NiranV Dean

In VRChat you have a 100-200 ping too when connecting from EU to US or the other way around. Even with a 40 ping everything is still highly delayed. Getting stabbed across the room even in a EU server is pretty common. You can see the person walking up to the place you were 5 seconds ago, carefully taking aim before placing that knife up your butt, all of which you can very comfortably watch from across the room. It is annoying but not neauseating at all. But VRChat is client authorative. Whenever you move or do any action it happens instantly for you but takes time for everyone else to be seen. Objects you pick up or drive are switch ownership to you, meaning you will take care of synchronising its movement for everyone else, this makes sure that for you the experience is always instant, its not so great for others to see in some cases. A clever trick is because you are responsible for the synchronisation of any objects you own, objects you pick up and carry around will have the same delay as you do, meaning they will be synchronised with you and line up properly. This is the default behavior and can be overridden by the world creator, some worlds use host-only sync which means you are at the mercy of a 200+ ping, making especially vehicle and world interaction lag behind a lot. Most people don't seem to have a problem with that tho, they go to flight or drive maps and deal with the incredible input lag of those vehicles and they don't vomit. What seems to trigger most people is simply... moving, like just moving, sliding around (rather than teleporting). The lag would be annoying but hardly a big problem for motion sickness, to fix it though SL would have to switch to a fully client predictive model (SL already predicts and interpolates movement but its not doing it ahead of time), this could simply be added as an option you turn on, making the client instantly emulate your movement while sending it to the server (rather than waiting for the server to report back), this will innevitably create cases where when theres physics lag the client may think you move faster than you currently can due to lag, making you bounce back to the synchronised position aka rubberbanding, this once again could be combatted with more predictions based on ping and region FPS (both of which the Viewer knows of).

Time to do this. I'm so infinitely sick of seeing these human content creators being full of *****e and false information and fueled by pure capitalism and laziness. I was going to write a full Guide on how to optimize models just to show everyone how stupidly easy it is and why everyone should be in rage against any creator daring to come up with cheap excuses like "too much work" "takes too long" "isn't worth it". So here it is, step by step how to optimise any and all models. To start off, i prepared a little scene, i loaded my optimised Wicker and the original unmodified Wicker. To showcase how much you can optimise while retaining (or outright improving or adding more details) i've put both side by side, gave them the same blendshapes and the same materials. The left one is the original, the right one my custom one. First thing you will notice is that the left original is missing some hair in the front, again this is the original, it doesn't come with said hair, that's an addon. The polygon counts are as follows: The original unmodified version has 67.338 polygons. My modified version has 45.961 polygons. How is this possible you may ask? How does the modified version retain the same quality and even add more stuff to it? Optimisation. Taking a slightly closer look we can see that some areas on the optimised version are less dense. Particularly the head and claws are very apparent. So the first step to doing optimisation is simply identifying where we can optimise, you don't need to be skilled or a professional to identify some key places, just looking at the wireframe the head, tail, claws and feet seem like a good place to start. After identifying where we can optimise we can start right away. While in Object Mode: select the model and change it to Edit Mode: Now in edit mode we can start editing the model, the easiest way to optimise any model if you have any normal non-*****ed model is simply smart dissolving edge loops. That sounds incredibly technical but is incredibly easy infact. In the top left, next to the Object/Edit mode dropdown select the edge selection mode: this allows us to select edges (instead of single vertices), other methods involve using the face selection mode. With the edge selection mode enabled simply select any edge you deem unnecessary like so: I chose the claws here because as you can see they are incredibly dense for no reason, perfect place to start optimising. Now that we selected a single line you'll probably say "but that's going to take an eternity and you'd be right... but there is a very nice feature called edge loop selection. Simply hold down the ALT key and click an edge like in the above example, Blender will automatically follow along that edge until it loops around or ends. As you can see it selected the entire line all around looping back to the beginning. Now the actual magic. Dissolve. Blender has an amazing feature called "Dissolve", it essentially does all the work for you. With the edge loop still selected, right-click anywhere and select "Dissolve Edges" from the dropdown. It should be almost at the very bottom. Now what this does is it "removes" the selected edges and makes all necessary changes in the UV and the mesh to keep everything properly connected. The result is this: And that's essentially it! The same can be done with faces (and vertices although i HIGHLY recommend not doing that). Simply go around, select edge loops, dissolve them, make sure you hold SHIFT in addition to ALT to add to the selection rather than replacing it. A nice, quick and easy starting optimisation would be dissolving every second edge loop. Like this: Right-click, dissolve edges, boom. Optimisation! Can you tell the difference unless we zoom this close? You can't. This little 1 button press and 2 click (times 10) maneuver which can be done in less than 30 seconds just saved us 600 polygons. 600! On a single claw. Do this to all 4 claws on both hands, thats 8x600=4800 polygons just gone. Each claw taking 30 seconds, that's roughly 4 minutes for 4800 polygon savings just on the claws alone! We haven't even looked INTO the hand, the claw extends into the hand itself, geometry that is invisible to the user and could be mostly removed or reduced too. All of this can be applied to the foot claws, the head mane and if you really want to the arms and legs too without impacting visuals. We can easily get this model down to 45000 polygons before we have to start looking into less obvious places. Note that this entire process, from start to end takes roughly 30-60 minutes depending on the density and quality of the mesh, more density means more clickies but even if this was to take an entire day (which it can take if you split it up a bit, take your time and really look into every single tiny corner and optimise for even the smallest gains) it is something you as creator only ever have to do once for a model. You can also always keep a high-poly copy in case things go south. Remember to make multiple saves whenever you have done a chunk of work. With that being said, human models will most likely not look like the above avatar with fancy rectangles, they will most likely look very scuffed with triangles all over the place. Blender does have an option to automatically convert to rectangles but i'd not recommend that. In that case its their own fault and they are pretty much out of luck, edge loop selection wont work on these so they will have to manually select each edge, this is obviously going to increase the time by a lot but again still something that only needs to be done once. What they could try is selecting everything (CTRL + A) switching into the UV Editor and selecting everything in the UV Editor too (CTRL + A): And using the UV - Seams From Islands menu option This will automatically mark all edges as seams in the main render view: All red lines are UV seams, what they could try is selecting everything except the seams and then trying the Triangle To Quads option. Here's a triangulated mesh reverted back to quads: It does a... somewhat okay job but it breaks at seams, which is why you may want to deselect them beforehand (haven't tried it in that picture) it will leave them intact and allow you to manually optimise them. Anyway. The main point of this guide was to show you that it literally takes 5 seconds, 1 button press and 2 clicks to do an optimisation. 30 Minutes tops to do some rough optimisation given your model isn't a flaming dumpster fire of doom (and even then you can still optimise it but it will take longer, you might want to consider retopologising it alltogether). In that case i'd recommend this video i watched: It's an amazing guide and makes retopo really easy. Combine that with the Data Transfer modifier to transfer all weights from the original model to the new retopologised one and it makes optimising a breeze! I used the above video tutorial to retopo a really badly made hoodie and i'm quite happy with the outcome.

