materials.
Thank you for your insight and for the opportunity to jump in on the discussion! Optimization is something that we all need to be talking more about.
Frajs
Chiimera Thanks for the reply, I do think its good to actually have a reasonable discussion about this topic as the flaws we have very visible at the moment are very noticeable.
The desired end effect of my suggestions would be to have less Very Poor avatars by giving people incentives to lower poly counts on their models.
It is not an argument for great performing avatars to get worse performance.
The absolute majority of avatars in vrchat is currently ranked as Very Poor, and we would probably be better off by redefining that and give creators incentives to improve.
Cultivate more focus on optimized models that are Good, Medium or at least Poor, instead of most users running around in Very Poor avatars with basically no limitations.
Chiimera
Frajs
In that case, I would say we need more official documentation (or officially endorsed, if not created by VRChat itself) to enable people to hit those targets.
I feel that the wiki has tried to do this, but from my experience it is hard to navigate. Many of the official docs are outdated or lack information as well.
Although I understand your desire to shift more people into the "medium" and "good" rankings, changing the measurement metrics doesn't actually improve performance, and in many cases will permit worse performing avatars under the pretense that they are (falsely labeled as) better.
Castell
Chiimera, Most models today go around between 200k to 600k polys, cause 70k is seen as impossible by the majority of people, it makes the system useless since the other limits are actually reasonable. What Frajs is sugesting is that if we increase the ammount of polys per rank more people will try to conform instead of completely ignoring it like we all do right now :)
Frajs
Chiimera I think you might misunderstand the intentions here a little bit.
Or perhaps my explanation was poor, which I do apologize for if that was the case.
But the intention is to give people who usually make Very Poor avatars a reason to get better and lower their poly counts.
Its not intended to make good performers worse, but to make bad performers become better.
Give people a reason to care, instead of it being a reason not to care if you get what I mean.
Chdata
Frajs Eh, you do have a reason. Your avatar will be blocked less by others.