#open-beta-discussion
19 messages · Page 27 of 1
one day ill find a bug that makes the users pc explode and i'll report it and i'll be seen as a hero
I meant more the main menu. Where even adding the vrc+ button long ago made it feel crowded. The quick menu would be too small for it
Take my money! 
God, imagine having money
i always get so confused when people call them ads
i dont see the mcrib being promoted to me on the quick menu ever
i had to disable toggle voice to speak, push to talk no work either
i kinda get it, some people see the banner format and immediately assume ad
i try to replicate 
the banners (or formally, the "info push system") are meant to be a way for us to nearly immediately communicate with people in vrc
and one of them does include please give vrchat money
actually i need to enable the beta banner, whoop
The art style is pretty poppin' too
seeing a new banner does get me pogging
idk who does the graphic design for them but you love to see it
are there any plans to update the sync rate for OSC?
at the moment OSC + face/eye tracking is too slow, network wise, to be usable, without smoothing anims.
something something interpolation better than update rate for that use
forgot, scroll up 
It's structured in exactly the same way as ads, and takes up a lot of space, so it's a reasonable assumption
you can hack this by adding the params to your menu and keeping the pie menus open while you're doing the tracking and that helps
it'd be nice to have that update rate all the time.
without having the menus open
maybe if new banners and information would show up once and then go away, freeing up the space and making it less annoying while still having a way to communicate new game changes and other important information?
if they just show up once and go away nobody would remember
Personally I never had an issue with the banners
It would be much better. Like, having a banner cover a couple rows for a brief second before getting out of the way, and then having a dedicated button in the quick menu to get back to banners
Or even just have a toggle in the settings menu to hide those banners
we talked about it a bit before up there, but here (AND AGAIN, i am not the feature czar, this is me paraphrasing, this is not me making guarantees):
- the radial menu thing is an unintentional side effect of having OSC, its kind of a bug but not really, not a super big fan of people using it the way they are using it, as it complicates us making... well, basically any changes there. we have to account for this odd "spacebar heating" use case.

- avatar params is not meant to be the final form of face/eye/etc tracking so we aren't trying to push for that. the "final form" is likely some kind of native on-descriptor definition driven by OSC and any native solutions we end up implementing. networking for that would likely be its own network stack or piggyback on a higher refresh rate extant stack like IK/pose.
- faster update rate isn't necessarily the solution here-- a bit faster might help but really interpolation is what'd help. right now there is zero interp
but then how long do announcements last? I don't need to hear the same information once per session for months? Also about remembering, does everything need to be remembered? It is great to convey information, but a lot of people wont care either way, or if they do they will access it in other ways like through this discord.
tbqh banners haven't changed, they are off topic for this channel 
like as an example open Beta announcements need to stay around the entire open beta
so what's everybody going to do with those extra parameters to be more on topic
Ok, thanks for the update and information! Really do appreciate it. Is there a roadmap or time frame for when we might see this face/eyetracking addition? As in--is it something we'd be likely to see this year, or is it further out than that?
No roadmap, no estimate, again, this is me tossing out my wishful "magic wand" thinking here as a result of translating internal chatter and convo into external digestable format that's bound to get screenshotted and passed around until it looks like a ifunny post. 😄
However, needless to say I'd love to have it sooner rather than later!
wha
doxxed
lol
i dont want eye tracking, that means true eye contact will be a thing
i dont want that
scary
Oh god yeah
unironically this
peepoLeave
peepoArrive
people instantly noticed that I'm ADHD as fuck and my eyes go everywhere instead of focusing on the person I'm talking to
Introverts like me will have it real rough then, having to actually use your eyeballs to look at people. Scary
xD
I use face tracking and that adds a lot to expressivity. been debating if I want eye tracking or not
if I do get it, I will probably have a toggle for it bound to a gesture lol
Just set up a 3.0 toggle to switch between real eye tracking and fake eye tracking. Easy
^

people will have to deal with me not looking at them just like they do irl
you can do that by having proxy bones for the eyes
now with 256 parameters this is even easier!
not my problemmm
1 pair of bones for real eye tracking, 1 pair for fake eye tracking, 1 more pair that the eyes are actually weighed to. put rotation constraints on the latter with the other 2 bones as a source, and use an animation to switch around the weights of those 2 sources
i can go even crazyer with my fx layer as if it wasent cancer enough
again, thanks for the information!
i mean you can also just not get a headset with eye tracking, it isn't like that's going to be a challenge
sure thing!
at least people won't be as disappointed that i'm not as "confident" as i seem in vrc lmao
that why u leave the slider on shy
I mean you do have a point, but then I can't do the Rock eyebrow face can I
with the vine boom bound to it
eye tracking is one thing, but eyebrow tracking is simply one bridge too far
the technology isn't ready
heard from a source that Cambria is supposed to have that
i mean i heard from a source cambria is supposed to print money and have a functioning fusion tokamak built in
lol
i heard from a source cambria has a built in coffee maker
bought
sold

it'll be interesting to see how accessible Cambria's face+eye tracking is to VRChat.
coughy time
new project cambria leak

hold on kido i wasnt on beta

@wind talon any clues if Cambria's face/eye tracking would be accessible to VRChat? 😏
i have no idea
fair enough!
Is the increased parameter limits here to stay when it gets pushed to live? A friend of mine come to belief it's only temporary.
it's only in the non-live open beta
Yeah it will be a permanent feature once this update eventually goes to the live branch.
Nothing is 100% sure for beta though
i am very curious how your friends came to that conclusion
the answer is "yes, of course its here to stay"
Indeed.
Well one friend, apologies, they got mixed out when it was mentioned parameters being used for facial tracking was a temporary solution and I guess they got mixed up thinking you were talking about parameters being increased. Thank you! Sorry for the confusion!
oh! okay, i see. nah, the param space is here to stay. the face tracking thing is just because a native solution (driven by OSC) is what we'll probably eventually have, so using avatar params isn't the final form of face tracking in vrchat.
it's a great "preview mode" tho.
Oh boy, more parameter memory to play with!
