#avatar-optimization
1 messages · Page 36 of 1
Lots of people want a weighted system but that feels much more contrived rank labels are just artificial and kind of meaningless at that point
because on average its not that laggy
no they are more accurate
having a lot of bad properties is not the same as having one
You would no longer be meeting a specification
yea but thats not the point
the limits arent what cause lag
the combination of all them is
if u rank low on all avatars and they lag more then me then why should my performance rank the same for having only one of those categories being low
As the rankings are at the moment, they don't state your actual avatars' qualifications but instead if your avatar meets all the standards at the same time for any given specification
A weighted system doesn’t change lag either getting. Making a weighted system so a laggy avatar can be rated medium instead of poor doesn’t make it less laggy
its obvious that the other user has a much less performant avatar
no thats not the point
it wont be rated medium
im saying say you have excellent everywhere
but poor dynamic bones
u should be dropped to good or at least med
because overall the character is ok
obviously if u misuse dynamic bones then u should go to poor/verypoor bc of how over the limit the value is
but again to be ranked the same lag-wise as someone who has an even worse avatar
is what im saying this avoids
Diminishing returns system essentially
yea each value has a punish value
instead of it being like a minimum value
the lower the score the more punish it has to the overall
Dynamic bones are the biggest problem with performance in the game as far as uploaded content is concerned. Your basically asking if you optimize all the stuff with far less real impact if you can get away with being more heavy on the worst thing possible and get away with a better rating? No...
no that makes absolutely no sense
because it would still rank the avatar as poor
the lower the score the more punish it has to the overall
so if u go crazy on one property it will hurt u
but just because its a little lower ranked doesnt mean the entire avatar is
Obviously dynamic bones would have further impact in terms of value weight
yea but the idea is that a combination of low scores should make them much lower than they are now
Getting excellent in everything except dynamic bones does not make your avatar any less laggy. Lag only cares about your worst category
and a few low scores should reduce optimizatoin impact
Does lowering the refresh rate up dB performance
and no it doesnt
obviously having several poor properties
is much worse than having 1 or 2
which is my point
and obviously having very very poor things still impact performance significantly
also my point
which is what i was talking about and how it could satisfy both issues
If an avatar is laggy then they went over the limitations established to such a degree that a diminishing returns system could land them in poor/very poor even if everything else is excellent
^^^
If dynamic dynamic bones are bottlenecking your frame rate it doesn’t matter if you have 1 material or 500. They aren’t the bottleneck.
thats not what were talking about [2]
If an avatar is laggy then they went over the limitations established to such a degree that a diminishing returns system could land them in poor/very poor even if everything else is excellent
but if its not that low then obviously its not that laggy and it shouldnt punish them that far
and also again what this prevents is people doing the minimum possible
to make all categorys medium so they can be medium
when in this system they will rank as poor or lower for that
the point were making is the more something lags the more punish it has to the overall score
its not that each item is an individual decider
Nobody is asking for laggy avatars to be accepted, just for a better and more accurate rating system that works in a weighted manner instead of absolutes of 1's and 0's
so the more laggy categories you have the more punishment your overall rating gets
and the more lag in that category that punishment is multiplied
If you have a certain number of dynamic bones you will be rated poor or very poor. It doesn’t matter how well optimized any other aspect of your avatar is. Each limit was chosen in isolation. Being good at one thing doesn’t earn brownie points to get away with worse in another aspect. Lag is lag.
yes but again
that makes no sense that argument makes no sense
and it is completely irrelevant bc as we already stated
if its that laggy then their score will be lowered anyways
the point of this system is to make sure that people who have laggier sets of components and more of them get far more punished then people with a lot less lag
it has nothing to do with 'oh ok theyre only bad here so give them excellent'
it depends on how bad
and how many bad
Dynamic bones = bad
yea lol
Dynamic bone collider = also bad
Preddy much
but the argument we making is that with new system its more accurate at measuring lag / optimization
It can't measure lag, because it's not dynamic and it's not relative to your PC specs anyway
The values may change, but the way they are set now is an indicator of where they'll be set to if they were to change
i get that but read the above points we made
There's reasons for all these values, they weren't chosen randomly
i get that [2] but read our points
there is no 1 deciding factor in the optimization anymore
its a collection of punishments
Yeah, because they add up too
the more punishments the lower the score obv and the stronger the punishment the lower the score again
yea but my avatars sometimes get poor just because of one or two poor properties
and got the same rank as someone with almost everything in the red
That's how it works
Personally i'd prefer a weighted system like such myself as it seems fairly reliable and fair for the end user
so why would i be ranked the same
when clearly my avatar is much better optimized
Me and my friends were playing with a specific laggy avatar last night. It was fun but we all agreed it didn’t have a place in a public instance
Because dynamic bones are bad, if it has dynamic bones it isn't really performant
Regardless of optimization
If you want to keep them then you need to understand that they're not performant, you can have a version with and one without
Some clever weight painting can reduce your dynamic bone count drastically
medium/poor for second to last
bc that way people who get a low score on less categories arent punished as bad as people who get low scores on more categories or people who get lower scores on fewer
cloth and dynamic bones being weighted more than the other components
its not just about dynamic bones thats not my issue
i just dont like how it slumps people into undeserving categories
when this is a more accurate estimation
If DB is a problem for now and the future, why not changing for Graphics.Blit Scripts or something else instead of continuing using the DB ?
https://vrchat.canny.io/feature-requests/p/graphicsblit-scripts
like everything on an avatar of mine for example is good or excellent and 1 thing is medium with another poor, should be at least medium right
Sorry but your system seems less intuitive Wulfe. It feels arbitrary
wym
its designed to punish users who make less optimized avatars not users who rank like one of their properties low but not significantly enough to cause that much lag if any
Each category of the ranking system was chosen in isolation. For example when they decided that 50 dynamic bones was medium, it was done under the assumption that every other aspect of the avatar was excellent. 50 dyn bones alone makes it medium.
