Feature Requests

Please check out the following guidelines before suggesting a feature! Off-topic posts will be deleted.
Support for VRM? 3d Humanoid Avatar Format for VR?
As I'm sure you're now aware, Vtubers and VRChat in general is currently on a moonshot rocket ride to mainstream in Japan. Due to this, DWANGO and others have started OSS initiatives using Khronos GLTF standards to create an optimized "platform independent" 3D avatar file format. They already have published a unity asset that allows you to import and export them from FBX and other meshes incredibly easily, and they are designed to be embodied by humans in VR, making them EXTREMELY well suited to VRChat. In fact, I think they were created with VRChat in mind, if you look at their mission statement - they really do seem that way. It may already be possible to bring VRM into vrchat since the format fits a lot of your package standards. https://dwango.github.io/vrm/vrm_about/ Either way, direct acknowledgement and support of the format would go a long way in south east asia I believe, as they have somewhat decent copy protection considerations for creators, as well as optimized setups for normals, meshes, materials and armature. The MMD PMX and PMD file format is WILDLY popular in vrchat for good reasons as I discovered way back in 2015, but there are things that turn off the japanese creator community about this (sorry for that :( ), mainly models getting ripped - and the fact that they are not optimized for vr IK or human embodiment, and not generally light weight. This is designed as a complete solution to all that, and they have a lot of backing and this initiative comes from talented devs and huge companies, like for example niconico. Give it a look!


Folders for Uploaded Avatars
We want some way to group my uploaded avatars list to make it easier to select avatar. 日本語要約は下にあります Abstract Current VRChat doesn't have any way to manage avatars except for avatar favorites. Japanese have uploaded many avatars and will more when we follow official recommendation, one avatar should have one outfit. According to my survey, over 33% of Japanese have more than 50 avatar variations. We don't want to use avatar favorites and they are inadequate so we want some alternative for uploaded avatars. As a one alternative, I request folders for Uploaded Avatars. Why not avatar favorites? There are two big reasons. First, we want to use avatar favorites for public avatars. Even though we have uploaded many avatars, We want to use public avatars. Currently, avatar favorite is the only way to keep public avatars so I want another. The other is that number of avatar favorites is not enough for this usage. We only have 6 avatar favorites. I want to group into more groups. Why Japanese have many avatars? In Japanese community, almost everyone have multiple outfits for multiple bodies, so we actually have many avatars. I surveyed about this on Twitter for Japanese VRChatters. The result says about 33.1% have more than 50 avatar variations. Someone says they actually have uploaded over 500 avatars (they upload one avatar for one outfits), and another says they have 1000 avatar variations in total (they upload one avatar for multiple outfits). survey: https://twitter.com/anatawa12_vrc/status/1672077245165256732 500+: https://twitter.com/narazaka/status/1672154813684146176 1000+: https://twitter.com/MekaNyankoVR/status/1672466157927006208 However, current VRChat doesn't have way to manage uploaded avatars. So, we want some way to manage uploaded avatars. The reason why many people have many outfits is that there are huge variety of outfits for avatar bodies and many people bought many of them. In Japanese community, many outfits author make some or huge variation for many avatars, so It's easy for people to add, combine, and customize outfits to their avatars. For example, in Capettiya and mumu , popular outfits and avatar stores, many outfits are made for about 40 avatars. VRChat may think most one doesn't have such a number of avatars by server metrics. However, there's a reason why the number have huge difference. Many Japanese combines many outfits into one avatar violating official recommendation that one avatar should have one outfit. Instead of uploading avatar by outfits, many Japanese uploads one avatar for one body, and they change outfits with expression menu. That's one of the reasons why Japanese Avatars will always have many skinned meshes and PhysBones, and huge VRAM usage in their avatar. There are multiple reasons why they do that but one of the biggest reason is that with many avatars, we cannot choose avatar easily from list if we do (or already). (Previously, Not keeping the calibration settings is another biggest reason, but it's now fixed with IK 2.0) Some Similar Canny There is another Canny to improve problem with many avatars. I strongly want folders for uploaded avatars but those canny can improve problems with many avatars. Switching Avatars from Expression Menu and/or Contacts Menu Design Suggestion I made two image for design suggestion for Avatar Folders, one for Main Menu and the other for wing on MM/QM. I hope this may help you to understand my image for this feature. 日本語要約 アップロード済みアバターの管理方法として、フォルダ機能を追加してほしいです。 多くのアバターを持っている人は、アバターを切り替えたい時、目的のアバターを探しづらいと思いますが、VRChat公式が推奨している「1アバター1衣装」を実践すると、更にアバターの数が増えてしまい、大変です。 そのため、アバターを管理しやすいようにフォルダ機能がほしいです。


Add Support for Streaming Mipmaps for Avatars and Worlds
This feature request is to add support for streaming mipmaps for avatars and worlds. Steaming mipmaps ( unity docs ) allows textures to be downscaled and dynamically loaded into VRAM. It does increase the size of each mipmapped texture by 33%, so package sizes will be larger. However, it allows textures to be loaded dynamically based on distance. This would mean that a texture that traditionally may be 2.8 MB DXT1 at 2048x2048, may be under 44 KB if it is reduced to 256x256 in VRAM. So in worlds, this means significantly reduced VRAM usage when texutures are a certain distance away. Same with avatars if they are a certain distance away. If tuned properly, this could mean that avatars that may traditionally take up 200 MB of VRAM, can instead take up magnitudes less. Often, 2K textures look just fine on most avatars from even 0.25 meters away. So being able to unload them for a smaller one would be significantly useful, especially for large meetup events. Fax did mention that VRChat currently does not have mipmap streaming. The reasons are the following: > Most notably - some user avatars may lack mipmaps, have them disabled, or have a maximum priority. This would adversely affect mipmaps with streaming enabled: Once you run out of VRAM, they’d appear blurry and stuck at a lower mipmap. Users might run into this issue maliciously, or accidentally. Considering that the VRChat application currently allows avatars to be created with mipmaps, but they aren't shown, perhaps there is a way to allow the client to choose if they want them enabled? I do know many games have texture streaming as a togglable option, so perhaps VRChat can do this here? I would even be happy if this was an experimental feature that had risks associated with it, if it's for the sake of testing. Fax also mentioned: > Unfortunately, that wouldn’t fully solve the problem, as avatars uploaded without streaming mipmaps enabled effectively have an ‘infinite’ priority. I see this as a non-issue, unless there are some other contingencies. Unity docs mention that it can load non-mipmapped textures alongside mipmapped textures. Considering VRChat doesn't have mipmap streaming already, this means that the game will be identical with or without mipmap streaming if all avatars don't have mipmap streaming. But in the event that half the lobby does, then there will be at least that many avatars that will save on resources, even if the others don't. Also, perhaps (idk if this can work), but this may also mean that the avatar hider can unload textures properly? At least in theory, if the avatar is not loaded, it may load the smallest mipmap? This would save a LOT of VRAM for users that want to use the avatar hider. This feature would help a lot for people that do not want to optimize their avatars. I do think VRAM over-utilization is one of the worst offenders in terms of VRChat performance. Solving this issue would help many, especially those with less VRAM.


Load More