It is! OSC is great and I love what's people have been doing with it
Yeah I’ve seen some mind blowing things with OSC

wut
I might put face tracking on my avatar for the lulz now that we've got double parameter bits lol.
it is certainly a parameter costly thing to have, but now we got more room to handle that with feature-heavy avis
Also desktop updating parameters as fast as VR now is big for me, cause I'm a desktop player.
so i found a bug now the fun part is finding out what actually broke
might just be since i've not been desktop for a while, but i don't remember my hand sticking out while grabbing props working on every avatar before
OMG
Oh my God thank you so much vrchat
Doubling the parameter space friggin awesome
wasn't able to replicate the mute button getting stuck, but now i cannot do gestures on desktop

I think it's going to take months and then it doesn't
I'm incredibly impressed
And excited
Doubling the parameters spaces one of the best they did
im so confused i have a animation. works perfectly fine for myself and on live but on the beta no one can see it
i see it working pefect for myself but no one else can
reproduce it, post the repro steps with the avatar id as a canny
ya i narrowed it down to the specific animation will make a version i can just release and let others tweek at it and see whats wrong
think i finally found why the mute bind just dies 
used controller and mouse/keyboard at same time, mute and gesture buttons dont work on controller or keyboard now
is it safe to use the new SDK for uploading avatars for use with Live?
heard no
new project cambria leak
le canny
I thiiiink as long as you don't go over 128 bits it should still work but I could be wrong
that's what I figured but just wanted to make sure
That's a big beta holy moly
beta^2
Just out of curiosity, what was done for the significant performance boost for Phys Bones?
gave them gamer juice
they have a reduced update rate at a distance
ok thats actually smart
upvoted
Ahh makes sense
that's the most user-facing thing about it though, apparently there's a bunch of other more technical things like just a general code optimization pass
would love to see benchmarks
good to know ^^
@ Table xD
like an 80 people lobby with people spread out at different distances
https://feedback.vrchat.com/open-beta/p/1197bug-animation-works-fine-localy-and-on-live-unable-to-be-seen-in-open-beta @wind talon made a post and uploaded a version of my avitar with it i removed everything possible its jsut a basic toggle that turns on a cube in your hand works on live just fine but on beta its local only im so confused

cough wish that was an option we could play with cough 
||https://feedback.vrchat.com/avatar-dynamics-reports-and-feedback/p/1181-update-rate-setting-for-physbones||
tfw tracked internally


Is it know what the distance is for that starting to happen?
not sure
only one way to find out ig
it's kinda hard to tell because this only changes the simulated rate but it still gets interpolated every frame between the simulated positions
ahh that's good
I hope this wont cause any problems for stuff like grabbing & pulling, so that they will be at the same position for everyone.
basic functionality was tested but yeah, always more stuff to be found out in the wild
It shouldn't hopefully, since if I remember correctly the Phys Bones will always move to where they are for the Person who uses the Avatar once you let go from them / pose them.
I wonder how long until it comes out to live 
soontm
probably not long
Wonder if it's 20m like it was for dyn bones. :v
That's a lot of parameters
would make sense
ya lol
cannot scroll main menu with left joystick anymore 
https://feedback.vrchat.com/open-beta/p/1197-bug-cannot-use-left-joystick-to-scroll-main-menu
Time to make my cherub even worse ONCE AGAIN
I didn't even knew there was one xD
Oh, I didn't realize beta chat was back. lol
Nice, added steam deck controls in new open beta, but do videoplayers work on it aswell or not? I don't see anything about that in release notes.
Also somehow the steam deck microphone in vrchat sounded terrible for some reason - others people could hear themselves and the sound was cutting out. Gonna test it out now, hopefully that is fixed now or in the future.
Not much VRChat it self can do about Video Players on Steam Deck as far as I know.
related, this old issue came back
~~https://feedback.vrchat.com/open-beta/p/navigating-with-joystick-in-the-main-menu-is-reversed~~
made a new canny for it anyways 
https://feedback.vrchat.com/open-beta/p/1197-joystick-scrolling-is-inverted-again
If its a proton issue then maybe vrchat team or anyone who knows about how to fix it to contribute on proton github, hopefully in the future everything works, its been kind of sad not being able to watch anything or listen to music with friends on the deck or any linux distro in that mattter 😦
Yeah it's a Proton issue if I rember correctly
@storm fossil The issue is a LOT MORE COMPLEX than that, You can read more about it here https://github.com/ValveSoftware/Proton/issues/1199
It boils down to codecs and WMF
Aye, thx, will read on it.
I see you're in the NEOS Discord. In their case, videos work fine on Linux because they're using libvlc for audio/video stuff instead of any system-native multimedia helpers exposed by Unity.
Which isn't a good solution for VRChat, because Neos devs despise libvlc, and would much rather use basically anything else.
oh? 256 parameter memory now? i am definitely getting eye tracking with my facial tracker now
now i can pair every avatar with eye tracking and mouth trackin pretty much its funny i always wanted it to be more because i would just go over the limit most of the time. 256 is def perfect i think
Yup 256 is great and in one year everyone will complain about getting even more xD
(hopfully not I really can't think of anyting you would need more than that)
i mean with the dynamics system you do not even need parameter slots
inb4 people using ints and floats for literally everything :sadge:
i made an avatar with sheath unsheathe. swing sounds and hit sounds takes no parameters
Btw should issues with IK 2.0 be posted on the Open Beta cannz or should we still use the IK 2.0 canny for that?
oo can scroll sideways with mouse wheel and the Mouse5 back button also workssss
i wants to scroll horizontally while holding shift like literally every other program ever, but think that would need to wait til MM beta
Also noticed in the new open beta that the crouch key is mapped on steam deck, but not the prone key (z on desktop), also prop rotation is done by holding L1 - using left trackpad for it would feel much more natural, and you could map prone to L1, also L4 L5 R4 R5 would make alot more sense for gestures since these buttons are right under the fingers.
try holding the crouch key
when i was messing with the gamepad stuff, holding the left joystick down would put me into prone
Bro 256 avatar memory 
Really depends, stuff which necessarily has to use floats (basically where you control the motion time) are pretty expensive if you use a couple of them.
yeah but you can now have lik 32 of those, I think that should be enough xD
I just hope people don't get the idea "I have 256 bits now, so I can add another 100 layers to my fx controller"
i have no idea what to do with all those bits
Agreed, that should be enough now I hope, I did hit 128 pretty consistently (I usually would use like 140)
Yup same I can now finally do some OSC stuff xD
Btw what happens if I load an Avatar with the new SDK and more than 128 Paramters in live? 
I guess everything above 128 wouldn't get synced
All I need now is less unexpected behavior of physbones with scaled bones in the chain (like making hair shorter with scaling a single bone), then my creator life is fulfilled xD
Yup, that works, though not quite confortable or intuitive. Also obviously a button layout screen would be nice with all these new controls ^^
All I need is access to parent constraints for quest (even if its with harsh limits)
I have seen some canny post that were set as tracked internally but can't seem to find em anymore
And it looks weird, now imagine being able to use something like int4 instead of int8 xD
when did the Quest Creators canny board get removed 
i have more than that before the update
scary
As said by the devs before, Constrains acutally have a really high perofrmance impact which is why they aren't on Quest. They also have to keep Quest relatively good perofrmance whise, because if they don't do that, they could be kicked out of the Quest Store.
it might still get added with limits
Which is kinda sad, I still hope that we will get them at some point in the future :/
like physbones
If they do it that would be the only way
limit it to like 10
10 would probably be to much xD
Ah i see, wasn't sure about the performance cost, for me it's mostly an annoyance because have to put duplicate objects on an avatar if i want it to swap positions, which both increases poly count and material count
not really thay arent that bad but people will put 100s
Quest is havealy CPU limited, I'm honesly suprised that they even got Phys Bones.