yea but then why would cumulative low scores not put people lower than someone who has very few
compared to dynamic bones any other optimization is already negligible
yea but thats not my point
u can still weight cloth and dynamic bones respectively to offer more punishment
but if someone has way too many materials way too many DB, way too many cloth verts (all low but not very poor)
i dont see why they arent ranked as low as possible
and why im on that same level
If dynamic bones were rated poor with everything else excellent, it would be worse than dynamic bones being rated medium with all other things also medium. What you are saying would eliminate any drive to improve the avatar
Imagine that an avatar literally doesn't have dynamic bones, got a single value categorized as poor, whereas everything else is excellent, they're automatically considered poor
^^
People would just play the min/max game and keep their laggy dynamic bones
Imagine that an avatar literally doesn't have dynamic bones
excuse me
also why would an avatar with poor everything be as bad as an avatar with poor in one category
doesnt add up
also as u said 'they were chosen individually'
And that's why it's based upon diminishing returns, no amount of minmaxing would let you have a laggy avatar due to db
so that means medium in another category is also actual medium
so unless the developers had no idea how much lag things caused in retrospect
poor should lag nearly as hard as another poor
otherwise i dont really see the point in them having the same rank value
i mean u could always just set up control avatars
where u increase everything to its limits until it lags as much as another property
individually
Dynamic bones is being used as the crux of the argument whereas it'd obviously be weighted the most and of course have the largest diminishing returns
it will still function as its supposed to
but the thing here is
medium isnt going to be a medium punishment
the punishment for medium is actually lower than medium
which is why it can rank users to poor and very poor for having terrible ratings
even if none of them are in that rank
Wait wait, how does this rank thing work again?
The established one or the one we're talking about?
The established version
If any variable goes over a treshold you get automatically placed in such regardless of any other optimized sections
I.e. Excellent everything and Poor polycount as an example lands you on the Poor category
Because it’s less of a rating and more of a specification
Perhaps the problem is the name
no thats not the issue
the issue is that it is a specification
and not an actual rating of who is more optimized than who
You understand what meeting a standard means right
yes but again that is not the point
ur clearly missing the fact that there is way too much innacuracy
and like the majority of avatars currently in vrchat are poor and medium just because of this issue
when in reality a lot less of them are
its just the system thats not that great
Essentially you are asking to overall, make the rating of the avatars fall
Because that’s all that would happen with your changes
The problem is that people are seeing it as a rating whereas it's actually meeting a standard, which makes many users annoyed or even relunctant about others' avatars even if they're in a general sense very well optimized i suppose
So wait hold on, depending on the type of unoptimization on a person, does that mean you get demoted in a way?
literally no argument u stated is relevant to the solution
no basically
the lower the categorys in total the stronger the punishment
and that punishment is multiplied by how low the category is
current its like like a standard
its not set up like this
so having everything low gives you an awful rating
having a few things low gives you a less than great rating
and having like one thing low doesnt take off too much
If this is for meeting standards i'd say keep the "rating" for the user not the public, so they know their "rating" as it stands it's essentially a ranking system that anybody can flip out over
because its true that avatar is pretty well optimized
compared to others that might share the rank with the current system
Oof. Atlasing
yea but that wouldnt be necessary if it was more accurate
also if there was a way to judge how much performance an avatar uses up that would be super cool
but i dont think things can be measured like that
all avatars load their properties synchronously so
That'd be pretty neat yeah
u could also set something up like if someone uses a gesture and it lags it could show in a log
or someone loads and it lags
I’d personally like to know the stats of any avatar I wear, as well as those in the same world as me worn by others. Rating ain’t important to me@but I do want the stats
actually a log would be a great idea
a log of like particles when they are enabled
I'm all up for statistics, but this is essentially a be-all end-all rating system, if it's for meeting standards, keep it to the user
and objects
We currently have that.
that shows up in the game
Yes
for vr users also
I can deduce rank from stats anyway.
yea but i mean like
a log where particles, sounds, objects, and their owners are all logged so u can
see who is lagging the game
or who is spawning all those particles
and click on the perosn to block them
Sounds like overhead and a cause of drama :x
or to see their menu
Yes you can, but at that point it's the user deducing it, not a system selectively stating it out loud
not really u can just listen to the events locally its not that much overhead
and also it would be easier to find crashers
and u could also make it report avatar cloning
in case people clone an avatar u blocked
or if it could block it automatically like in the feature req
Oof I don’t like the direction this is going. The more stuff you add like this the more drama there is and people stop having fun and just start going meta talking about stuff like this in game
what
none of these should have a negative impact really
possibly the rating system but not the other suggestions
im sure a lot of people would like a log to help fight lag, spam, and crashers and avatar blocking to prevent cloning from ruining the game
People are getting butt hurt about a system designed to help them optimize their development work. After release of it my personal avatar passes for excellent now and loads from blue man in less than a second. So the system is doing the intended job by helping.
yea i get that but
were not saying the system is bad
were saying its inaccurate
if u have an excellent avatar good 4 u but
Does the system require other users to be aware of what the end user needs to know (the ranking), aside the avatar statistics for such?
Other than the ranking i'm all up for optimization and avatar information
I don't really think it is inaccurate. There is no accuracy to be had since it has to take in account all computer types. Beefy CPU with wimpy GPU? Wimpy CPU with beefy GPU? Number of polygons will affect those differently.
yea but thats not what i mean
talking about multiple low scores
vs few if any
the ranking system says a lot of these similar cases have the same performance ratings
when obviously the former is much worse than the latter
when u look at stats urself
ur using the same process were talking about
how ur weighting categories and checking how many are low
As the ranking establishes standards to meet i think in my honest opinion that the end user should be the only ones seeing it, as it causes drama over something that can be easily explained and passed over by looking at the actual statistics
independent of the avatars actual ranking
and ur like 'oh ok this avatars actually not that bad'
that process is a more accurate estimation
Gonna be honest if the rank wasn’t public I wouldn’t have optimized. I wanted my green star and I wanted everyone to see it
If it was to establish a visible ranking then one would expect it to be actually reasonable in terms of acertaining the actual values
I like having the icons pop up over player's name plates. Teleporting into a room and seeing the world start to jitter I can immediately find the culprit to disable their avatar.
well having a green star kinda sucks tho bc it means theres not a lot in the avatar
if u rlly wanna make it look good the rating system typically slops you into poor
but again thats bc of all the cool things u put on it
which is somewhat reasonable=
only if the majority of the properties are low or any very low tho
And then you see very poor avatars that literally cause no lag over a single variable, bam blocked
yea dude this tank was ranked poor but their cannon
instant 4 fps
mine is ranked poor
buuuut is it really..?