What are you on about? Until just 2 months ago i had essentially zero knowledge about Blender, let alone how to do any professional stuff. I have only ever been very slowly doing absolutely minimal moving vertices around because thats all i could do. Now that i'm fully committed to optimizing my stuff i looked up a couple tutorials and started doing a lot more. I'm literally a blender noob. These people have been working with Blender for a long time, there is NO excuse and NO valid argument that these people can't do the same, watch a couple videos and follow them step by step. This requires absolutely 0 confidence, 0 knowledge, 0 skill. ANYONE and i mean absolutely ANYONE can do it. I use the "source" (me) as example because i am bad at everything, i have 0 skills and like everyone was an average user at some point, until i decided to do more and actually started working on it, i use myself as proof that everyone can achieve these simple tasks. I made a Viewer with 0 coding skills, learned coding while doing it, i started making and optimizing models with 0 Blender skills, learned a lot about Blender and optimisation. I'm saying that anyone, especially well known content creators can do what is required if an idiot with 0 skills like me can do it, if they just wanted. I have seen a cripple who's entire body was paralized play and finish a solo round Mass Effect 3 on Gold difficulty, something i still have trouble today and i was pretty good and have 2 fully functional hands. He beat what i still can't most of the time, with his goddamn tongue and a stick in his mouth, how's that for your "invalid". But the ignorance and stupidity is not surprising considering that such a response came from you. I wouldn't be surprised if you are one of those that believe no-mod protects from copybotting.

