Flea Yatsenko
-
Posts
489 -
Joined
-
Last visited
Content Type
Forums
Blogs
Knowledge Base
Posts posted by Flea Yatsenko
-
-
There's no need for that. I am telling you what other people do and why there's so many problems with spec norm and why no one uses it.
PBR is a different rendering engine. We are not being forced to ALM. ALM is getting replaced with the PBR renderer. Forward rendering, I.E. the old way with ALM disabled is also getting removed from the LL viewer.
*shrugs*
-
4 hours ago, Zalificent Corvinus said:
I wish that would be true, but I suspect reality will be different.
Biggest problems with spec/norm isn't the system, it's people not bothering to RTFM, and assuming the spec map is a "specular strength map" instead of a "specular colour" map, and forgetting that their software is probably exporting DirectX based normal maps, not OpenGL based normal maps.
That was my point though, SL used a different spec map, normal with alpha set to openGL. Whenever you'd use a default export for any software for spec norm, it wouldn't do it right. You had to read the wiki page then set up a custom export preset. Hardly anyone ever did.
I know spec norm and PBR rendering usually isn't compatible. Which means SL probably has to convert the spec norm to PBR somewhere before it's rendered. Which means those incorrect spec norm maps are probably gonna be converted to PBR and they're probably gonna be very wrong. Which is probably why people are complaining about stuff looking drastically different. It was probably never done right in the first place but it worked in SL.
WIth spec norm, even downloading spec norm from texture libraries would give very wrong results. With PBR you can easily download the maps from sites and they'll work the way they're supposed to. Same with exporting from blender or whatever. It is easier to not screw up making content for SL with PBR.
I still think baked and only diffuse texture is still going to be very popular. I think content creators who want to do PBR are gonna have to support baked textures for a long time, especially with scenes and stuff. Even if ALM is removed, no one really used spec norm which means most people are still using baked textures so I don't think much is gonna change unless you have stuff that used spec norm, especially if they used spec norm the wrong way but it still looked fine with ALM.Exporting and importing with GLTF is way better.
Spec norm was doing more damage than good. At least with PBR it's gone.
-
8 hours ago, Zalificent Corvinus said:
It's been 9 years since SL got ALM-Materials, with normal maps and specular colour maps, and 99.99999% of SL still doesn't use those. PBR isn't going to change that. Flat, matte 2003 style texturing will STILL be flat, matte 2003 style, the only difference is more load on the servers sending the blank PBR data, and increased load on the viewers, checking the blank data is blank before rendering the old 2003 version anyway, and, the difference in HOW that's rendered, with the new lighting.
Spec Norm is a total pain to use. It never looks like it does in SL with whatever tool you're using. Trying to make a metal pipe with a modern workflow like substance painter or BSDF in Blender is a nightmare. You have to jump through so many hoops. I have all sorts of weird presets for exporting various spec norm materials for SL because just simply exporting some things do NOT look right at all in SL. PBR lets you export the GLTF from substance painter, blender, etc, and it looks almost identical in SL (minus environmental lighting).
Spec norm never worked right, there was never any consistency between SL and blender for me. Ever. Import would always be different. Mainly because pretty much everything else uses a PBR workflow and if you want to export spec norm you have to convert the PBR to spec norm and it seems like there's absolutely no standard between software for handling spec norm. Even if I did manage to keep spec norm workflow looking alright the full baked versions always looked better anyways.
Yes there's a lot of amateur content creators in SL. Converting to spec norm is a lot more difficult. PBR makes it a lot easier. In fact it's easier to export GLTF from substance painter or blender than it is to bake it. PBR has some really nice features but I think the real goal is to make it easier to use something other than flat baked textures on everything for creators.
Just having a lightweight implementation of PBR and the related workflow is going to make things way easier for creators to figure out how to make good looking stuff in SL.
-
1
-
-
I've got a 6800 in Linux and the PBR viewer runs a lot smoother than the ALM one. I am really confused what people are talking about with performance being poor. The frame rate numbers are similar but the frame time seems a lot smoother. If PBR is running badly you can just turn down reflections and it's way better. Disable them if you need to. I am pretty sure that's the most taxing part of PBR to run.