I optimized a single avatar i use just to avoid witch hunting and people losing it whilst my avatar never caused any issues to anyone, even in a nigh full room, that is the only reason i did it fear, over criticism and possible accusations
That's what the system has brought upon me
Not to have a shiny green badge or be seen as a person that optimizes avatars, the old ranking system still hates me to this day even though i never went over the polycount before and after its inception, and caused nobody any distress, so i inherently dislike this system aswell, another label to tag me, i'm honestly getting tired of trying to be optimization friendly as it keeps feeling like it doesn't even matter anymore
I mean... promoting the concept of optimizing your avatar is a great idea for it helps with performance on lower end pcs and overall performance for everyone. Question is if it was a good idea to use this system to “motivate” people to optimize their models
For some, yes. For the trolls on a certain avatar website... It has the opposite effect.
Guess i'm not one or the other there
I'm all up for optimization and helping people in that regard, especially promoting it with methods that retain a models' original concepts in mind, i however, am not in favor over this system at all, and if anything, it demotivates me more than it informs me
My avatar looks identical to before the rank came out, but now it’s rated excellent instead of poor. I had a ton of hidden geometry, I was using custom animators instead of idle override, my bounding box was huge, and my avatar was split into like 12 skinned mesh renders for no reason. So yes this system helps people who have no idea what they were doing and had no idea they were unoptimized.
Finally:
Are you saying there are ways to make the count be inaccurate?
known since closed beta
Because it’s not measuring lag
If you know enough to intentionally trick the system then you at least know your rating is false. You are aware your avatar isn’t suddenly optimized and less laggy just because toy tricked it. So the system isn’t really for you.
peopl are doing it on accident
Then that’s a bug
yes. its inaccurate.
I think inaccurate is different to a bug
its inaccuratedue to bugs...
extra mats being read is another ive heard of. no particles, nothing hidden. read 13 and not 6 if i remember right
Oh right that happens when a user reuses mats and it tries to count the total mats at times and at times the actual used mats
I've had it jumping between 2 and 6 on my model
basically why i dislike the ranks, besides the lack of weighting. its broken
big bounds due to pen but excellent everywhere else? very poor
most users dont undertand what the numbers and stats mean. they see icon and react
SDK or client? Or both? The SDK being super inaccurate is known
both
i was refrring to ingame though. for this convo. forgot the sdk had one since its worthless
ah that too. i believe in some cases the other users see the "true" or corrct stats, but not all the time
one spoof was undetected in testing we did while others worked clientside only
I just have a feeling this system is going to be used to test the general public consensus on avatar optimization and build a new system based on such that blocks values over the norm
It's guidelines for now but that may change in the future
well the patchnotes say that
somewhat in the polycount part
but its probably more just to help with awareness. i hear a lot of "wait tht number matters?"
especially db script stuff
I mean atleast it does good in that front, but i still would much rather see avatar details indepth rather than the current system, that sort of encourages blocking of users without much of a second thought, and that's not even that far of a stretch
Only thing i don't like about the system?
The ratings, that's about it
Avatar details are a complete a-ok for me
Devils advocate: if someone blocks for such silly reasons would they really have been worth interacting with
Yes actually, many people have that kneejerk reaction as they may have had many previous bad instances with bad users
And may be completely civil people
And I’ve had someone say they will block excellent for being boring. You can’t counter crazy
Never heard of that one before, but everybody's different to an extent, so it's not entirely surprising
Different is a nice way to put it :p
Eh, true
But do keep in mind i completely understand the system, i would just prefer it be implemented in a better manner, or one that wouldn't encourage such behavior from users that either don't want issues from avatars from past incidents or even from warnings from fellow users
It's just my personal opinion on the matter anyway
Well I have no preference. I wouldn’t wine if the public rank went away and it doesn’t bother me. If someone wants to block poor avatars they will be very lonely
its not a sily reason to them, theyre told tht theyre being lagged
by the description andd everything. to a less knowledgebble user its a good reason to block
Well what I am saying is they will be blocking like 70% of users
Let them be dumb x)
uhhhh or yknow
Most people have common sense
fix the issue thats misinforming users
Well I suppose. I just wonder if this is a lot of worry over something that just isn’t happening
I’m just not hearing people saying they are getting blocked
I’m not hearing any complaining about rank
any spread of misinformation is bad, and yes, people complained here
Or even discussion about it outside of this discord
Not every user that's been blocked is going to start talking about it, especially when the system is being so vehemently protected
Plenty of discussion about it in the creator communities. You wouldn’t care if you weren’t part of those
I do think the system should be dynamic. For example if a prop would make you poor but it’s inactive, don’t count it unless it’s inspected while active
im not even part of any and i hear itout of discord plenty
That’s the point you only hear it on discord
You don’t tho
"out of discord"
It’s worrying over a thing that isn’t gaping
They just said they hear it out of discord plenty
stop, please. youre literally just wrong.
Well alright then I just don’t hear it I spend a bunch of time in public
How skilled are you at making avatars
How do they even know their avatar is being hidden
its a topic i hear almost nothing good about. its inaccrate and misinforms.
^
Well, there's 2 kinds of block i suppose, the avatar sense, and the literal sense
I hear plenty about it in worlds, vrc canny, reddit, and discord
I mean I would expect people to drag their feet wanting to optimize and hate the system
the potential for it to be good is the nly positive thing i hear
It’s like being told to clean your room
I understand the need for optimization but it literally doesnt measure optimization
Of course no one likes it
That’s the problem
Please stop playing devil's advocate with such poor points
People dislike it because it's inaccurate. People dislike it because it's misinforming. People dislike it because it's poorly weighted.
Yes, there are people who just don't want to optimize but not everyone with a Poor or Very Poor avatar is a world maker with 2500 collisions
some are mutes with markers, or people with otherwise very performant avatars but too much of something that's not even significant
It limits creativity by misinforming people who don’t understand that the guidelines aren’t reflective of actual optimization
Well everyone I work with is just all “I’m not getting rid of my colliders”
To correct your analogy. it's more like people being told to clean their rooms because it's filthy, but some people only have a few socks laying around while others are sitting on week old food
“Wtf why did you Atlas this”
You haven't been working with all that many favorable people then
No they are great
I still think, renaming the ranks (along with better weighting) would help the situation...
