LL Reaches Out on PBR

[Fluffy Sharkfin]

23 minutes ago, Luna Bliss said:

That reminds me of the advice I give some landscapers -- to put the less resource-heavy stuff to the back (although I'm not sure every sculpt would qualify these days, even if they are  less LI), and then put all the high LI, very detailed mesh up front covering up the less desirable stuff in back.

There are a few tricks used by environment artists that could help to improve performance in content heavy areas (and plenty which just aren't applicable to SL).

There's also the issue of people cranking up their LOD settings to "make things look better" which causes a lot of content to appear with very dense polygons.  The issue isn't just the number of polygons but also the fact that small polygons are incredibly wasteful to render since the GPU renders pixels in "chunks" and when a polygon is smaller than the chunk of pixels being rendered then any pixels that don't contain parts of that polygon are discarded.  So if you're rendering a bunch of tiny polygons that only cover a single pixel then you're wasting 75% of your GPU time (chunks are most commonly 2x2 pixels in size).

I posted a video a while back which explains it all in depth along with a practical (if rather extreme) demonstration of how an RTX4080 can be reduced from 100fps to 3fps.  The video also references white papers from both Nvidia and AMD detailing how this affects pretty much all GPUs made in the last decade and, although it gets a little technical in places, does a pretty good job of explaining the principles and why trying to render millions of tiny polygons is a really bad idea.

 

Edited August 25 by Fluffy Sharkfin
extra info...

[BilliJo Aldrin]

2 hours ago, Scylla Rhiadra said:

Oh, I don't know about that! I keep sticking my hand into the fire here, don't I!

Look, this is easy. The higher and more complicated the barriers and obstacles are to getting into a game of any sort, the fewer the number of people you'll attract or retain. And god knows SL needs to attract and retain.

HOWEVER, the key point with which I began this entire thread is that LL has at last seen the light, recognized that they've endangered or at least potentially damaged the platform by making it tougher for people with old or low-end machines, and they are, apparently, doing something about it.

If they do it well enough, relatively fewer older users will have to stare at YouTube videos working out what a "dual core processor" is (always the most fun part of playing any game, amirite?) or a "discrete GPU" (one that won't tell your gf that you've been watching virtual porn), and newer users will be more likely to not find the obstacles to trying it out off-putting.

All the while retaining SL's move into PBR and a bright, shiny new future!

So it's all good, really.

If they have seen the light, why don't they just say ooops we made a big mistake introducing PBR, and get rid if it.

Its not like they have to compete with a newer and better version of SL that someone else brought out, it doesn't exist, and probably never will.

[Fluffy Sharkfin]

1 minute ago, BilliJo Aldrin said:

If they have seen the light, why don't they just say ooops we made a big mistake introducing PBR, and get rid if it.

Its not like they have to compete with a newer and better version of SL that someone else brought out, it doesn't exist, and probably never will.

Yeah and it's not like they need to maintain a certain number of customers or make money or anything like that, they can just leave SL running forever just as it is and won't ever have to care about the fact they're a business not a charity.

🙄

[Porky Gorky]

59 minutes ago, Luna Bliss said:

Yeah the only negative thing about doing that (forcing optimization) is that it might eliminate those who just want to create, mess around for fun, and never plan to have super professional skills.

LL should add a free library of pre textured, modular mesh assets that are built to a relative scale and which are easy to snap together so users can mix and match elements. Something similar to what is available in Fortnite Creative, simple enough so a 10 year old can effectively build a level with practice. 

New users could then play around, building from this library. Once they have exhausted the library it may inspire some users to go out and buy from content creators or to learn to create content themselves. Creating an asset library should be easy enough, just farm that job out to the moles.

[Zalificent Corvinus]

6 minutes ago, Fluffy Sharkfin said:

they're a business not a charity.

A business doesn't drive off actual paying customers on the vague hope that non customers will magically appear to replace them, for no actual valid reason.