-
They did change the lighting system quite a bit. I am very happy with it, I almost always used a custom environment in my viewer and the PBR settings are great stock, maybe with just a little exposure setting changes. Just the lighting is a huge improvement, let alone using PBR materials or not. Just need to wait for people to get a grip on using reflection probes and it'll be really good.
-
1
-
-
10 hours ago, animats said:
Steam has about 20,000,000 users online right now, and they all have gamer PCs. SL has about 45,000 users online. Plenty of room for growth there.
SL on a phone screen is just too cramped for a big virtual world. It's a nice capability to have, but too limiting.
Not all of them do, some of them are on laptops or mid range computers.
https://store.steampowered.com/hwsurvey/videocard/?sort=pct
Roblox averaged over 70 million daily active users in Q3 2023. 500 million downloads on Google Play.
I'm not fond of smartphones either, but the market there is huge and it's a lot more lucrative (i.e. they will spend a lot more in game).
Phone is also a matter of convenience. It'll be a lot easier to get on SL on a phone than PC.
Mobile viewer will never ever replace the desktop viewers. You are right. No one is going to give up their desktop SL for a phone. But it should be able to bring in a lot of new users.
Not to mention it's based on Unity which means they are gonna have easy porting to a lot more than mobile.
There are lots of actual, legitimate games for mobile with worlds a lot bigger than an SL sim.
-
3
-
-
20 minutes ago, arton Rotaru said:
Last info I remember to the question; "if the mobile client will support PBR materials?", is a reply of a Linden that was something as follows. "We expect the mobile client to support the features of the desktop client." (Don't quote me on the exact wording. pls! Nor do I remember where I read that information.)
Although that isn't a "clear" answer, but enough for me to certainly skip blinn-phong materials in any future build I do.
Well it's using Unity so we can make an educated guess. Only thing I can find is this old forum post on Unity forums talking about how PBR can be disabled if the settings and turned too low. So I guess Unity can support PBR but it's going to be up to LL what they actually use.
https://forum.unity.com/threads/pbr-on-mobile.585388/
The wiki is kind of ambiguous. They kind of make it sound like mobile solutions will update to see PBR. Almost sounds like they aren't discussing their own client.
https://wiki.secondlife.com/wiki/PBR_Materials#PBR_Stand-In_Textures_and_Outdated_Viewers
EDIT: I sure wish they were a little clearer on all of this. I'd want to know if they're not going to allow third party viewers that don't have PBR, if mobile will support it, etc. Kind of hard to adopt this is new products with so much uncertainty.
Do you make new products with no Blinn-Phong diffuse, have a client not support it, then have people buying products with no textures? Or do you spend the time and effort to support old stuff only for it it to never get used when those clients are disabled? None of us know AFAIK. And it stinks because I would love to go full PBR.
-
1
-
-
5 minutes ago, Qie Niangao said:
but furniture and reusable building components really shouldn't have scene lighting baked into the materials, and that's what I fear will repeat from the bad old days when everything baked onto the diffusemap.
I fear that too. Baking is only meant to take care of shortcomings of the hardware or engine. The lighting in PBR is really good, people just need to figure it out. I also worry there is going to be a pretty big amount of residents who despise PBR and will never use it, and people who can't figure it out so just go full bright with bake. PBR is good enough you can just bake props with AO and ignore scene lighting and the viewer will make it look good. ALM didn't really succeed in that very well.
I don't think it'll be as bad as materials were. PBR is an industry standard and you can pretty much use any major game engine workflow or tutorial to figure it out. PBR has a right way and a wrong way. Materials had a bunch of different good ways, a lot of bad ways. And none of them gave you consistency between what you were using to make stuff. PBR fixes a lot of stuff. It just will be a mess because you really should be supporting legacy Blinn-Phong. Which means you have to texture once, package up, then texture again, package up.
My other big fear is people won't support Blinn-Phong, mobile viewer comes out, and people are buying things with no textures, with pictures that very clearly show textures. I'm also pretty certain there's going to be a group of users who hate PBR for whatever reason and do whatever they can to make sure a third party viewer supports ALM.