Theeen...
More people are simply uninformed.
It just needs weighting
They just don’t know anything about optimization and are mad they can’t get their stuff in a public world pedistals
Oh that's a world issue then
I've ran into SO many people who just don't know and never knew that dynamic bones were such a big cause of lag. WHy do yo uthink this is?
Oh, so they're uninformed cool
They are informed they just don’t care
I don’t think you understand the legitimate issues people have with the system adeon
"they don't know anything about optimization" so they're uninformed
They are taking about making clonable alts to just stand here where the pedistals would be
10 clients, all connected tothe game at once, eating up your bandwidth and PC resources
If they’re that dedicated to an idiotic idea then let them
So your PC is now rendering a dozen or more ybots (assuming they've got at least that much brains to block avatars), on top of any users that join 10 times
so you're rendering 120 ybots, the world itself (lighting, mats, etc) 12 times
Well that is my main brush with those who hate the ranking system
If someone has the resources to do that then tbh more power to them. the 99% of people do not.
Stupid, uninformed people? You did nothing but prove my point.
How do you suggest to improve it
The ranking system should exist but it needs weighting
The same way everyone has been saying. Weight the system properly and make sure it scans correclty.
I don’t see how weighting helps
....
Because ur misinformed
^
If it's weighted properly, and applies the diminishing returns system wulfe recommended it should make alll the difference
Will the laggy avatar be rated better
No
"laggy avatar" stop
Nope, they wouldn't
Then what does it change
You're proving our point. you equate Rank to Lag
Infact, they'd be rated more accurately
Very Poor = / = Lag
The point is that some avatars that aren’t laggy get marked as poor performance
With weighting that would stop
Mostly
For the most part yeah
Let’s be real the majority of poor avatars are due to insane dynamic checks
If not that then materials
True
So is that an argument against a properly functioning system?
Weighting will change nothing people will still hate it
We don't know the true stats, but it sounds reasonable to assume such
EXCEPT WEIGHTING IS WHAT EVERYONE WANTS
You can't see the forest for the trees dude
Talk of bounding boxes and such is not why people hate it. That’s an edge case that will make YOU and a few other people happy
Me? It does'nt affect me whatsover nice try though
MY bounding boxes are always Excellent. Lots of people use pens etc
Most agree that the system should exist. We just argue that it should be accurate. And it’s happened to many avatar creators
You're willfully being ignorant and doing nothing but proving our point.
Open your eyes. If things are properly weighted then the people with otherwise performant avatars won't be ranked as poor/very poor.

And it's supposedly getting a weighted upgrade in V2...
The main reason people dislike the system is because
It's weighted improperly.
It's inaccurate
Or it exposes their unoptimized avatars with 2000 collision and 50 colliders.
You could have the best optimized avatar and 1 marker on it making your Avatar very poor performance. That is an issue
People arguing against the system just don’t understand how avatars work
Saw a avatar where every loose part was it’s own skinned mesh. That’s the stuff that needs to stop. And 300+ dynamic bones with colliders
Same goes for people with 1 poly avatars and tesselation shaders on them that lag entire lobbys
People will just do the minimum needed to min/max. The system is strict right now, weighting will be far too lenient. If you tell me I can make my avatar slightly more laggy and stay excellent of course I’m gonna min-max as much as possible
Bruh u just don’t understand how avatars work
Go for it I guess
You don't understand how numbers work.
Ur proving ur ignorance
You say it’s more accurate I don’t think so but if I am wrong and it’s more accurate than yes it’s better.
Diminishing Returns, if something goes way too over the limit the system will amass the weight into the negative zone, so a laggy avatar will still be considered poor or very poor
Your reasoning as to why it's inaccurate has concrete proof to being wrong.
Ok last attempt
Just to be clear adeon... The first day it launched, tupper and aev were saying that it was planned to be weighted, but it will be implemented later. So it's not an invalid point to ask for weighted when the devs themselves want it like that
Honestly maybe it should just measure draw calls
Look, I pose a simple question to you.
Are you telling me that you believe Very Poor Bounding and 256 Dynamic bone collision count are equal in the performance hit?
I'd like a simple yes or no, there's really no other answer to the question.
This isn’t about what your avatar is rated. It’s about how laggy you can make your avatar and still get a given rating. Right now the system may be inaccurate; but it’s inaccurate on the side of being overly STRICT. What will making the system less strict do? Make the average avatar less optimized.
What...
Wat
????
Alright, one more time here
That was my last attempt.
From the very start we've been talking about the rating system
I found an amazing way around the system. I just optimize and it never catches me.
I’m tired of trying
I've got 10 avatars. They're all ranked very poor. All for different reasons.
You can't succeed Adeon because you aren't correct.
Weighting is implemented, and suddenly some are good, some are Very Poor, some are Medium. How are any of those avatars more or less laggy
You can be optimized and still have the very poor rank tho.... that’s our argument
You can't even seem to define "laggy" when also arguing your point.
That's because the system isn't finished yet. Which is why weighting and other things need taken into account.
He argues against proper accuracy of "lag" detection because he thinks it will magically makes avatars more laggy
Right now next to no one is trying because no matter what you do, one point can just drag you into the shitter.
Excellent avatar but you have a snail pen? Very Poor. Good avatar but you've got a snail pen? Very poor. Medium avatar but you've got a snail pen? Very Poor.
It's like lets say I make a really amazing map that renders in benchmarks amazingly well. However it runs bad when being used cause everyone cast shadows and the world renders twice due to being in VR per lens.
you can "detect" it's lag but it's inaccurate
Proper weighting and you'll probably hit Medium or Good assuming no other factors, and with proper weighting those "other factors" will be accounted for.
It’ll be a more accurate comparison of lag between avatars
I’m not arguing against weighting. I’m arguing against more relaxed requirements for a given rank.
If you weight in a way that does not let you make an avatar with excellent everything except dyn bones suddenly get a better rating than I don’t care.
If your avatar still ends up being "very poor" due to the weighting, it'll still be "very poor"
These people are not asking for less strict requirements
@amber hemlock then you aren’t arguing with anyone here
That....that's what weighting is my guy
Weighting is literally "Dynamic bones mean more than polys"
I am interpreting weight as in “weighted average” right now, it’s just a floor round
It's giving things a certain value in terms of "weight" in which they are accounted for "dynamically" which makes it more fair as you actualy get accurate values
Right now excellent except for all but one very poor is very poorly. That very poor counts with 100% weight. You are saying it should be less than 100% weight
Several people are typing....