From what I've been told they actually are
exactly its why personally 10 is reasonable but who knows
i remember on the "constraints on quest" canny, tupper said something about wanting to tell something preventing quest getting constraints in the future but the canny was deleted 
do you have a quote for that? because ive only ever heard the opposite, since of course constraints arent even tracked by the sdk at all, indicating they have next to no impact.
I Myself would say 5
its relative thay are more expencive then previously thought not insanely expensive
I mean, I assume that they have been tested in a quest build to actually see their performance within said environment?
it was previously believed to basically be free i still wouldn't call them expensive
though havent seen actual numbers
kinda curious
they're free in small numbers but they scale disproportionately. there's a huge jump to multiple milliseconds around 50
ya sounds about what i expected
by all means i wouldnt mind a hard limit on it if it truly does cause an impact like that
i was happy about the physbones limit
i want 1 single parent constraint 
theres alot you can do with just one
Hmmm
it for hair in first person 
Man, I'm so exited to mess around with the new max amount of pram bits.
Now I can make more OSC magic, with out having to use weird compression tricks. 
I can see why, but at the same time I would get annoyed by my hair blocking the view approx 2.48 seconds after loading the avatar xD
i would be using for twin tails in the back cause they can be grabbed with physbones now
Actually I wonder now, for parent constraints do they always have some impact on performance or only when they are set as active?
Of course there is some performance cost but I'm asking more like, does setting a parent constraint to active cause the performance to go down at that specific moment?
Not disputing the performance impact, but was curious if anyone had a reference to the parent constraints discussion? I'd like to know how much evil I am visiting on others by not slapping objects throughout my armature.
Ah okay, goot to know ^^
A hard cap of like 5 sounds reasonable then I would say @lyric ocean.
(If you think about bigger instances)
to be fair thay could make more than 2 or 3 make you very poor and at that point your blocked by default
with a higher hard cap
but its up to them
Worst case scenario would be like 32 players where everyone has 5, so around 160 parent constrains. Then again that is worst scenario and I doubt such an scenario would pop up
read #open-beta-announcements to see why! c:
Just want to ask and confirm because brain slow rn, but is the new SDK only working for open beta? Like how I would want the increased parameter memory and wanted to use it in the live version for example, would it work?
The open beta is not network compatible right now, so live is stuck at 128b for now.
Alright thank you so much!
If you upload and Avatar with more than 128 Bits, it will work fine in the new Beta, but they everything beyond the 128 mark will be ignored in live.
ik beta =/= open beta
So 11946 is not network compat
ik beta is still network compatible, open beta isn't
I saw that
I haven't had time until now to read it fully
I am so anxious for 256 bits to reach live
if you go above 128 on live, all parameters are ignored
not only what went above 128
I imagine this open beta won't be in beta for too long
¯_(ツ)_/¯
oof okay, good to know
If you had paid attention to the OSC github page the 256 parameter thing was mentioned, you'd know this was gonna happen, just never imagined it would so quickly, but I love when I am pleasantly surprised.
VRChat has been rapidly evolving over the past 2 years.
Yup same, I knee they planted it, but not that it would happen so quick xD
So exciting, I'll finally be able to implement eye and face tracking into my avatars
I like making multipurpose avatars
Yeah I don't have a HMD with face or eye tracking so I sadly can't
Valve, where is my Deckard????
I expect this will become more common in the next few years and more headsets will integrate this tech.
Yup, hopefully
Based on steam vr data mining its likely the deckard is 1.7 or more years away
I actually started modeling my master complication watch yesterday
Knowing that I do not have to worry about it having too much functions taking too much memory is great
It's still in proof of concept stages
I'm gonna cry
Valve is very focused on the steam deck technology at this time and that technology is a very large stepping stone to creating a standalone headset that's x86-x64 compatible (has compatibility with the steamvr library)
They don't want to create something that will be outdated in 6 months
they want a headset people can keep for atleast 2 years before being obsolete.
For example is there SteamVR tracked headset that is available at this time, that has all the features of the Index with no drawbacks, that costs the same or less than the Index?
Keep in mind while the Index full package is $1K the headset alone is $500 and all other steamvr tracked packages and headsets cost more than the Index at this time
While some include features the Index lacks, they have a drawback, usually in terms of audio or FOV
Or adding audio costs $150-$300 USD and it costs more and is inferior in terms of quality of audio (both mic and headphones/speakers)
Yup the Index is the best allrounder for me, especially for VRChat
Did they update the osc update rate for facial tracking? So it doesn't look choppy for others?
no, they brought Desktop parameter sync in line with VR
The current update rate is 5hz as far as i remember?
around there
#open-beta-discussion message
#open-beta-discussion message
in short no, but avatar parameters aren't meant to be the final form of face/eye/etc tracking
and yes, the update rate is 5hz
I thought they improved in new open beta to IK speed
nothing in the change log said that
Right. Just not yet. This is on the roadmap.
They have expressed interest that OSC will eventually sync at IK speed
Not really, they plan on implementing Face Tracking in to VRChat without needing to use Paramters etc.
Even better
The Sync Speed of Paramters will probably not change.
I think it's OSC parameters that are slower
There's jitter with the facetracking OSC app
Not sure about that
This is in the canny
I've also used it
Locally its finee
Remote users see some jitter
Ah okay, yeah I remember. It's slower but it gets faster when you use the Parameters in the Expression Menu and have it open.
Once this beta is live I am going to use my vive pro eye and facetracker more
Such a weird "bug"
I have haptic integration done already
Just need to add eye and face tracking to my main avi
Why is my space bar only working sporadically
There's no dirt under it
As soon as it's commercially available and I can afford it I'll plug my brain into VR and never look back
What about (Finally enough Params)
Good idea. Gonna change it when im home 👍
Why are like 90% of the bug reports from @inner kayak ? xD
the menu scrolling made me angy 
cause i reported same thing during qm beta
Yeah, that's complelty understandable xD
I didn't even have time to test this update yet, it released at like 3 am for me and I'm not home untill 12 pm xD
Well, unless you didn't miss something like your work it's fine I suppose xD
If you say so 😅
Another open beta without List<T> exposed to udon : (
The more people that report bugs and make feature requests the better. This is the only way the team will know what the problems are. They don't have the time to sit here and discord to hear all of the issues that people are having. They're busy working. The feedback canny is the tool for communication between the team and us. This is something very important to communicate to other users in VR chat that are frustrated with how things are. If they want things to change encourage them to put feedback in the canny rather than complain about it to other users.
i think they were just joking, obv we know why the canny is there
Anytime I meet a user who's upset with certain things I always make this recommendation
IK, I wirte a lot of bug reports etc too xD
I just found it funny how currently 90% of the ones in this current beta are from @inner kayak xD
Of course most people in this discord are aware of it. Most VR chat users are not
Even users that have been the platform for many years are unaware of its existence
they said input got redone when afk button no work, so i lookie at anything else input related 
Yeah ik, some one I talked to also didn't know that VRChat is planning on a better face tracking integration. (He was mad / sad about how bad it currently is)
I think it would be a good idea to communicate this to other users in the menu or something.