[Extrude Ragu]

You guys still at it? 😮

[Zalificent Corvinus]

3 minutes ago, Porky Gorky said:

LL should add a free library of pre textured, modular mesh assets that are built to a relative scale and which are easy to snap together so users can mix and match elements. Something similar to what is available in Fortnite Creative, simple enough so a 10 year old can effectively build a level with practice. 

Pretend this isn't some AAA Game, with a Modders Creation Kit, where players can build their own levels with stock parts, using some snap-to-grid option that's enabled by default for the modular tiles.

 

In fact, don't pretend.

 

[Luna Bliss]

I think somebody will 'be at it' until we all burn up in hell together...

[Porky Gorky]

8 minutes ago, Zalificent Corvinus said:

In fact, don't pretend.

No pretense required, whilst SL is not Fortnite, it would be totally possible to implement a system similar to this. It will obviously be clunky, finicky and generally problematic, but that is the vibe in SL so why change things now. Having such a library would be better than not having such a library.

[gwynchisholm]

I feel like just the other lighting settings like shadows and stuff do a lot more to negatively impact performance than PBR does

Example, heres a place that is oriented around PBR content with all the fancy settings turned off, no reflections, no mirrors, no shadows, etc

and here it is with all those settings turned on

PBR materials and stuff are hard carrying making this look good, with very little impact on performance compared to the lighting, i get like 50fps in this scene with all the lighting and reflection stuff turned off, and maybe 10-15 with it on, and it still looks really good at 50fps with all the lighting off because of the PBR objects

i try not to think of it like, PBR is tanking performance, i think its more like you can really reduce a lot of other settings that more heavily impact performance, and the game will still look pretty good because of PBR

 

context:

"CPU: Intel(R) N100 (806.388 MHz)
Memory: 8015 MB (Used: 4768 MB)
Concurrency: 4
OS Version: Microsoft Windows 8.1 64-bit (Build 9600)
Graphics Card Vendor: NVIDIA Corporation
Graphics Card: NVIDIA GeForce GTX 750 Ti/PCIe/SSE2
Graphics Card Memory: 4009 MB
Graphics Card Memory (Detected): 4009 MB
Graphics Card Memory (Budget): Unlimited

Windows Graphics Driver Version: 30.0.14.7212
OpenGL Version: 4.6.0 NVIDIA 472.12"

This is not a fancy computer, i got rid of the fancy computer because it was a waste of time, money and electricity. And im staring at PBR stuff and having no issues. Im not really understanding where this gripe about the performance impact is coming from, because im barely seeing it. If this nugget of a computer is playing SL fine, what on earth are some of you playing on?

 

[Fluffy Sharkfin]

4 minutes ago, Zalificent Corvinus said:

A business doesn't drive off actual paying customers on the vague hope that non customers will magically appear to replace them, for no actual valid reason.

They also don't allow their product to stagnate for years and lose a huge portion of their customers but here we are.

Since I don't have a crystal ball I can't tell how well their attempts to improve SLs appeal to a wider audience will turn out but I'm hoping for all our sakes that they manage to do something to draw in new residents because, regardless of PBR or any other recent updates, SLs population continues to decline and has done for years.

Also, exactly how many of their customers have they driven off so far given that there are still multiple non-PBR viewers available that apparently won't be deprecated in the near future and a whole world full of legacy content for people to enjoy, not to mention a large number of creators who probably won't even bother with PBR for at least a few years?

[Fluffy Sharkfin]

16 minutes ago, Porky Gorky said:

LL should add a free library of pre textured, modular mesh assets that are built to a relative scale and which are easy to snap together so users can mix and match elements. Something similar to what is available in Fortnite Creative, simple enough so a 10 year old can effectively build a level with practice. 

New users could then play around, building from this library. Once they have exhausted the library it may inspire some users to go out and buy from content creators or to learn to create content themselves. Creating an asset library should be easy enough, just farm that job out to the moles.

All of this!

3 minutes ago, Porky Gorky said:

No pretense required, whilst SL is not Fortnite, it would be totally possible to implement a system similar to this. It will obviously be clunky, finicky and generally problematic, but that is the vibe in SL so why change things now. Having such a library would be better than not having such a library.

And this too!

The idea of having modular building components is no different to all the full perms building kits that people are using already.