Not sure what you see as the average user but seeing that VRChat runs at 20-36 FPS for the vast majority of people i'd say no. 36 FPS (Quest 1 with async reprojection) is completely fine. If you get anything but the Quest 1 you'll have 90-120 FPS max (that means you'll most likely run at reprojection framerate -> 45 / 60 FPS). I can very confidently say that after almost 8000 hours of VR (5200+ of which are VRChat) I recommend you sit down. As user of a shared world we should all be striving for good quality and optimization. Not to mention a lot of content creators make money for literal unoptimized trash, which could be optimized in less than an hour. It's not hard. I learned in a single day, it takes less than an hour to optimize a model and is a thing you only ever do once (and can also be pushed as update). There is no excuse we don't have "game-ready" assets especially from so called "professional" content creators. I'm sick of hearing this load of *****. There is very much incentive for each and everyone of us, especially content creators to optimize, we all share the same world, we all see the same thing, that means WE ALL suffer together from bad optimization. YOU make others suffer with your bad content.... and then some content creators even have the audacity to complain about SL's performance and put the fault at SL/LL. For once i have to say something nice about SL, for an overly complicated burning dumpster fire of a mess that is cobbled and held together by bandaid it is running surprisingly well considering how incredibly inefficient all of this mess is executed and the amount of trash it has to process. I can tell you that Unity breaks apart far before we even come close to an average scene of SL. Second Life doesn't need to be made for VR to sport optimized content. And their payment is every modder's payment -> word of mouth, good rating, downloads/uses, popularity, all of which translate into more units sold in SL.

What PekeNL said. Both needed packages are linked on the pre-download page and the FAQ!

Every other application immediately shows tearing, very noticeable. SL does not and has never in 15 years that i am in SL now. I'd see it, i see even the slightest tearing blindfolded, i hate it so much.

...but triple buffering is off by default. I don't have an override for my Viewer (or any for that matter) either, its globally off. So unless SL forces it thats not either.

Yes. I keep forgetting to turn off avatar movement by default.

Interesting, not only did i learn that the "composite rendering" how everyone called it back then in DWM that prevents tearing without VSync is called render ahead (that explains the driver setting) it also doesnt exist in OpenGl, now that begs the question, if its not render ahead that is preventing tearing in windowed applications with OpenGl, what is? SL will never ever have tearing, regardless of your settings or Vsync as long as you do not start SL in exclusive fullscreen (which to my knowledge is not possible in any Viewer and was removed a decade ago because is was a crashy mess), something is still preventing it from happening. Any insight on that?

Right i forgot you cant remap buttons in FS. I was under the impression that wss possible by now. In BD you can full remap all common actions plus some extra.

You are spinning uncontrollably because you have a controller plugged in. PS4/Xbox360/XBoxOne Controllers all tend to report very tiny values while the joysticks are in their idle position, commonly called stickdrift. Stickdrift should be very minimal and in the latest versions of Black Dragon this should not happen anymore due to the increased deadzones. If its still happen make sure that your controller joysticks are indeed idle (touch them and move them manually into a centered position if you have to). If that doesn't help go into Preferences - Camera - Flycam Configuration and adjust the Pitch/Yaw deadzones. Alternatively you could also simply turn off avatar movement for controllers, or select None in the Controller selection dropdown, or simply unplug your controller(s).

B for jump? You are either a madman or a Dark Souls player. Or both.

I think Ana is referring to when the user manually deletes these folders. It is... a very common thing for users to immediately go in scorched earth rather than deleting only the settings file for instance. I've had several instances where i was offering to look into their settings and manually fixing them (before asking them to delete the settings, making a backup beforehand) and before i could even finish my sentence they were already in the Roaming folder deleting everything because that's what they were always taught coming from Firestorm.

To be fair i'd very much root for the GPU being the cause here. Viewers have rarely been hitting the GPU (unless you enabled and set excessive shaders such as Depth of Field all the way up or had a low sun angle with shadows) so usually there would be a good chunk of free resources to work with, i'm saying that with a GTX 1060 which was hovering around 60% most of the time (with everything enabled). Now that the CPU can work down a lot more stuff, the GPU is also being worked harder, which in turn leaves less free resources for OBS and OBS from my personal experience has always been a very heavy performance hitter on its own and that was with free resources to spare. Now with your GPU resources being far more "limited" i suspect that your GPU simply can't keep anymore since OBS is quite a heavy GPU resource hog (i was told using NVidia's Shadowplay codec is supposed to reduce said impact close to zero, like Shadowplay/Share itself, i highly doubt as that was the first thing i tried and it did not make any difference). Personally i'd suggest giving Shadowplay/Share or the AMD equivalent (if you don't have a NVidia card) a try, they are very good at what they are supposed to do (recording the applications) and unless you are writing completely and obscenely large resolutions at insane bitrate and framerates on a super slow HDD you should not see any problems (infact in that case the recording would simply break and you'd see dropped frames inside the recording, your framerate would be fine tho). As far as i understand they literally catch the rendered images and save them to your disk, images that would otherwise simply go to waste are simply being saved from going to waste and instead copied from the GPU directly to your disk which eliminates 99% of all overhead making it almost free recording. (Which makes you wonder why this hasn't been a thing much sooner than 10 years ago)

The Viewer will first clear the old .old file, then rename the current .log file to .old and create a new .log file. There is only ever 2 files at all times, the currently running session (or one you had before closing) and the one before that.