Perhaps I'll make a feedback canny you report about this
It is, there is a banner linking to the canny.
I felt the same way. Although I read the GitHub and the team already expressed the intent to change this.
I was willing to wait patiently for it to occur.
Half of that plan has already come to pass.
i don't expect it will be long before face tracking will be built into VR chat
Likely before project cambria becomes released
the github stuff too hard to look through, not everything is in one spot
Most of the stuff from the dev stream from last April is also now done. I think only UI 2.0, UI 2.5 and the creator economy are missing
creator companion too
Creator Companion?
well idk if that was in the stream thing
This information was pointed out to me elsewhere by another user of VR chat. This user is responsible for the OSC application that drives the B haptics suit
Every time this user makes a prediction about what's coming to VR chat it comes about two or three weeks later
It occurred to me this user was paying close attention to the canny and github's
This is the best way to predict things that are coming to VR chat
The team does indeed communicate to users of VR chat you just need to know where to look
companion app thing that helps manage packages for your vrc projects ..?
idk the canny for it is public so peeked there
Ahh that think, I complelty forgot about it xD
Hmmm... interesting this sounds intriguing
I suppose time will reveal the full nature of this "creator companion"
pretty sure even though it’s closed beta, the download link for it is just sitting public 
Yeah tbh I hope a dev stream later this year so we can get a bit more of an insite of what is going on.
least i think i remember stumbling on the doc link from something while ago
edit- remembered where and yah it just sitting public by vrc team 
I've noticed there are quite a few users who have expressed the difficulty in the learning curve in terms of creating VR chat content
too tired, looked spooky
Perhaps this creator companion may help bridge the gap
Btw what happend to the VRChat Dev Labs, wasn't that also a thing that was planned?
VR chat dev labs?
Why is that?
I think it was supposed to be a place where the devs show more of what was currently going on
That will go a long way in terms of communication between users and the team
How small is the dev team in the VR chat staff team?
I already know it's not very significant
do you have an exact figure on the number of individuals involved on the dev team?
I don't think there is any offical information on how many devs there are / who does what
I already expected this answer.
(at least not that I know of)
There doesn't seem to be much public information on this
my persumption is that it is not a huge number
no it's not, but it's also not tiny
PC game development teams can include hundreds in some cases
At least in terms of AAA games
VR chat seems to be somewhere between indie and AA
The VR game contractors only includes 3 developers for example
One of them personally contact me to give out game keys on my stream
Originally I thought this was some kind of scam because it never happens to me
I was quite surprised when I figured out it was legitimate
I'm way too small a streamer to receive keys from developers
Although interestingly enough this is happened twice now
Has anybody tested boolean parameter driving ?
You're going to have to be more specific in what you mean
If it still syncs at a different speed for the the user client and the observer clients
Well if you drive a boolean in a controller, if there is not a time delay introduced it will work for you but not for other people seeing it
The parameter type bool has been a part of the VRChat for quite some time
The time it takes for the animation to have an effect depends on your transitions
but yes there is an amount of time that it is synced across the network
VR chat documentation should have an answer for this question
It does not
The thing is, I found a solution that works, but in the changelog there is a mention of "- Various fixes and improvements to internal memory management processes"
So I wondered if it could influence this also
Ah man, really wished they had increased the sync rate for the osc driven parameters, 5fps is really not that much
They've expressed interest in doing this in the github. At least for facetracking
The Legend Of Zelda: Breath of the Wild had 44 programmers I think. Just programmers. There's also designers, modelers, sound and music people etc etc etc
Coders aren't always the only developers of games. Development is a much more loose terms these days.
You can create a game with modern game engines without ever touching code.
Yes that is why I said there's also X and x etx
Asset creation is a part of developing games.
But the programmers is the only number I remember
However the Dev team of vrchat Imma say is probably around 10-15 people in my opinion
Possible.
there are 66 listed on linkedin. I'll say that a few are unrelated trolls but the vast majority are valid and there are also a few who are on the team and don't have a linkedin, so that figure is a good estimate
That is more than I expected xD
You can find the approximate number of employees on the internet, it's public information. Just look up the company info itself
There is also information about investors
And also Phasedragon (who works for VRChat) just said it lmao
that's a major lowball estimate there~
Yeah it really was, I was expecting around 40 - 50
The thing is, people outside US and Canada are contracted, so they probably do not count as employees here
Also, on jobs.lever.com/vrchat there is 19 (!) open positions
If I upload an avatar using the new SDK not using the new parameter space, will it work on live?
Nope, the parameters will all sadly just be gone
Dang. Thanks.
btw where did you got that info from?
2 minute google search. This is all public information
I meant that specific website but okay 😅
I think it would be cool if we could use LB and RB to switch between the different Tabs in the Quick and Main Menu
oo
And actually use the B button on an Xbox controller as a back button
ohh sad, I can't right click buttons anymore to just play the click animation without actually doing what the button should do xD
was always using the to play the open / close sound of the QM wings xD
can still click and hold to play the sound before the animation 
true
Uh oh, am I the only one whos new controls on steam deck stopped working (except gyro) after reinstalling vrchat? When switching back to stable it works like before.
put in a canny
Hold one of the shoulder buttons (R1 for me)
And then use the joysticks for each hand
hm nothing happens
Btw did anyone else have issues with UI sounds just being gone?
I have no idea how that happend
It's just gone for me right now
Rejoning has fixed it
Okay it's the world which is weird
the issue happens on live
Made a detailed canny about this!
Currently the B button is only used to close either the Quick Menu or Main Menu, my suggestion is to make it work more like a Back Butten as it does in other Games.
https://feedback.vrchat.com/open-beta/p/1197-feedback-make-the-b-button-work-more-like-a-back-button
updooted
I think this part in the announcement reveals the true intent behind this open beta
This leads me to believe it will go live much sooner than the Avatar Dynamics beta
Depending on which elements go live first
If it's the whole thing it may take a month or so (likely less)
Hooooold up, does this mean I can upload something with more than 128 params?
Like... right now?
only 128 will be on live , but sure
Ah, I should hold back then for the moment-
this open beta is what will go live (after some bug fixes of course)
there is no splitting or anything
Someone said it just doesn't work for live (becomes 0)
I think
The only question is time.
I imagine as always it will happen sooner than I expect
They did this with the physbone update.
While the increase to 256 bits of synced parameter space is a no-questions-asked literal improvement over 128, that's still not really enough if you really need to get into the meat & potatoes of doing cool stuff with your avatar without having to gut everything for 1 specific feature.