Try and "cheat" by being excellent in every category but go overboard wtih dynamic bones? DB would be weighted higher on the scale than everything else to begin with, meaning your avatar will be ranked accordingly
wait .. we weren't talking about weight painting?
🤔
lol
Lol
We weren't?
🤔
And the fact you're Excellent in every category BUT DB means you're not as bad as you could've been, even if it does still hit you to Very Poor
Right now your worst category counts as 100% weight! In a weighted system, it would be less!
We could switch how we are discussing this. Lets say every bone causes another 1ms of render time for arguements sake. an avatar with 200 bones wil render at 200ms vs a 65 bone avatar at 65ms. If every dynamic bone causes 10ms per transform it effects. So if you have 5 dynamic bones on both those avatars. one will equal 115 ms the other 250.
@amber hemlock they are not asking for a less strict system. They want the categories within the system to have a more accurate weighting to them depending on how much the individual categories affect your avatars performance. So say you have only 2 categories, polygon count and dynamic bone count, the influence of the amount of dynamic bones should be “weighted” higher, compared to a slightly higher amout of polygons. Overall this doesnt make the rating system more strict or less strict but it makes the individual categories more accurrately “punishing” depending on how much potential influence they have.
Also, please correct me if i got something wrong*
Sounds about right to me, just want accurate rating based on the total weighted value of every category
Well there u go thanks EXS 😃
Instead of a 1 decides all sort of deal
Yeah, if anything that would cause people to optimize better because you won't just be hitting Very Poor for something stupid like your pen and just giving up
Really, why bother if my pen puts me to Very Poor? I may as well go ham on my DB and stuff.
Or my poly, or anything else
PEn just being the main example I use since it's the most relevant one due to the amount of people using them.
Hoping my rigid body arm avatar gets a rank above very poor lol since it’s very well optimized but uses to rigid bodies not affected by gravity
I understand that but how is that not more relaxed in the case of dynamic bones? It just lets you get a bit more dynamic bones in because other things nudged your rank up
just put a keyboard on your forearm and point at the damn letters 😄
Does the bounding box size really make a noticeable impact?
Yes, the "lag" you cause is stil lequal....
bounding box size prevents your avatar from culling
It does in some cases, yes. It's culling related.
so it renders no matter where you are facing
Yeah i know that...
So if you're perfect in every regard but bounding it's no big deal
But if you've got a pen and 500 DB, you're never not culling those DB unless they're 20meters away (or 12.5 Tups as I discussed a while back)
It can cause some issues if there's alot to render constantly, but that's about it iirc
If you think it’s possible to make an artificial weighted system that fairly represents lag impact sure but I think a system like that is far too arbitrary to be fair and too complex to explain why people got the rank they did
It's actually the proper way to do this system and they even know that and plan on implementing it afaik
But everyone disagrees with me so I’m probably wrong
Essentially it comes down to what rank you want.
Yeah, I'd rather have people trying to optimize, even through min-maxing, than no one trying because something shitty puts them at V ery Poor
But at the moment its pointless to optimize because if you want f.e. a simple marker, you immediately get very poor even if all other categories are well optimized.
Yeah I agree that sucks
But they really do have a huge bounding box
Of course its different depending on what rating you want
Just like they made an artificial category called collision check, instead of bounds. They could have a more reasonable category to rank against
But at the moment it doesnt matter what ranking you want or how well you optimize because 1 category thats bad, instantly gives you very poor like f.e. with a marker
They dont have artificial categories
One thing I don't quite get is why colliders matter if we have collision count to begin with
Bounds and collissions are 2 different things.
For example multiplying the bounds by some other aspect
Idk
But there may be a reason for that since DB is actually something I'm still learning new things about
A lightweight avatar with huge bounds is clearly not what was intended. Or perhaps; allow a better way to have markers
Colliders woudl probably be weighted fairly low since the collision count,as far as I know, is what matters.
Only using colliders to touch your ears? Thenyou've only got like 6 collisions at best
Here's an example of the material count being a bit on the nose, i have 2 materials reused on the entire model but it accounts the total ones
IMO dB components doesn’t need to be counted. Unless there is something I don’t know about
@amber hemlock even so, the fact for having a lightweight avatar with f.e. 9/10 categories on green should allow for at least some compensation on other categories
Inspecting the avatar with a second client
Lol
Yeah Components and Colliders seem redundant when there's Transforms and Collision Count, but hopefully Tupper swoops in on this one hint hint
There's only 1 thing i miss from 5.6.3p1, the double jumping due to the tiny lag it incurred
It was neat
I think I get it. I was just seeing dynamic bones as the only real source of lag
They're the main source, as far as avatars go
With any average being a way to min max a little bit more in
I've heard IK scripts outside of the main game's IK can be bad too
but they're pretty uncommon
IK is the main source of lag from what i recall, but nothing can be necessarily done about that
They said they’re improving it. They intended to do it in the last update
Mhm, it kind of went south, but hopefully it'll get improved
Yeah I think it was IK, on top of chairs constantly moving, on top of this scan actually scanning every time an avatar loads and not "applied" to the avatar on upload
out of those three, IK was most likely the worst offender by miles
Oh correction on my part, nothing can be done about that by the user
Yes they can. Block full body users lol
feelsbadman
Once the IK is fixed we can go back to like 20+ users with next to no IK related lag
I'm still under the impression that, until we get an in-house DB script, the IK fix on the network is gonna be the last significant performance boost we'll see for a long time
Since there's not much else to improve on as far as FPS goes on their side
True
I mean in general lag applies to a lack to resources to maintain what is currently happening, so if the IK gets improved, lag will dissipate exponentially, even with DB's that otherwise would increment it alot
Im most excited about it
Hopefully it'll land nicely
Yeah, I couldn't care less about chairs or ranks or whatever
the IK was what I was looking forward to the second I saw those patch notes
Yep
I was more looking forward to stats personally, as i like knowing what's going on with my avatar, but the IK was certainly up there on my list
I tried it but my visemes got fucked
I could've sworn that instances when created were essentially p2p, so the "host" gets transferred to another user, sharing that users' network speed, i always thought that was the case since users managed to stay in worlds talking with other users without connection to the VRChat servers, or atleast the repository for their items such as worlds and avatars
Some things are host related I think, like streamed content
Instances still linger for a bit after everyone leaves though, so some of it is server-side
Be nice if u could host instances on your computer so you can have worlds that can remember users for leveling and stuff
Block knuckle controller users excessive IK use :p /s
Hmm, would it be IK since it's not skeletal data? I'm guessing that'll be an option but i think it would jsut be animations
The ik fix made you go cross eyed for everyone instead of just you btw
What does "Could not load type 'Unity Engine .XR .XRNode' From assembly 'Unity Engine" mean?