The benefits of having a library of modular assets isn't just that it gives users a reason to build things in world instead of just buying prefabs (which most of the time don't really resemble what they want so much as being "close enough"  because they're sick and tired of scrolling through the marketplace).  If people are all using assets from the same library to create different things that means more assets precached locally and less unique assets to download.  Shared libraries of textures would similarly reduce the amount of VRAM usage rather than having 30 slightly different brick textures to download because everyone is using textures from different sources.  Of course you can still have the option to upload your own textures and meshes but a library of high quality premade assets would be an invaluable resource for everyone and I really don't see how there would be any downside to it.

[Ksenia Elcano]

4 minutes ago, Fluffy Sharkfin said:

Since I don't have a crystal ball I can't tell how well their attempts to improve SLs appeal to a wider audience will turn out but I'm hoping for all our sakes that they manage to do something to draw in new residents because, regardless of PBR or any other recent updates, SLs population continues to decline and has done for years.

You don't need a crystal ball. Instead, make a forecast based on how successful LL has been at previous attempts to improve the platform and attract more residents. So what is their batting average?

I can think of only one success, which is Bellisseria. Other than, LL has been a fail factory, with Sansara being the most colossal failure. So my forecast is that LL attempts to improve SL to appeal to a wider audience will not turn out well.

[Fluffy Sharkfin]

Just now, Ksenia Elcano said:

You don't need a crystal ball. Instead, make a forecast based on how successful LL has been at previous attempts to improve the platform and attract more residents. So what is their batting average?

I can think of only one success, which is Bellisseria. Other than, LL has been a fail factory, with Sansara being the most colossal failure. So my forecast is that LL attempts to improve SL to appeal to a wider audience will not turn out well.

Right, so LL have failed in the past and therefore should just give up trying?

Bellisseria did nothing to attract new customers, all it did was encourage a bunch of existing customers to upgrade to premium (and make a few premium alts so they could play Game of Homes).

What do you think is going to happen if LL does nothing at all and the population continues to decline?

[Zalificent Corvinus]

20 minutes ago, gwynchisholm said:

in this scene with all the lighting and reflection stuff turned off,

Congratulations, you just made your PBR look no better than Blinn-Phong, and in fact, due to the new tone mapping and the sub standard rendering engine combo, with the de-saturated over contrasted thing, and the glowing cyan neon water, it actually makes SL look worse.

 

So, your experience basically says "PBR, it's worthless".

 

[gwynchisholm]

So on the modular building thing, other games have done this, and it will be appealing to a different audience, but not all audiences. Anyone whos really into the idea of creating something in this virtual world is going to figure out how it works, given enough time. And modular building systems are pretty limiting and not that appealing to the people who would with time, make something impressive.

Roblox did this in 2011, where they introduced a modular building system, everything done in cubic form. Terrain, building blocks, furniture, etc. All of it fit a cubic grid. This system was right alongside the conventional building system which was the Studio program, which is laid out like 3D modelling software and is not easy to use for a complete novice.

There was still some population who looked at the modular building and went "ok now what" and learned how to do more. But a lot of people just felt it created a dichotomy of creations in the game, those which were made with that modular system, and those that were "real".

And years later they took it away, its just a hidden feature in the studio program. Instead they focused a lot more effort into making studio more accessible, and teaching new users how to use it. That ended up being more effective in terms of getting users to understand how things are made ingame, and what they can do with the platform.

But man almost any game did this, and theres always some higher level of design that people will implement and if the result looks good, people will learn how to use it. I presume a bunch of people here have played habbo hotel? *ahem*

Roller Stacking

Basic principle, all items in habbo fit to a two dimensional grid. Only certain items can be stacked on top of other items, like on table surfaces and such. But there are these roller items, you put something or yourself on top of them, and they move in the direction the roller is facing. If you roll items into eachother, they can occupy the same tile, and will retain the height of the base of the roller.