I think the Viewer is recovering because its getting disconnected, thus breaking out and ending whatever it is that is causing the Viewer to freeze in the first place. I've seen this in the past. These are very rare special cases, something that needs to be so extremely specific, something probably not coming from the Viewer itself directly unless it is an infinite while/for loop that is skipped/broken out of when the Viewer detects that it has been disconnected, another thing i could think of would be your log files, you might want to check your log files when that happens, the Viewer might be spamming an incredible amount of log entries (we are talking hundreds of megabytes in a couple seconds, up to gigabytes of log data) which can cause the Viewer to stall to a point it doesn't recover until the situation is solved (e.g a disconnect breaking you out of whatever function or loop that is causing this spam). You can find your logs in: C:\Users\PCNAME\AppData\Roaming\VIEWERNAME\logs ViewerName.log and ViewerName.old

I was already informed of this yesterday. Apparently this is being shared on virtual secrets (another SL drama website), see below. Not only does this person not have any idea, nor was this person patient enough to simply do a reanalyse (which is what i did a couple hours ago), netting a fresh test of the same file and to no ones surprise all 60 apps detected it as clean. Jiangmin (like a couple others) as the name suggests is an asian antivirus, they are very well known to throw false positives at basically everything. Also whoever made this immediately started fearmongering to whirl up some more drama, clearly targeting to shoot me down. Not the first time. Shown in the above picture the very statement that i made the #1 copybot Viewer is proof enough that this is completely untrustworthy and someone who is simply out to attack me. I have never and will never make a copybot Viewer. Prior to Nirans Viewer i didn't even have any coding knowledge whatsoever and the development of Nirans Viewer into Black Dragon, all the way until today can be easily followed. To adress the elephant in the room directly. You most likely did not download the wrong file. To understand why antivirus apps go off you have to understand how these antivirus apps work, they are no magical all-defense application that will protect you from all viruses, in fact they are actually quite stupid and make a lot of assumptions. Most of the essential work they do is just check something (like the auto extracting WinRar) against their database of reported "malicious" behavior, if they match up they will flag it as potentially malicious, keyword being "potentially". They do not know, they are simply warning you the user that this file is acting similar to common viruses (according to their database), that is a self extracting archive is a very common thing to abuse (just like any installer), in fact most of the actual viruses cannot be detected in the first place because they base their reports on their database. If i were to write something malicious i'd most likely not write it like a common virus, which would make your antivirus unable to detect it as it doesn't follow any commonly reported behavior. Antivirus apps are basically just a giant database that compares behavior, behavior that has to be reported first, this means people will get infected and probably into trouble long before the virus is deteced and reported to the database, potentially preventing future infections. Anyway, as with anything if you do not trust it, send it again to VirusTotal, have it be retested, make a mental note of who actually reports it, make a manual scan with Defender or your Antivirus of your choice, it is very likely that its simply a false positive. Also do not ever download these files from anywhere but the official websites (that is in my case my blog which leads you through the pre-download page to explain some common problems and then into the actual download page: here) To adress these, they are as you already explained quite normal, Google Drive can only scan files up to 64mb? i think, maybe 100mb... so its just warning you that it couldn't test it, nothing special about that, the other being Microsoft (either Defender or sometimes your browser) giving you a warning that this file might be unwanted, this is because again its an executable (which are often used for malicious stuff) and it also doesn't have a MS Certificate (never had), that's also completely normal. You can somewhat safely ignore them, again don't just blindly trust everything but also don't freak out when something is up, simply read what it tells you (usually its just a harmless warning) and then decide for yourself what to do. The Viewer is fully open source, anyone can look into it and compile it themselves and if anyone believes i might have done something with the installer itself, you can essentially skip it, its a self extracting archive made with WinRar, this means you can simply open it with WinRar and manually extract the files without using the auto extraction expansion.