Yeah that's actually what I was suggesting even though I knew it was impossible due to their vacation hiatus
Just remove ranks for now, and leave the numbers. They're at least somewhat accurate.
Once they're actually accurate, leaving them in makes much more sense since they're not false representation of optimization after they're weighted
The fact is a lot of people don't understand what some of the numbers in the Stats are and why they matter, and to what degree they matter, so the symbol is somewhat important, if it's correct.
@shut sequoia Redundant/extra/unneeded components have a non-zero performance impact, so that count will likely remain
@mortal violet You're probably using the wrong version of Unity. You need 2017.4.15f1
;_; muh scripts
That is too general purpose. You have to look at the code for DB.
Generally speaking, splitting up scripts to separate objects isn't a huge deal, but with DB, the cost is higher than it is with some random C# script.
Then tests need to be done in that regard to acertain such
I’ll take a look at the code and see what I can gleen. I only know pseudo code atm so probably not much
I like my green star. Don’t take away my green star D:
My cute hair is worth the medium
i cant find Unity 2017.4.15f1
@mortal violet
Direct DL Url (64bit)
https://download.unity3d.com/download_unity/5d485b4897a7/Windows64EditorInstaller/UnitySetup64-2017.4.15f1.exe
Is it actually true that DB has significant overhead per extra script?
I'm looking at the scripts right now and I cannot see any expensive stuff outside of the "for each dynamic bone:" loop
Everything before that is just setup, and everything inside that loop will be done roughly the same number of times total anyway, when comparing one root bone or individual strands.
Well it's nonzero, but Tupper was saying earlier that DB had more overhead compared to most other scripts
The only "overhead" I can see is the script checking if it's time to update already yes/no
In case of lower update rates
Well, it has some overhead for gravity checks for some reason, but these are all pretty cheap operations and they're O(1)
So it's the same relatively low cost for every script
With two scripts, you'd have two loops going, wouldn't you? How expensive is each iteration of the loop?
I thought it was the colliders that multiply the overhead, not the actual script
You can often put the db scripts one level up and exclude like 90% of the children like put it on head and exclude all the eyes ears etc
I'm able to keep the scripts down to 4 on most things I tried though on one or two I had to cut out a piece of clothing because it wasn't in the right part of the heirarchy
Also it wastes a transform because the head counts as a transform even though I don't intend it to move
can we pin the link to the model chart ?
the colliders arent too much overhead
each dynamic bone script goes down the bone chain
calculates a few things based on colliders iteratively, stiffness, and radial distance
and also simulates spring forces
but if u have like
4 colliders
that means every bone in the chain is gonna handle 4 colliders
or at least check them
I mean the same calculations are going to be done (to an extent) with multiple scripts (that is a script per root of chain not multiple scripts per chain), and will infact be far more reliable (aesthetically/functionally speaking) rather than a singular root as dynamics bones pivot based on the root bone, having a single one limits a users' freedom for using DB in general as they're mostly restricted to some set values as the single pivot can't deal with all of the possible angles the same way
Some actual testing between single root and multiple roots should be done to actually acertain if it actually causes any sort of reasonable frame dropoff to well, have such a bad fame of being done as it's adding more components into it
Started working on a new optimized Pen today. Works and uses way less vertices than a trail render. Issue is it required me to use a fixed joint/rigid body. Meaning my Excellent went to Medium but that's still better than the Snail Marker's Bounding box.
I have a few other ideas to play with and will see what comes out of it
Medium is still pretty decent for high end computers.
At least from my informal testing.
Why more then a controler drop to good🤔
Medium is alright, Good is more of a goal while Excellent is a fancy little goldstar
i cant make girl avatars so i will ask can somebody that want it make a girl avatar for me?
They should change medium's name/icon to be less harsh maybe lol
Since it seems to confuse a few people
@heady smelt If you have a budget, join the VRC Traders discord in #community-servers-old
No one here will make you anything for free
I finally did an atlas! Now I'm optimized!
how do i go about combining these pesky skinned mesh renderers?
ah nvm got it
i wish the CATS texture atlaser would atlas diffuse and normal maps at the same time
I do too
Pmx files don’t use those kinds maps 😭
Update material combiner and Cats
I am a weeb though
@surreal topaz just updated, it definitely only combines the diffuse textures, to get normals you have to re-open the saved blend file and delete the diffuse and set normal maps to be diffuse
You might have to use material combiner itself rather than going through Cats
While it takes longer hand atlasing will usually give you a higher overall quality, especially around the eyes and such
Unless cats really overhauled their automatic atlasing recently
yea i got mad when the combiner did that too
but apparently all the images keep their resolution
so i could just copy and paste the normal maps into the atlas
and save a copy
but yea there need 2 be better way
@fervent lintel Cats uses a new atlas tool now, it's so much better. You should try it out
Is it the equivalent of hand doing it now? I use cats for most stuff but I like the fine tunability of atlasing by hand
@sudden zodiac
Cats automatic atlas is lossless iirc
Except for Textures that loop over their bounds or are very close to the edge of the Textere. They may bleed into the adjacent Texture
But technically they were already bleeding over to the other side of the texture
yea the current atlas tool is a lot different
the old atlas tool would bake all the diffuse to a texture
more reliable but u couldnt do it on a per material basis
the new atlas tool just maps all the uvs and images to different sections in a transparent texture
it also makes the uvs/textures neater than random unwrapped mess
how transform are calculated ?
