Copying the tl;dr from the bottom to save time:
TL;DR: Base avatar and clothing creators who sell their stuff have no financial incentive to make optimized or mobile versions, as you really need to do it from the ground up with the intent to make an optimized or mobile version. If end-users are willing to shell out regardless of how optimized an asset is, then not optimizing literally raises their per-hour wage, and while some blessed creators are willing to put in the work anyway, it’s hard to fault anyone who isn’t. As such, other creators modifying, kitbashing, and otherwise customizing avatars will necessarily have a significantly harder time optimizing post-hoc, especially for mobile. You really need to be willing to dedicate a lot of time or channel some sort of hyper focus to stick with it long enough to not only learn the techniques but apply them to ever-more-unoptimized assets coming from above.
-–
As someone who has spent literal hundreds of hours optimizing my avatars (including all quest/ios versions being under 20k polys and ‘poor’ rated), I wholeheartedly agree that optimizing for Quest/iOS can be incredibly painstaking and requires a lot of time and effort. There is sadly no ‘magic bullet’ to quickly get an avatar built for PC under the ‘very poor’ mobile limits, and whether we like to admit it or not, there is a WHOLE LOT of entitlement and elitism among a large percentage of PCVR users who are fortunate enough to be able to afford and operate a PCVR-capable machine. I’m willing to bet even if there was such a one-click optimization button, a fair amount of users wouldn’t bother clicking it. Imposters are sort of an attempt toward something like that, but we all know they’re far from perfect, and even if they were, some people would still insist on using fallbacks.
To make a truly optimized Quest/iOS avatar, a creator pretty much has to build it from the ground up with optimization in mind. The number of creators offering ‘quest versions’ of their avatars is vanishingly small (and in some cases it’s just a prefab with the mobile shaders and everything else - polycount, meshes, materials - still wildly over the limits), and the only way that number is going to grow is if the people who make the avatars feel the money they earn from selling optimized versions is worth the extra time and effort required. That’s not a dig at creators; this stuff is *hard* and they deserve to get paid for the time and energy they put into making things. And as Wub said earlier in the thread, there are many goals and targets for optimization, some of which require very different skillsets and/or workflows.
Maybe some creators would be more willing to make ‘mobile optimized’ avatars if the limits were adjusted, but that’s one heck of a big ‘maybe’. The VRC staff that post here have agreed countless times that reviewing the avatar limits is something that needs to happen, and, yeah, it’s frustrating when things some of us think should be a much lower priority or off the roadmap entirely (I will die on the hill that is boops) take precedent, but adjusting or replacing something as fundamental as the performance ranking system is also *really hard* - especially for the mobile limits. You have to remember there are a lot more standalone android-based headsets than just the Quest line, and Meta undoubtedly ‘ruined’ VR in a lot of ways by (among other things) insisting on making the device a ‘gaming console’ in and of itself, rather than a PC peripheral that lets you experience games in a new way. Also by artificially lowering the price and purposefully taking a loss on every unit, skewing peoples’ expectations of what a VR headset should be, do, and cost. It let a lot of people who otherwise would never have considered or been able to afford a headset get one, but whoa mama did we all pay a hefty price for it in the end. Remember, if a product is free or cheaper than it logically should be, YOU are the product.
The point is, there are a lot of different pieces of hardware with a lot of different specs that, ‘officially’ supported or not, people are using to play VRChat. Making it ‘tolerable to play with strict performance limits’ for 90% of that userbase is a much better strategy than ‘entirely unplayable for 70% but with looser performance limits for the 30% who can run them’.
That being said, that entire missive is entirely about optimizing for Quest/iOS.