For the main example that prompted this post, this is what the parameter list would look like if you wanted full eye and face tracking with the current version of VRCFT:
- EyesY
- LeftEyeX
- RightEyeX
- LeftEyeLidExpandedSqueeze
- RightEyeLidExpandedSqueeze
- EyesDilation
- JawOpenForward
- JawX
- SmileSadRight
- SmileLeftOverlay
- MouthUpperUpRightApe
- MouthUpperUpLeftPout
- MouthLowerDownRightUpperInside
- MouthLowerDownLeftLowerInside
- MouthUpper
- MouthLower
- SadLeft
- CheekSuck
- PuffRightUpperOverturn
- PuffLeftLowerOverturn
- TongueRoll
- TongueY
- TongueX
- TongueSteps
- TongueUpLeftMorph
- TongueUpRightMorph
- TongueDownLeftMorph
- TongueDownRightMorph
The numbers represent the limit of floats you can have with 256 bits. You get left with 32 bits for other things. If you wanted both a tail puppet and your ears to react to sounds around you with 'VRC-OSC-Audio-Reaction' to go along with your full face and eye tracking (not exactly a unrealistic combination of things to have on a single avatar), guess what? You're plum straight up out of parameters again, so not even any bools for outfit changes or other basic on/off things.
With a increase to 512 bits, now we're at the point that you'd have to be really trying to fill that list up with even the craziest of custom setups for cool things.
Just no...
If you really fill up 256 bits aka 32 floats then you defiantly doing something wrong...
Also currently doing Face Tracking with parameters is just a temporary solution, VRChat is planning on doing it in a more official way, so without you needing so many parameters for it.
And if you really need more then 256 bits for all your toggles etc. you should really consider splitting your Avatar up in to multiple ones, for the better performance of everyone.
Honestly, no reason not to split up your avatar bases into multiple ids if you have different outfits and all that at this point especially considering calibration saving is a thing now. 😉
No more 300 skinned meshes on your avatars please. 
well, the avatar menu UI is one reason... not a lot of options for categorization there
avatar wing is a thing 😅
it is, but the problem is there's limited ways to organize it. You have basically five categories - "all avatars" and each favorites group, and can sort by either upload date or creation date. You can't, e.g., categorize by avatar base, or by the occasion (unless you only need five categories)
Main Menu overhaul plz save us 🙏 🙏 🙏
oh god i've never seen it that bad ;-; have you?
worst i've ever seen is eboy avis with like 60 meshes
There's a customize your own spartan avatar floating around that comes to mind.
Every time someone loads into it, always comes with a nasty hitch
ah yes
the one with every single armor piece from Reach and a few guns available to change between
i know the one
512 parameters
just add 512 bit param support as part of vrc+
but I want to animate all 20 materials on my avatars hue shifts independently tho ....
That would be 20 out of your 32 floats
Truth be told more parameters are always good to make cool stuff like osc interactions (not including face tracking)
but i need more space to make single toggles with ints!
Make your single toggles with floats and just trigger everyone that makes anything in sdk3
You could actually do that now, you can have up to 32 floats lmao
aren't the floats in the animator 8 bit mini floats? so they would use the same amount of memory then the 8 bit ints ^^
Yup, they are
Yeah, but just about everyone has been using Ints and bools for their toggles. Using 1 float per toggle would set a few people off I’m sure of it.
well, yeah
there are reasons for it tho
of you use something like proximity, or the height as input
I mean for stuff like proximity contacts you don't need to use a synced parameter
I'm not sure when to set my contacts to Local Only. I am using both default and custom collision tags. It seems like animations end up local-only depending on what I do. The parameters are defined on my ExpressionsMenuParameter list.
If you defined them in your Parameter List they will be synced over the network.
why only hueshifts? full rgb main color controlls with three floats and a blendtree :p
Contact parameters are run on both the local and remote machine regardless so defining them in your parameters list isn't necessary. If you choose to do so though, you should check "local only" otherwise, you don't.
i did that for a while cause i had to set a float from 0-1 in a blendtree 
now i have bool and use parameter driver to set the float
cause i’ve learned 
Hue shift requires one float. RGB requires 3. More floats means more memory means less memory for other parameters
TL:DR hueshift is better than rgb
i will write sarcasm after the sentence next time.
- PhysBones have received a **significant performance improvement!**
Btw how big is this performance improvement?
(Not because of limiting the Updated Rate for people who are far away but because of the code optimisation)
will it work if i upload an avatar to live with the new sdk?
the params dont go over 128
generally no you shouldn't be uploading for the live version using beta sdks
I think it simply won't show up on live
as long as you don't go above 128 bits it will work on live
Have you tested this
yes
How long does it usually take for Open beta to get pushed to live?
words cannot express how excited I am for the double parameter memory
Is it normal if you use the beta sdk and upload an avatar with the doubled parameters to live the avatar menu is disabled like i cant click any button in the avatar 3.0 menu
I think if you went over 128 bits it gets deleted on live
yea but why disables the whole menu tho,should still be able to use to 128 memory space
Because it's a beta I guess
yee. Hope the beta gets out soon so i can make my avatar more complex
I been stuck at 127 for months and would be nice to add some senders/receivers for animations^^
Contact Senders / Receivers don't need a synced parameter though
*years
from tutorials i seen it needs float parameter but that has to much memory
(unless you really need them to sync perfectly)
Contacts can be ints, bools or floats depending on what you want. They also don't need to be in your Synced Parameters...
idk why there's a limit after all, could make one in safety settings but dunno
That would be a really stupid safety setting...
And as I said, you don't need to us a Synced Parameter for Contacts and you also don't have to use floats...
I mostly follow the tutorials from Sippbox and that's what they used i think. Im not familiar with the reciever/sender yet only adjusted physbones stuff
Sippbox hasn't even made a tutorial for Contacts yet...
back then i used int for cloth toggles but i learned bool is easier
and less space
Sippbox made me understand Physbones easier but you are right. It was someone else, my bad
but ye Sippbox is still the one who made me understand Ava3 much easier. Quick and easy, not a half hour tutorial x3 but yee hope they make a tutorial on the advanced stuff too
You can also read the official docs
That is what the official doc says, but doesn't answer my question. If I use a custom contact (for example a pen makes contact with a book, activate a parameter from ExpressionsMenu to activate an animation) do I make that local?
if you're toggling between less than 8 different states use booleans. any more than that you should use an int
the one downside to using booleans in that scenario is having to toggle them back off, unless you set up a driver
im toggling between 8 states with three bools, since 2^3 = 8 possible states :p (but thats basically a situation of each bool can be toggling something independendly)
i mean yeah if you want to encode your menu options in binary go for it
If you only plan on enabling one option at a time, using an int value for multiple settings is great
but the systems that use ints in place of bools are really heavy, bools are much better if you need more than one thing enabled at once in a set of changes
any way that we may get something like "int2 / int4 / int6 " stuff, so one may use the easiness of int without always having to use 8 bits?
ints may have more of a network impact because they take up a whole 8 bits which is overkill if you only have like 5 outfits, but bools have a bigger framerate impact because you'd need either a whole new animator layer for every independent bool or a ton of states and transitions
Maybe eventually, there should be enough room to generally not worry about that now though
from what i saw during testing its better to have less layers but more possible states
I'm firmly in camp "absolutely do not put multiple outfits on one avatar" 
yeah that's the real solution
I really hope that calibration saving helps people to stop doing that
I never used the avatar wing until calibration saving was a thing, now I love it
Avatar wing and expression wings are amazing, I love them.