In terms of dynamic bones?
yes
it's how many bones are animating due to dynamic bone scripts
if you put them low level, all their children will be animated and the count will go way up. put them only where needed and add exceptions for stuff that doesn't need to be animated to keep the effected amount of bones as low as needed.
Also, if a chain doesn't really need to move from it's parent, consider merging bones in blender.
just look at the script for it
not everyone can read C#
you can explain that in english
yea but its spring physics and simulated forces/collision
I'm fairly sure he was asking how the "dynamic bone effected transforms" category of the avatar ranking system is calculated
not how the DN bone script works 😛
transforms are how many bones the script is considered to be affecting
So if each color is one script, it (should?) work like this
Red is affecting 5 Bones, Blue is affected 4 Bones, and Green is affecting 3 Bones, so total you've got 12 transforms.
Lots of transforms = Bad since dynamic bones are the biggest influence players have on frames
I made a video on fixing some texture issues and some other issues people may run into when merging textures https://www.youtube.com/watch?v=LGg3fd98T1k hope its helpful
YES I know there is some mic noises caused by my phone on the table. Sorry I was not about to rerecord a hour of work just for like 10 sec of bad audio
also actually it affects any bone with a following bone afaik
unless u use end length
You're never gonna want a script on every bone though @shut sequoia
Only on the root bone, once
13 bones per pigtail, 1 per bone lol
Wtf
not something i use in public since its a billion transforms
lik 190 total for the avatar
but the hair is cute
:0 oh is that what that does
If you have 13 bones per pigtail, you can get it down to 13-14 transforms total and keep the same effect
By just having one script
Yeah, curves can alter the dynamic bone values the further up/down you get
oh well shit
So in this case, I'm assuming you want less inert and less stiffness as it goes further down to get the same effect
i couldve been doing that this whole time
oh damn, literally just make a line diagonally down and its the same
Oh wait maybe not, it's kinda weird
ohwell, i'll mess with that and test it in vr later
It might take some fiddling, and I'm willing to bet you also need to draw a line like that for elasticity
Starting at like 0.1 and going down to 0.01
You can right click a key in the curve editor to change the values more precisely
In case the editor freaks out or you need precision
Actually, curves are indeed effective for those general purposes, doing on a per-bone basis does give you far more control over the entire chain as a whole (as every value is editable without a varying curve that may or may not apply to them properly at any given time) but i consider that going a bit overkill
yeah 26 bones with their own scripts is 100% overkill
It can be used of course, just like everything, but it is indeed a bit over the top haha
All I really do is put the inert in descending order from .9 to like .2, and the curve does that so i'm just uploading one as a test
i almost never use curve tho
only for stiffness when doing stationary root jiggles
Also you can go at a higher level and exclude a lot of stuff but you can't exclude anything in the chain to get there. For example I put dynbone on head and exclude almost everything and it will count Head, Root_Hair and Hair_1 etc compared to putting it directly on the hair
However I never knew about curves. How do you set up curves?
click on the box beside
Omg how have I never seen that box. Gotta do some experimenting
im having an issue whit the spine and i dont know how to fix it
if you read in cats plugin when you apply full ik it tell you no not wory about the warning
ive uploaded this avatar before and i didnt have this issue with this avatar before the new unity update
@swift crow u can ignore the warning
it wont let me upload it.
that is not true, screenshot of your error, because you can 100% upload with spine = 0
other error ?
aren't your legs connected to your hips? straight? and yes, what is the other error
is my model optimized enough? https://cdn.discordapp.com/attachments/277632761496469516/528418350309834752/unknown.png
yes
how do i refresh the sdk to show the proper amount of dynhamic bone components/
press arrow in top left
Hm, not working
idk if sdk can detect transforms and stuff
always doesnt show my dynamic bone counts
but basically
amount of bones in the chains is transforms
and checks is transforms * colliders per
but idk y it doesnt show
can i get some help in 3d modeling please i need help with automatic weight painting
you can get zero poly
and of course dynamic bone counts show zero because https://vrchat.canny.io/bug-reports/p/sdk-avatar-performance-reports-always-report-dynamic-bone-counts-as-0
you can have o polygone
yeah it's called disabling the mesh renderer
555,000 triangles. 90 bones. Time to add Dynamic Bone on it. :O
Does the public world bot detect deactivated objects on an avatar? Because the optimization stats ingame dont
it does from what i've seen
Okay so i did automatic weight paint for clothing i put on a model but parts of it are moving i dont want to move when the chest bones move. when i go to weight paint mode the entire piece of clothing is blue what the heck am i supposed to do?
come to think of it the entire model seems to be blue?
but it moves just fine in vrchat am i doing something wrong? am i supposed to select bones in weight paint mode? because i cant find a way to do so clicking wont work neither will shift clicking
?
anyone?
i don't think bumping your post is going to bring in any more people who know. if someone reading knew, they would've answered...
k i feel like it has before though ive done the same thing and someone answers
your best bet honestly is finding a weight painting guide on youtube
i got it
K
i coudlnt figure out how to select bones properly in weight paint mode
that was why everything was blue dopy dee dopy dye
is edge splitting models efficient? i have this bean can model im going to use and i gave it edge splits to make the normals more sharp, but i dont know if it's a good idea
Edge splits shouldn't be necessary, you can have hard edges even with a smooth UV
Can't you use the auto smooth and set a degree?
well i use auto smooth if it's 100% needed
if not, i manually do it
also, if i didnt use edge splits it would look like this, which imo looks ugly
Hmm, in edit mode you can select some edge loops and ctrl e to mark sharp
You can use split normals instead of a edge split modifier
avoids making extra edges
I believe it was alt click to select a loop
the object will still have a seamless UV
no i dont edge split with the modifier i literally just click a group of faces and split them
Erm. I think if you remove doubles afterward the result is the same regardless
you just end up with split normals
Yeah, I think marking the edges as sharp would be a better option
But I think you have to have auto smooth on for it to show

do not do auto weight paint for clothing
if there is skin
again use transfer weights
hi, is the proper solution to be able to select Shaders when using an Atlas map to set the .fbx to external materials mode?
cause it just kinda makes it turn purple when I try to select a shader after that >_>
oh wow just realized after updating sdk and stuff all my textures are purple now... oof
are cubeds shaders busted?
they work fine
yeah I did, same
do you have cubed's shaders in your project?
does the console give you compile errors?