-
1
-
-
6 hours ago, Qie Niangao said:
Why should it cost a lot more to upload, once the pre-PBR viewers are cut adrift (and the sooner the better for that)? Even so, at most upload a baked diffuse map for them and if feeling super generous re-use the PBR normalmap that should work fine. After all, they'll mostly be non-ALM users clinging to whatever they can see, so they'd never see that normalmap nor specularmap if we bother to upload them them.
Otherwise shouldn't PBR usually use three maps, same as Blinn-Phong? In fact, depending what's being created, the albedo might just be blank with a tint. Also, again depending what's being created, the whole material should be re-usable for multiple products, perhaps with tweaks of the float parameters, won't it? (Or are creators going to bake scene lighting into the AO channel? I sure hope that's not how this will actually work.)
The official protocol is to upload the original diffuse with Blinn-Phong since that's a fallback if PBR doesn't work. Then you upload the three PBR maps, sometimes 4 if you're using emission. If you don't upload the original Blinn-Phong diffuse, any non-PBR viewer sees your object with no textures. But honestly it depends what you are making, if you are using tiling textures and not baking it's not that big of a deal. But if you want to bake or use textures that don't tile (because you are making something organic or natural and it needs wear or whatever) you might have some problems. You're supposed to have the AO in the ORM map, doesn't mean you need to. But from what I played around with it can make a very big difference since it's way better than the AO the viewer uses and depending on what software you're using, it can add shadows to the AO map from things other than geometry. Unless something changed, a non-PBR viewer will not use the albedo map if it doesn't support PBR. You can try looking at PBR stuff right now with an older viewer and see what I mean.
GLTF expects AO baked into the ORM texture. Doesn't mean you have to, but with what I make the baked AO makes a huge difference in quality. The SL Viewer simply can't make high quality ambient occlusion. Even if other games can do better, it's LL's game engine and they have to do the AO themselves, it's not like they are using Nvidia libraries or anything for AO.
I mean you can use tile materials and put them on everything but making any sort of super high quality stuff is going to take a lot of work. Especially if you have faith in the mobile viewer, because tiled diffuse with no AO will look horrible on mobile since the viewer probably won't be able to make AO or anything in real time. And pretty much every decent mobile game bakes most of the textures.
I really wanted AO on a separate map so you could tile and re-use the other maps and let the AO follow original UVs 1:1 but that was out of spec for GLTF so it wasn't really possible. LL really wanted to follow spec. But again it really depends what you're making. Some times tile textures work perfectly fine, like a modern and clean house. Sometimes they look absolutely awful, like stuff I make.
Don't believe me if you want but GLTF is supposed to use the AO map for more than just making parts darker, it also affects how light is reflected back.
https://docs.unity3d.com/Manual/LightingBakedAmbientOcclusion.html
Quote(Or are creators going to bake scene lighting into the AO channel? I sure hope that's not how this will actually work.)
Depends on what you're making, but it's supported both ways. It's an artistic choice and even a game engine like Unity supports baking every single ORM/Light Map for every object. In fact it's built into the engine.
-
1
-
-
Well. Unity decided to make people who use their engine pay PER INSTALL. There was massive backlash, and it was changed, but they only added revenue limits which makes me think LL would still have to pay for EVERY INSTALL on any mobile device. So there was a ton of drama with the engine LL is using for the mobile viewer. I wouldn't be surprised if someone crunched numbers and decided it could end up costing them money to release a mobile client. Unity backtracked a bit and SL is in a really unique situation given Unity's pricing scheme. So maybe they've worked something out. But considering what happened with Unity and how furious the internet was with them, I would expect that drama to at least delay the mobile viewer by a bit.
Second the last mobile client was cancelled because it was basically Speedlight except less good and residents started to wonder why LL was even bothering.
Third I'm pretty sure everyone on LL's level realizes high end desktop is more or less a dying platform compared to phones, laptops, etc and no matter what cool stuff they do with SL, if their only market is people with gaming PCs, they are never going to be huge.
Fourth the mobile client is one of the most important things to happen to SL. Mainly because if they can get SL to run on a mid range phone, that means it'll run great even on low end laptops. Which means a huge hurdle of SL running terribly on people's computers could be coming to an end.