it's really unfortunate that "I'm very poor because I have multiple outfits" is so normalized, makes it impossible to use safety settings
Download size limit does a pretty good job for me.
guess a lot is coming from "oh, 70.001 tris, so whatever" mentality as well
ah yes, using the multiple outfit excuse to hide the fact that each outfit is 150k tris 😅
btw, i am doing this, even though it are anystates, it seems to be better for frametimes doing that instead of three layers without anystates for toggling stuff (3 independend toggles in one layer combined)
I've managed to do it and stay under 70k but oh boy did it take a lot of blender time
wait did you put your outfits on separate blend files?
you don't have to do that, you can just remove the extra meshes in unity
it's a lot of extra unity time which is totally fair but it should have no impact on your blender workflow
no I mean getting the polycount down took a lot of time
oh well you should be doing that anyway 
:v
but that require breaking the prefab right
almost all my avatars are medium or better
not if you remove the component
ohh i see
All of my avatars are their own blend file with meshes merged 
unless I specifically need to enable or disable them
same, one mesh for head for blendshape separation, other mesh for everything else :x
so i could have one model with multiple outfits in blender, then separate into diff avatars in unity by removing unneeded components for each dupe?
yes
i guess i'd still need separate models if outfit has different bones, cause removing those requires breaking prefab
Doesn't that assume that the outfits are all not merged though
but that sounds pretty useful
you could do it either way - keep the skin separate and now you have two meshes, or include a copy of the skin in each outfit's mesh, which also gives you the opportunity to delete the skin underneath the clothing to prevent clipping
weird
switching to the robot avatar while head tracking is set to animation will persist
That avatar swap method is unusual
ctrl+\
I know, I mean I wouldn't recommend it for anything besides the "escape" situation of being in a bad avatar
eh it persist to any sdk2 avatar it look like
not just the ctrl+\ robot
(is live issue too, so placed in Avatars 3.0 canny) https://feedback.vrchat.com/avatar-30/p/switching-from-sdk3-avatar-with-head-tracking-disabled-persists-to-sdk2-avatars

that's why I have about 30 versions of my avatar >.>
initially (when 3.0 got released and enable/disable things got so much easier) I wanted to make one avatar to rule them all... but I quickly realized that A) I don't need all the outfits everytime, and B) it's just terrible performance wise.
but I used the opportunity to create "unique" features for each avatar outfit ;)
(the jedi one has a working lightsaber, the stormtrooper one has working blasters, the bikini one has a beach chair, parasol and watergun, etc)
and I'm always advocating to very poor avatar users to either optimize themselfs, use more optimized avatars or at least use very poor avatars only in small instances >.>
but that is kinda an uphill battle with those e-girl/boy avatars <.<
and I get the same response quite a lot: "Get a better PC", even tho I probably have better hardware then most of them, and I don't suffer that much from the bad performance then others...
well, that is a "so so thing" in my opinion. "very poor" covers a bazillion avatars, a lot of then unoptimized and actually very bad for the performance yes, but i know a couple of "very poor" avis which outperform medium/poor ones. I think avatars have developed too quickly on multiple areas and the current performance ranking can only give a rough estimate now, but does actually not reflect the impact of said developements
I'm talking about actual very poor avatars (not those who may have 76k polygons but 60+ materials, 300k polygons, 4k collision checks and so on)
dont forget the 100 4k textures to kill the vram 🙂
ah, yes >.<
I mean, I use 4k textures from time to time, but it's usually justified, as the other option would be to use several 1k,2k textures
but then it's maybe 1 or 2 textures
same, i only use 4k if i have a lot of detail on a large area (at least 50% of body height has to be that texture, for example a summer dress), i am aiming for maximum 120mb on thrys vram calculation tool, but mostly end at around 70-90
thrys vram calculation tool? may share more info about it?
thry has a vram calculator, where you can put in your avatar game object and it takes every referenced texture/mesh into account to calculate an estimate.
found it, thanks :3
even if it shouldnt be accurate, its very handy to look at potential wrongly setup textures which end up way bigger than needed
mhm!
I usually just guesstimate it myself
hm... one of my most heavy avatars is also in the 120MB ball park...
in my example there the both top textures are both 4k ones, but one with and one without alpha channel, seems like even that makes a difference (the 10mb one doesnt require alpha, so i disabled it)
oh tf
thanks to it I found a wrongly setup mask... that used 10MB >.>
like he did xD
just reduce the resolution of it, a mask usually doesn't need that high of a resolution
and thats sadly a thing, avatar creation advanced a lot and so did the quality and resolution of textures people use, imagine a "good" rated avatar using 8 materials, but every single texture (really, every single map etc.) set to 4 or 8k. that would be half a vram nuke, but its not getting recognised by the performance rating
from tuppers comment:
"8192MB of VRAM - 2048MB (for world, OS, app, etc) / 80 people = 76.8MB each."
that isn't that bad...
yeah, thats doable , but it feels like a lot of avis are more in the 200-300 range, not including the 500+ nukes one encounters every now and then
yeah...
I think that why I put a tracking control on the first state in the base layer to prevent that.
same, first state does basically set a "defined state" to ensure nothing carries over from other models
what the purpose of the open beta
to test new features for breaking changes, before they get pushed to live
For the VRChat community to be able to preview upcoming features and find bugs in the future updates
also gives avatar/world creators some time to test and update their stuff, if needed
triple A studios do that too, but they call it "release" :P
I mean this one
it already got answered
A lot of stuff changed other than IK 2.0 and it might cause problems. We want you to find them.
I'm surprised you're unsure of the purpose of this OB in particular Franada. :v
Even booth users, I have my game set to 25mb avatars, most booth avatars can't reach that with a couple of outfits, but then, so many have 200-500k poly 60mb+ avatars that you don't see them using but ONE OUTFIT FOR ALL. -_-
Yes, 60mb is a already too much, even 30mb is not good at all, that's already unoptimized and has unnecessary stuff in them. yet complain game needs fix. xP
even E-boy/girl outfits can get down to poor with mad good shader work if you slightly work on optimizing them. But these ppl rarely do or care and still complain on game.
I think the heaviest any of my avatars ever got was around 40mb max
VRAM goes brrrrr
cache also go brr
a 100mb avatar is one seventh the game's file size lol
also, download time - there isn't any good reason for an avatar to be that large
yeah there really isn't xD
my normal one with all my fancy stuff on it is 54mb and my more optimized one is 34mb
(that's still a lot but wayyyy more reasonable then 100mb)
heck, my main avatar is 1.14mb lol
Hello guys, I'm new here and also I'm new in unity, I have these problems, And I don't know how t fix the problem? Anyone can help me?
wrong unity version
This page lists the current Unity version you should use when creating content for VRChat, as well as links to the correct version for both Unity Hub and direct downloads. Current Unity Version The current Unity version in use by VRChat is Unity 2019.4.31f1. Ensure you are using this exact version! ...
Ehhhh, So I have to download another version?