Before I go off on a rant about PC avatars: Tutorials. Yes, there are tutorials on optimization, but so many of them are just not any good. You have the ones with perfect audio and editing that tell you absolutely nothing you wanted to know, and the ones by people who know what they’re doing but their mic is a tin can and a piece of string and/or they have half the screen covered by their pngtuber avatar. I’m sure there are a lot of really great and helpful ones out there that have appeared in the few years since I started doing this stuff, but I honestly haven’t looked since they weren’t there 3 years ago when I needed them and I’ve worked out my own workflow at this point. And the older ones with higher viewcounts are more likely to appear first in a search, which means anyone brand new to this is going to get poor quality or outdated information. I can empathize with people who go looking for how to do this stuff and get frustrated when they can’t find anything useful, and if your free time is limited, it’s easier to just not bother.
Anyway, old man rant ahead.
PC optimization though? Man, all y’all pushing over the very poor polycount for PC are just bad at this and need to learn some restraint. 70k is a ludicrously generous poly limit for avatars in a game where the average effective per-eye resolution is barely 2k x 2k and a likely majority of users need corrective lenses anyway. People walkin’ around with half a million polys and 8k textures is like trying to push uncompressed 7.1 dolby surround through a pair of $10 wired earbuds from the discount bin at best buy.
Physics simulations in Unity games are not even remotely accurate enough for 30 physbones with a thousand collisions each to be worth the time it took to weight paint everything, let alone the performance hit of that many bones.
We all have dozens upon dozens of upload slots for avatars and it takes mere seconds to switch between them; there’s no reason to have hundreds of thousands of polys and multiple menu toggles’ worth of clothes on a single avatar. I’ve seen an avatar with pants that couldn’t even be toggled yet for some reason it had a pair of 30k poly briefs underneath??? Like dude what are you doing? (Note, this is the same argument given to and ignored by people worried wearing their avatars with nsfw bits in public will get them banned. Bruh, just upload a version without the wang to wear in public, it’s not hard, pun intended).
There’s also no reason for every accessory or piece of clothing to be an entire separate skinned mesh renderer with its own material. UVTileDiscard is trivial to set up in Blender and with Poiyomi and VRCFury on the Unity side. There are tons of tools to help with texture atlasing, but rearranging/combining UVs in Blender really isn’t that hard either. And even then, does the clothing HAVE to be toggleable in the first place? See above: just upload a version without the clothes!
Clothes and accessories aside, again, 70k polys is a very generous poly limit in something like VRChat, and if an avatar base by itself exceeds that limit, I’m sorry but you really need to do better please do better I know you can I believe in you. Rexes are hands down the most repulsive furry base available and I’ve never actually seen one that wasn’t ‘very poor’ rated, so I have to assume the bare base by itself is already over the 70k limit and there is absolutely no reason something that looks like THAT needs more than 70k polys. It’d probably look a lot less like a drowned rat if half the mesh was just straight up deleted. I don’t even want to know what the polycounts on some of the ‘hyper-realistic’ looking ones in Furhub are.
The irony in all this is the people who get the most upset when you tell them you have ‘very poor’ avatars turned off tend to be the ones who have ALL avatars turned off by default (and are the most dismissive of mobile users). It’s especially funny when their reasoning for having all avatars off by default is because too many people use ‘very poor’ rated avatars and it bogs down their framerate.
Ok rant over.
TL;DR: Base avatar and clothing creators who sell their stuff have no financial incentive to make optimized or mobile versions, as you really need to do it from the ground up with the intent to make an optimized or mobile version. If end-users are willing to shell out regardless of how optimized an asset is, then not optimizing literally raises their per-hour wage, and while some blessed creators are willing to put in the work anyway, it’s hard to fault anyone who isn’t. As such, other creators modifying, kitbashing, and otherwise customizing avatars will necessarily have a significantly harder time optimizing post-hoc, especially for mobile. You really need to be willing to dedicate a lot of time or channel some sort of hyper focus to stick with it long enough to not only learn the techniques but apply them to ever-more-unoptimized assets coming from above.
“Just figure it out. There’s no playbook. You’re making the playbook. Draw the f**king owl.”