I know it doesn't like the 'tests' folder of post processing stack v2
but cubed's shaders still work and compile fine in 2017
Shader error in 'CubedParadox/Flat Lit Toon Lite': failed to open source file: '..\CGIncludes\FlatLitToonShadows.cginc' at line 143 (on d3d11)
guess ill try installing once again lol
yeah grab the latest version too https://github.com/cubedparadox/Cubeds-Unity-Shaders/releases
is the path to the cginclude inside cubed's shaders' folders?
ah, might be an older version of the shader floating around in your project
any way to force overwrite the package contents?
pointing to a file that was moved in a newer version
you can delete the folder
and reimport
ah yeah that did the trick on the import window
but yoiu might have an older copy somewhere else in the project
lets try that
do you know the answer to my previous question about the atlas mapping?
nope that fixed it, thanks 😉
ok cool
does gravity on dynamic bones only affect the first joint?
Use force instead of gravity, gravity seems to not function properly
It should bend all of the bones in the chain
Or well, the first bone, and then the rest follow suit
when i tilt the head forward the pony tails stills kinda stick out straight
i can see the connecting bone of the ponytail flex down towards Y though
Try these values as a test http://prntscr.com/m168rr
ill give them a whack
Use it in play mode so it doesn't actually save it
That way you don't need to revert everything
Oh hey, that's what I use for my ponytail too!
that pulls it down through the body
my settings were pretty similar on another avatar heh
to those ^
I'm using them for wires so they jump quite a bit
May need quite a bit of tweakin, but will probably give you a good reference
hmmm
tilt the head forward and the hair still kinda sticks out backwards
maybe something i just have to live with
Huh
Here's the wires
Not too sure why yours is sticking backwards, are you using a single root for it all perhaps?
single root?
im using the head as root bone
is there a way to make dynamic bone spheres link the space between ?
like how colliders can be extended?
so it cant clip through between spheres on arms and stuff?
Dynamic bones is more of a simulation than an accurate collision based script, so collisions will for the most part always look choppy, even when multiple colliders are used (which they shouldn't, gotta say that before somebody drop kicks in)
i have a capsule collider on both arms
and the long hair gets "stuck" on the arms and hangs over and it looks great
but it can slip through the collider if the hair colliders have a gap between the spheres
Mhm
no way for it to simulate a collision space between the dynamic bones?
Not that i know of, aside adding even more, in which there'll still be gaps between
ah nah, dont wanna do that to potato people lol
Eyup
so hey my arms are really long on my avatar and they are not bending all the way they wont go straight no matter what height setting i choose. how do i fix this?
FullBody?
no
Just arms?
vinny i dont think you can if they arent proportionate?
just arms
i could be wrong
They need to be proportionally accurate to your own to bend properly
hmm that sucks .
If your arms are shorter, it'll bend even with your arms extended, if they're longer, it'll stretch them out even if you bend them
i get no errors when uploading either
That sorta deal
most miku models are pretty accurate
but when you get into custom models things get hairy
this isnt mmd
Well you could try to shorten/lenghten the arms a bit to test
Unless it's extremely detailed, you won't really notice much
how would i do that
Oh is that fran?
yeah i put her in a virgin killer sweater
Neat
I think you can possibly just do that in pose mode in blender
Then you can set the pose to be the resting pose
Which makes it so that that's the original shape of the model
how would i shorten the arms ive had to edit in blender and reimport and re assign all dynamic bones over 6 times now
do i have to go back to blender again
sorry i was busy typeing i didnt see the messages
i figured that
It won't look like the greatest thing in the world that's for sure
But it's a simple fix to an otherwise complex issue
what is this new bounding box error in the build control panel
Pretty sure there's viera models out there... probably not fran specifically, but something you could probably setup
or someone completely remodel it and make a high poly version but no one that has done that before has released there model ive tired contacting them but they dont respond
why does it think my avatars bounding box is gigantic
Probably a broken shapekey
Yeah, otherwise your model won't react to light in maps all that well
Especially if the shapekey is absolutely huge
hmmm wonder how to identify it
Since it'll try to apply lighting onto your model based on the location
Which will be everywhere
haha
You just gotta go through it, one by one in either unity or blender
i mean how do i see the bounding size
Also, if your model has any spawnable items, in the sense of shapekeys, you'll need to check if they're in the middle of nowhere
nah no spawnables on this one
No hair changes, clothe swaps, glasses, hats y'know?
Hell, even facial emotes could be anywhere but the head for whatever reason
the god damn shirt cuffs arent weight painted right >_<
dam ff xiv i have to wait till this summer to make my viera character
Hmm...
You'll need to go ahead and check each shapekey one by one then
If no shapekeys are causing it, which you can test by deleting them all, re-exporting, checking bounds, then it might be a stray mesh or bone
Mhm, for the most part that's how you find ehm
Try deleting ehm all and re-export, if the bounds stay the same then it's something else
If the bounds return to average values, get all the shapekeys back and literally start deleting and exporting in batches of 10 shapekeys
When the bounds return to normal, you'll have found the batch of 10 possible bad shapekeys
yeah same error with zero shape keys
Then it's definitively something other than a shapekey, here's a way to sorta figure it out aswell
do you use a pen 
Go into edit mode, and select everything you can actually see, start with meshes, and press H to hide everything you've selected
nah no attached items
After every mesh in view is hidden, press A to select everything and X to delete
Everything that is hidden will be safe, and you can afterwards press alt+h to unhide it all
seperate meshes first?
Shouldn't be necessary
You can
You go into edit mode, and then you either press C for the brush selection tool or B for a box selection tool
And no, atlas' don't mess with bounds
Then after selecting what you can see, press H to hide
If anything is left in the center, keep repeating it
When nothing else is visible, press A to select everything and press X to delete it
Afterwards press alt+H to unhide everything again, and exit edit mode
Export the model again, if it still has issues, do the same for the bones now
This will guarantee that anything that you can't see, gets removed
ok i just have some bones left now
What do you mean?
Oh nvm, you're seeing the bones whilst in edit mode on the mesh right?
Aight, go ahead and follow my steps and it'll go through without issue
This should get it all fixed up
delete is asking me for edges/faces/etc.
hmm looks like something may have been floating out in space
or i missed something tiny lol
So the bounds are all fixed up?
that totally did it yeah