"The current Unity version in use by VRChat is Unity 2019.4.31f1."
yes, it's easy if you use unity hub
@orchid peak btw. that would be more a question for #avatar-help then the #open-beta-discussion channel ;)
Haha, sorry I really don't know that, I'm really new at here.
A third if you don’t count the predownloaded base avatars and worlds.
heckb
i update the sdk and now no toggels are working anymore
dunno how people manage that but ive tons of material toggles and stuff on my avatars and im ranging between 9 and 12 MB
the highest resolution i use on my avatars is 2k
in beta or in live?
in normal
yeah if you used more then 128 bits with the new SDK in live it will not work
Oh oki
Thank you
dunno what i would use that much bits for. Im at 60 or something. Though a lot of params are unused after updates now.
I'm at 127 currently
my vram usage is fine
but yeah everyone should use that vram calculator
meshes are making the highest usage of vram in my case
oh nvm
actually not
im still very poor though
I have my cache set to 100GB
How do you get this vram calculator?
jeez
sounds like someone needs to go through all of their textures and crank down the resolution as far as you can
but I like my 4k textures
you do not need 4k diffuse + 4k normal + 4k metallic
4k diffuse sure, but 2k normal and ~1k metallic
set the resolution for each texture individually according to their needs
keep turning it down until it looses too much detail
I wish there was a way to just mass compress textures
there is, you can select multiple textures at once
but that's not how you should go about it
because chances are, one texture can have a lower res than others and you won't notice the difference
so
4k for base colors
2k for normals
1k for metallic?
as a general rule of thumb yes
the smaller the asset, the lower all of those become
a ring does not need a 4k texture
1k max
then normal 512 and metallic 512 or 256
you get the idea
the lower you can make the resolution for everything, the better
turning down the resolution by 1 notch is a 4x improvement
lmao rip
much better, but could still be improved
even better
I dropped all 4k textures to 2k
and I dropped all 2k to 1k'
most of the specular or metallic I dropped to 512
which shader are you using?
since a lot of shaders can use packed textures for metallics
so that you can have 1 texture for metallic/reflection/smoothness rather than separate
mhm
thanks haha
welll uhm yeah
That Thry tool seems a little uhh, not good
wdym not good
A lot of the info there isn't really true
it is
how so?
VRAM is not download size
"If there were 40 of your avatar you would take up 18,02 GiB of Video Memory"
This is false. If there were 40 of your avatar, you would take up the exact same amount. It can reuse the same textures in vram, it doesn't need to copy paste it 40 times
is that really important though?
I think they mean 40 other avatars of the same performance rank etc.
that is not really important though
the point is to put the amount of VRAM used into perspective
^
not actually because 40 people will use the same avi
That it should probably be written differently. Saying "40 of your avatar" is not right
it's just to display how much 40 people with such an Avatar would take up
it would still require duplicating the texture which does use extra vram probably not as much but there’s still an impact
also, I had no idea VRChat would just reuse textures in VRAM
vrchat smort
so basically this is saying
"with a 3090, you're gucci"
lol
I have a 3080
LOL
the fuck
did you just say br
even only the body of my Avatar takes up 34 MiB
yes
👀
nothing to hide lmao
yeah I can't show that
not unless I rename some assets

lol ok
discords whole "bad language detection thingy" never actualy works
go on
how can I reduce file size of a mesh without breaking it
show the full list
VRChat would get mad at me
Remove unused blendshapes
Also lots of blendshapes do destroy frametime as they’re not linear when recomputed
bros got the whole planet in their avatar
how do I see what blendshapes aren't being used?
You check and see
In your animations
interesting how a texture that is only 483kb in size can take up 21,33 MiB 
I have way too many....
yeah Rezillo has a tone of blend shapes
If you have a lot of moving blendshapes (like FT) it’s technically more performant to split the head mesh from the body
You can do facial but not eye with the index still
oh right
I can't justify getting the facial tracker without the eye tracking
it'll feel out of place
Pro Eye isn’t that much of a downgrade though
it kinda is
the index has plenty of room on the inside for some kind of DIY version
just need someone to make one
If you don’t use the high refresh rates meh not really
terrible mic, worse speaker / audio solution, lower refresh rate, lower FOV, more scree door effect, for the most part a lower comfort, needing to run extra programs to use it
what website is this
Versus is a global data-driven comparison platform, covering over 90 categories. Compare smartphones, cameras, headphones, graphics cards, and much more. With detailed tech specs, data visualizations, and price comparisons, Versus is the best product finder for a wide range of consumer electronics, from smartphones to PC hardware.
thanks
Anyway, let's get back to the open beta topic, this is getting a bit #general-media .
right
Pro doesn’t require extra software that’s the Pro 2 but yeah
Ah okay, sorry my mistake then.
already a thing
also, face tracking on index works just fine
SMH the vive pro eye doesn’t have a remote control such a poor design choice
Ugh can we PLEASE fix Index controllers stopping Root.T from coming over in actions/emotes? It's been years. T_T
users with index controllers have Root.T dropped remotely when they do emotes.
looks fine for themselves
i've never noticed anything like this with my friends 0-0
You'd need to see then playing an emote that makes them walk around
they'll be walking in place
this looks hilarious
https://gfycat.com/fearfuldizzygangesdolphin
up and down motion is broadcast, but any side to side is killed
really visible in MMD dance worlds that have you walking around the stage
one person will just be standing in place
Kinda funny because the person who made this bug report now works for VRChat.
I know.
this bug has been the biggest thorn in my side for 3 years. and since you look perfect for yourself, very little people complain.
it's had me tempted to just get vive wands to avoid the bug.
again you won't even notice it on yourself, you just look wrong to others
it pays perfectly locally
I have an AFK that sits me on the ground, everyone else sees me burried
Yeah ik, I mean I never noticed this on anyone else, or at least not as being a bug. ^^
Don't mind me, just being the squeeky wheel hoping some day it'll get addressed so I can go back to making cool emotes T_T
Well, let's hope it will be fixed ^^
me too
weirdest thing is, leave at least one tracked body part set to animation (has to be one that would visually ruin the animation), and the bug doesn't show up
weird
I discovered triggering the default AFK animation and going back fixes this, because it forcefully is switched from local animation to NetIK
Root transform still needs to be fixed though :/
This bug has been super annoying and really aggravating me because it only affects index controller users
which aren't a huge percentage of VRChat players so maybe they just don't have the drive to fix it compared to all of the other stuff they're working on in the background
Kinda on the topic of root motion being wacky, one of my animations are a bit busted in VRC with IKBeta. I posted about it on the canny, but after more testing I believe it is my animation in particular that is broken, as other animations I've tried that also animate root rotation work fine.
https://feedback.vrchat.com/vrchat-ik-20/p/root-motion-rotation-in-animations-applied-2x-locally
Live
IKBeta
Would anyone be able to point me in the right direction to fixing it and having it animate properly? I've got images and the actual animation clip linked on the canny.
that's because crunch compression only helps with download size and you can probably make that mask texture lower-resolution
^
every texture is decompressed before being put into vram
so crunch compression does not help
which is kind of why like detail Maps can be quite helpful so you're not having to make your textures 4K just to have repeating detail
Single material, 3.3k △, using RGBA masked 256px normals in @poiyomi's shader. 🧵👇
260
Back to shader: Can use seamless tiled normal maps per RGB area - tiny (128/256px) normal maps, but tiled around 50-100 times in the shader, giving you a higher effective TD.