Now apply this idea, but start stacking rollers on top of other items. Now you can roll items into a 3D dimensional space. This for a very long time was one of the only ways to get a 3rd dimension in your builds. And on god is it the most tedious thing you can do. But the end result is that it literally opened an entire extra dimension of building, and with careful placement, timing and organization of rollers, you can build things otherwise not thought of as possible with the limitations of the game.

People who want to do this, who want to make something extravagant, who will apply whatever effort it takes to do so, do not care what the tools are to do it. This is one of the least intuitive ways of doing anything ever, that entire build wouldve been done row by row because you couldnt go back and change anything without undoing work in front of it, since the only way to get items up high is to stack and roller them. If they accidentally deleted that golden piggy bank upstairs, theres no getting it back without undoing half the build.

And yet people still do this, they still learn how to do this nonsense by pure ambition to do so.

I dont think SL is much different, its really not that hard to build things in this game, its just not talked about much. I dont think they need some more accessible building tools, i just think the existing building tools could use some better documentation.

Edited August 26 by gwynchisholm

[Zalificent Corvinus]

14 minutes ago, Fluffy Sharkfin said:

The idea of having modular building components is no different to all the full perms building kits that people are using already.

And how many people buy and use those, she said as somebody who does...

 

[Fluffy Sharkfin]

Just now, Zalificent Corvinus said:

Congratulations, you just made your PBR look no better than Blinn-Phong, and in fact, due to the new tone mapping and the sub standard rendering engine combo, with the de-saturated over contrasted thing, and the glowing cyan neon water, it actually makes SL look worse.

25 minutes ago, gwynchisholm said:

it still looks really good

 

1 minute ago, Zalificent Corvinus said:

So, your experience basically says "PBR, it's worthless".

No, that's just your mischaracterisation, as usual!

[gwynchisholm]

2 minutes ago, Zalificent Corvinus said:

Congratulations, you just made your PBR look no better than Blinn-Phong, and in fact, due to the new tone mapping and the sub standard rendering engine combo, with the de-saturated over contrasted thing, and the glowing cyan neon water, it actually makes SL look worse.

 

So, your experience basically says "PBR, it's worthless".

 

most of SL looks like this:

so im fairly impressed with what PBR is doing in terms of general aesthetics with minimal performance impact, when the above is my baseline

[Fluffy Sharkfin]

Just now, Zalificent Corvinus said:

And how many people buy and use those, she said as somebody who does...

 

Too many buy them, not enough use them!

If they were free and people got to experience how easy it is to create things for themselves using them then they'd probably be far more popular!

[Zalificent Corvinus]

3 minutes ago, gwynchisholm said:

I presume a bunch of people here have played habbo hotel? *ahem*

Roller Stacking

Oh boy, somebody trying to explain "modular building" with reference to some lame PoS game.

Go watch some building tutorials for Sims 4, and see how limited your Crappo Hotel is.

 

[gwynchisholm]

1 minute ago, Zalificent Corvinus said:

Oh boy, somebody trying to explain "modular building" with reference to some lame PoS game.

Go watch some building tutorials for Sims 4, and see how limited your Crappo Hotel is.

 

[Fluffy Sharkfin]

4 minutes ago, gwynchisholm said:

I dont think they need some more accessible building tools, i just think the existing building tools could use some better documentation.

Yeah, I wasn't really arguing for an overhaul of the building system (not that there aren't improvements that could be made) but rather for the availability of modular assets that people can use to build with in-world instead of being trapped in this dynamic of creators providing prefabs and them choosing which one to settle for.

[Fluffy Sharkfin]

3 minutes ago, Zalificent Corvinus said:

Go watch some building tutorials for Sims 4, and see how limited your Crappo Hotel is.

 

[Zalificent Corvinus]

5 minutes ago, Fluffy Sharkfin said:

No, that's just your mischaracterisation, as usual!

It's not mischaraterisation.

 

Take away the fancy lighting/reflection options, and for non-metallic surfaces, PBR offers almost NO improvements over Blinn-Phong.

And the tone mapping they imposed for it, does make things desaturated and over contrasted.

 

When the revised "looks good" is compared to flat matte 2003 style, rather than Blinn Phong, that's a pretty low bar for "looks good".