Pretty sure this has nothing to do with the Viewer at all. This has been a common thing for many years regardless of Viewer and the reason you see this after an update is because the Official Viewer tends to do regular cache clears after bigger/important updates. The Viewer then simply fails to fetch your inventory for what are possibly an infinite amount of reasons. Go to an empty place (empty LL sandbox for instance), make sure its not laggy, clear your cache and relog, next session open your inventory and type something into the inventory search filter (this tends to "kick" the inventory fetching). Keep repeating until your inventory finally decides to load.

I've hopefully made it clearer starting at 4.1.3. As it now reads "Download Latest Release here:"

Bottom line is keep it off on Windows, there is zero reason to use it, it slows down the viewer internal processes, causes additional input delay and is a bad framerate limiter. On Windows its impossible for windowed applications to have screen tearing thanks to the desktop composite renderer which acts as natural vsync without fps limit, if you want to limit your fps use the inviewer solution.

SL is played by 15 year old accounts that still don't know how to open chat after all these years. Other things most people don't know which are universal across all Viewers: Basic Camera Controls (Alt Zoom, Panning, Orbiting) Movement Controls (Arrow Keys/AWSD, E/PageUp, C/PageDown, F/End) Common Shortcuts (CTRL + P = Preferences, CTRL + I = Inventory, CTRL + C = Copy, CTRL + V = Paste, CTRL + B = Build, CTRL + 1/2/3/4 = Build Modes) Common Functions (Enter = Open/Send Chat, Right-Click - Reset Skeleton, Right-Click - Hover Height) Useful UI Functions (Arrow Keys in dropdowns, Enter to select them, Arrow Keys in sliders for smallest increment changes) I keep making fun of that one incident where someone asked me over IM how to chat in local chat (after 13 years of Second Life) but coming to think about it its a sad matter that after 10+ years lots of people still don't know the basics. I really can't repeat it often enough but after just under a year you MUST know the basics, there is really no excuse not to. Learning the basics is essential and takes you a week at most, they are so basic you are going to use them practically all day. Followed by assuming the user can read. And finally the last but not least is assuming the user can apply common sense. Combine all of these three mistakes and you've essentially included practically all UI design that is not "like Firestorm" because "like Firestorm". I think we can neatly wrap all the mistakes up as "It's not like Firestorm"

If your GPU set itself on fire it was a bad GPU (like those RTX cards that burned themselves up in the main menu of New World), they were badly manufactured or weren't respecting their set limits. They were essentially running full power without turning the fans up to cool them. It was a combination of faulty BIOS and hardware issues if i remember correct. This is something that should never ever happen and will never ever happen unless we have another company forgetting hardware basics. The exceptions of course are the rare badly manufactured and not properly tested units that somehow make it out of production. Again this is not VSyncs fault and you should never ever use VSync with this as your reason. There is no risk of frame tearing, as described above Windows (since 7 at the very least) has a composite renderer that makes it impossible for screen tearing on desktop. SL cannot run in fullscreen, there is no way you will ever see screen tearing in SL unless you manage to disable the composite renderer (or the forced VSync on desktop that apparently is supposed to be a thing i heard)

Dafuq are you implying? Disabling VSync simply allows the Viewer to run as fast as your hardware allows. That is most commonly your CPU at ~20% and your GPU at ~50-60% unless you have a low sun angle, are indoors with a lot of shadows or you are using heavy shader loops (Depth of Field) in which case you'll most likely run your GPU at 100% usage and quickly start dropping your framerate at which point VSync becomes unnecessary anyway. None of these will make anything "pop". You make it sound like allowing SL to run at higher framerates is a bad thing. VSync's main purpose is to sync the refresh rate with that of your monitor to prevent screen tearing (which is literally impossible to get in SL anyway because SL does not run in fullscreen and any window/borderless application is incapable of screen tearing as they are subject to Windows's composite rendering which prevents screen tearing). VSync is nothing but a trashy way of limiting your framerate, which potentially adds unnecessary delay to input and in SL's case also slows the internal processes down making you rez and load things slower. I'd highly recommend leaving VSync off unless you are running hundreds of FPS and want to save some resources while the Viewer is active (there is an option to scale the Viewer's refresh rate back while in the background)

Weird. Are you perhaps using the Front View camera preset and just didn't realize it? Go to Preferences - Camera, scroll down and select the Rear View (or Group View by default) preset, selecting them will automatically switch to them. You can also switch between them via the old Camera controls window.

You monster.