You can combine these with an base nrm map for creases and baked normal details etc: (Blend with Base)
There are NO "modifications" of the VRChat client that are considered safe or allowed, no exceptions.
Any application that interacts or modifies the VRChat application is a violation of the Terms of Service, and using it will result in your account being banned.
(original message was deleted by the user who wrote it)
are we no longer allowed to have more than the recommended max number of phys bone colliders and such with the new sdk in the #open-beta-announcements ? cause i was able to upload my avi before i updated to that sdk, now i cant
On Quest that has been the case since it released. The Quest has a hard limit for Phys Bones.
Anyone that's been keep up with this beta, think it will be released this week? Or are there major bugs that need to be fixed?
Seems like a pretty simple update vs physbones
Just getting excited! 😁
Hoping it comes out soon!
I'm not trying to upload to quest though
Don't forget, this "simple" Update also has the new IK in it
hmm weird
But that beta has been out for a while now, though
Have the bugs not been ironed out?
There are still a few issues here and there
I also haven't written bug reports about 2 of these I think, which I will do later this day.
The phys beta was around for like 2 months and it still had bugs/issues on release. Nothing’s perfect, there’s always things to work on.
But testing things is never a bad move especially when it’s such a massive update like this one is
This beta mainly has problems with syncing rn. Avatar parameters and udon arrays have issues syncing. Some worlds are completely broken, especially with a lot of people in them.
The IK side of it is pretty much fine, it's mainly the rest of it that's pretty buggy. I doubt it'll release this week.
I'm gonna try switching build target to Android then back
I don't have any problems with there being a hard limit (Yes ik, this looks terrible but it's only there for performance testing.)
So this actually seemed to have fix my issue

weird, maybe the SDK still thought it was in Android mode
Either way, if anyone else has that issue now we know the fix at least
Sadly, I don't have an animation that moves my avatar away from root, so I don't think I can help. :<

I should mention that your poly count is nothing compared to mine, so don't worry about that regard, I have multiple hair and clothing toggles
just thought it a funny fact at how high my polys are
are you telling me your Avatar has over 800000 Polygons? oooof how tf did you manage to do that
It, uh, has over a million
14 different shirts
7 different hair
Like 8 or so pants
Tail and ears
A few misc items here and there
Multiple guns from Halo and Titanfall
Mind you almost all of them are disabled at any given time
just use material swaps br o
use material swaps for the shirts !!
Material swaps?
What is that, can it completely change a shirt from, say a t-shirt to a dress without the extra polygons that having the 2 objects would have?
no but it can switch between other shirts
So like using an int param to choose which shirt is active
yeah
Which is what I have already
itll help with the polygon count if you make the shirts into simple material swaps
my 10 pieces of clothing + the entire model is <90k
What is the poly count on these shirts and pants such that they add up to much of anything
surely you don't need over a million tris
Most of them aren't active at any given time, as I have quite a few int params to limit different spots to 1 item
Like only 3 guns can be enabled at once, and only 1 of those are bigger guns
I've made sure to keep the avatar optimized at runtime, even if it is a big avatar
So far no one has had any issues with it in any world I've been in
what about those 59 skinned meshes and 139 materials though
surely with the amount of toggles you said you have, you have a lot of meshes per item you're toggling, which could all be joined together
Nope, only 1 mesh per item, so most of those are also disabled at a time
As for the materials, it's mostly between shiny black, black, white, metal, and if the item has a specific texture. If it has multiple textures, I atlased them together
for things like props that you won't really be editing the textures of, using CATS's Bake is very useful
reduces it all down to 1 mesh/1 mat
if you have UVPackmaster2, it uses that when repacking the islands
I plan on reducing the materials and textures and such when I get most of the rest of my avatar setup, reason I haven't done it yet is due to not having enough of the avatar done to know how best to optimize it for what I need
So the tail doesn't phase into the body, along with the scarf
Multiple colliders are also for the bullets for the machine gun I reuse the same bullets for all the guns to reduce colliders
Gonna need a collider for the weapons, but I haven't set those up cause imma see if I can use 1 then parent it around like I do the bullets and lights and just parent it to each gun
Equip sounds and fire sounds for individual guns, with similar sounds being parented around to reduce audio sources
keep in mind that Angle limits are nearly free in terms of performance, unlike colliders. you can use angle limits to prevent physbones from going into the body
Issue about angle limits though is if you set the max limit based on pulling the tail straight up to the back, that means if you pull it straight out then you cant bend it upwards in the middle cause the angle limit at that bone doesn't know or care if the previous bone is at it's max or not
Unless I'm mistaken on how angle limits work
the limit determines how far each bone in the chain can move relative to their rest position
you can add a curve for the angle limit, to make it so that different bones in the chain have different max angles
you can use the Pitch Roll and Yaw fields to "offset" the angle aswell
and use curves for those aswell
Meaning that if you setup, say a chain against your chest, so the angle starts at the rest and can move out for each one to not let it go into the body, then if you pull on half of it, the other half is gonna stick straight out cause it's gonna already be at it's max angle that way
Which is the basics of what I wanted to avoid for my tail so I can still grab it and move it in front of me, but not have it phase through my body in doing so
Even this should not exceed over 1 mil. please, please, just use the good ol blender to optimize these meshes. Unity still background loads and renders these models, and being this unoptimized is not healthy.
the 3080 can handle about 9.5mil tris, you are 1/9th of the gpu
That is TERRIBLE
not to add that this is in a falling opengl test, meaning it is just OGL rendering the model with limited libraries
And well, Unity has draw calls outside of just placing triangles on screen
im pretty sure that gpus can handle more tris than that. the main issue is draw calls and such
nah, stress tested was 9.8 mil
that is A LOT
And well, this is without batch sizes, batch sizing can get you up to 14.2 mil
but the asset IS NOT batched
you say that but all these meshes and textures are killing vram
please import this https://github.com/Thryrallo/VRCAvatarTools#vram-checker
and check
THANK YOU
i really want to see the result, i cant imagine it being less than 200 MB
probably around 500 MB
btw can you already update sdk and use it in live?
yes, as long as you don't use more than 128 parameters
mine is around 500 MiB and you know how mine looks, I really don't want to know how bad this guys is
This is horendus
you are a model! (but with all caps)
please
if it isn't your materials and textures (which should only be a few and around 1024 to 1024 or 4096 to 4096 for atlas), then tune down on the vertexes, please...
My avatar takes up 21mb and it has a ton of toggles and comprises of a 2k atlas
The wonders of asset optimization
For real