-
2
-
1
-
-
Full bright is only for when you bake lighting, shadows, and everything in something like Blender. No way it will work with in world builds.
But yes, PBR is going to make every product take a lot more time to finish, cost a lot more to upload too. Price are probably gonna go up for me for a PBR version. At least if you ship two versions, you can get away with a customer using the old version until they can figure out EEP and PBR.
-
PBR is affected way more by the environment than materials or baked. PBR can look significantly better but you have to get the environment and reflection probes right. Otherwise it can look way worse, like "why is the blue from the sky coming into my underground tunnels?!"
There is going to be a learning curve. I think once people get the hang of it, it'll be amazing. That said, you can't turn PBR off in your viewer. But at least for the world, you can set a lot of stuff to full bright if it's baked and it'll look fine.
Even if you can't change EEP settings for the sim, you can change them in your own viewer. In Firestorm, it's just World -> Environment -> My Environments...
So really, there's two ways to go about fixing the way the scene looks. There is also an exposure setting in the graphics that you can play with and it'll make it look more like what you're used to.
You should always release a PBR and non PBR version of your products, unless you think the mobile viewer is going to completely fail. Mobile viewer won't support PBR last I heard. But no matter what a baked build in full bright, as long as it's not for an AV, is going to look fine like it always has. LL is going to eventually block non-PBR viewers besides the mobile one. And of course that won't look the same either. LL built in regular diffuse textures as a backup to PBR specifically for that reason.
There's always a rough transition period when LL gives us a major new feature like this. It was the same with sculpties, materials, etc. But Materials were massively outdated compared to other games (sorry to use that term) and PBR is an industry standard.
-
502 gateway error
-
1
-
-
On 11/22/2023 at 11:03 AM, Coffee Pancake said:
Yes .. and that should be presented in SL as it left photoshop. Not boosted a bit more by the local settings.
You shouldn't have to mess with exposure settings. In fact, you shouldn't even need to have that option.
Just saying LL's defaults for the PBR viewer are not good, but it's something easily fixed.
-
Happens in Linux too, running the open source driver. Either it's a problem affecting both drivers or the viewer. I tried editing the debug setting to change VRAM and it didn't help. It seems to get worse if you TP around to other places. It's so bad in Firestorm Alpha that I can't even get stuff to stay loaded long enough to get product images.
-
I think you are gonna find that 99% of marketplace images are already photoshopped and edited. I think almost all major merchants edit their photos before listing. I don't, but it's obvious they do because everyone has bought something without trying a demo that didn't look anything as nice as the picture.
The lighting is a lot different with GLTF/PBR but I found playing with the exposure settings can make it a lot better. I just think the default exposure is too high, I turn it down in my viewer. Even just lowering exposure to .9 stops it from crushing bright colors.
-
1
-
-
I think there is a bug or something with viewer detecting how much VRAM you have. I have a 6800 with 16GB of VRAM and with the Firestorm PBR viewer I get insane trashing. I saw them talking about it in the preview group as well. Seems like a new thing, just suddenly started happening, then I started using the PBR viewer and it got extremely bad.
Even if you had 32GB of VRAM, if the viewer is only detecting 512MB it's going to cause problems. Really hope there is a solution soon. This is so SL, the graphics get upgraded with PBR and everything looks way better but then there's a new texture loading problem and everything looks blurry and terrible.
I even deleted my firestorm_x64 folder to reset everything and it didn't help at all. I don't think this problem has to do with any settings unless there are new defaults and they are bad, or a new setting has a bad default that's breaking stuff.
-
I think SL has a very innovative userbase who can usually fix problems better than LL can. And realistically no one knows something like SL more than the users who can understand the problems. It's actually a really good thing SL users can fix SL's problems and it's something extremely rare that a lot of other companies could only dream of. I think LL kind of knows they are not entirely steering the ship. They are more like the people who keep the ship running and not sinking. And I think that's what makes SL so cool and filled with really awesome user generated content. It's a good relationship and it makes SL extremely unique, even amongst all these metaverse fad things popping up. I think LL knows to kind of keep their hands off the wheel and just let the users take control of most things. In world shopping was a pain, so we got xstreetsl. LL provided bodies were aging very poorly so we got mesh bodies. SL's users almost always fix LL's shortcomings and I think the current LL is kind of realizing that their role is to give its users the tools it needs to make SL better instead of trying to make SL better. And yes, I consider mobile client a tool to make SL better because it should in theory make for a lot more customers which would drive a big content creation push because there's more money and the market is larger.
The shopping experience is gonna get fixed soon, trust me
And yeah I had a rough day in RL so sorry in advanced if some of this post doesn't make sense.
-
1
-
-
Yeah but reflection probes can cause a ridiculous amount of lag. Not sure how to deal with it as a vendor, if I add it to each piece in a builder's kit it'll make building easier but the optimal way is to place it after the build is done where it's needed. There are going to be a lot of teething issues for sure. But PBR is going to be amazing and we're going to get a lot of stuff that looks amazing in SL that people aren't expecting.
-
There is also a big warning on PBR test regions that you shouldn't post PBR stuff on the marketplace yet. I don't think you need to rush into making PBR stuff, it's going to be a while before everyone is on a PBR viewer. But you should be learning how to deal with it. Just a heads up, I was ready to go full PBR before I saw that sign and started thinking about it.
-
2
-
-
Mobile client is using a game engine not developed by LL. It uses Unity. LL won't be responsible for all the backend and graphics stuff with the mobile viewer, only making SL work with another game engine. The current SL Viewer is basically a game engine LL has created to handle SL. It has to do graphics, networking, graphics API (OpenGL, Vulkan, etc) support by itself and LL has to make it.
Mobile viewer will be capable of whatever Unity Engine is capable of, which is a lot. Here are the platforms it supports
https://support.unity.com/hc/en-us/articles/206336795-What-platforms-are-supported-by-Unity-
Meaning mobile SL viewer could be ran on everything on that list (theoretically). For APIs (directX, OpenGL, Vulkan, Metal), here's what Unity supports
https://docs.unity3d.com/Manual/GraphicsAPIs.html
So yeah, theoretically, it's possible to get all sorts of optimized builds and LL doesn't have to do much work compared to the official viewer. LL has the potential to make SL much bigger than just a mobile client, but we'll see what happens. This is one of the most exciting things to happen in SL in a long time since it has so much potential to reach new users.
-
1
-
-
502 errors the last week have been incredibly bad. Worse than I've seen that I can remember. Happens on secondlife.com and the marketplace.
-
2
-
-
How dumb do you have to be to run a scam on a completely controlled, centralized economy like SL? LL controls every part of the L$, every transaction, etc. It's not like getting grandma to buy a gift card to "pay the IRS." How are you supposed to even get rid of your scammed L$? Even if you sell it in some shady way like not on the exchange, LL can still pretty much know where it's going, where it came from, etc.
-
SL has gotten a lot better with performance too. I have my old AMD FX 8350 I use sometimes and it is actually quite usable with an old HD 6900 and FX 8350 in Linux. Much better than I remember. Still not great on my AMD A4-5000 but that's a 2013 netbook APU so I can understand.
I think LL doesn't get enough props for what they've done with viewer performance, and PBR is going to shift massively from CPU limited performance to GPU limited. Between CPUs getting better and viewer getting better, the bar for running SL is lowering. Which is great because that can be a huge hurdle for new users, specially those who don't have proper gaming PCs.
-
1
-
Facebook
Instagram
Twitter
YouTube
Flickr
Code Cleanup & Memory Usage Question
in LSL Scripting
Posted
If you want to fix your existing code instead of trying a different solution, make the
if (llGetLinkName(counter) == "Engine Cell 1")
line instead match a variable, then add that as an argument to your function
Engine(string engineName){
...
if (llGetLinkName(counter) == engineName)
...
}
It also lets you get rid of
if (llGetLinkName(llDetectedLinkNumber(0)) == "Engine Cell 1") {Engine1();}//Cell 1 else if (llGetLinkName(llDetectedLinkNumber(0)) == "Engine Cell 2") {Engine2();}//Cell 2 //Else if Repeats more of above ^ that would other wise be placed here.and replace it with Engine(llGetLinkName(llDetectedLinkNumber(0)));