Jump to content

Future of the metaverse, and all that


Recommended Posts

2 hours ago, Drayke Newall said:

There would be no difference between Unreal 5 tech demo and a demo of 100 characters independently outfitted as those outfits irrespective of numbers would be created by the studios themselves therefore be optimised or at least made to work with the system. They are just simply assets just like a cardboard box would be.

I don't think so. First, notice how the tech demo skipped completely the character's specs. UE supports pretty high poly characters no problem, but those have a few limitations in regard of their skeletal deformation. Starting from the practical issue that the higher the polycount over a certain density threshold, the harder is to get good smooth and realistic deformations, considering that influence limit is still 4 joints per vertex exactly like in SL. After the working stage is completed and the rig works, there's something to think about the gpu workload that happens after it. If the polycount itself and the related deformation dataset isn't actually a problem for modern cpu to process in order to send them to the gpu, the actual deformation then happens in the gpu and it's updated every frame, with no trickery possible to skip frames for easier interpolated calculations like it is seen for the GI (see the lighting at the cavern entrance for example, when the sunlight changes angle). Multiply this for 100s of characters and it may become a problem in the cpu as well as in the gpu. Add to all this also the need for dedicated materials and textures, whereas environment scanned assets can be worked with scanned texture data made tileable and reusable, characters can't afford such technique and their shaders include subsurface scattering and transparency in the hair, which are an added resource intensive process when added on top of a regular shader without it, even though their new GI system is able to fake it some other way, it's still an additional computation task that scales up. So no, a character is definitely not simply yet another asset like a cardboard box is, even just under a mere rendering point of view. Then there is all the related and additional data required for them to BE characters, like animations and state machines that have to be loaded, the physics system for the character controllers and the scripts to make everything run, unlike a cardboard box there are many additional factors. 

  • Like 1
Link to post
Share on other sites
  • Replies 371
  • Created
  • Last Reply

Top Posters In This Topic

Top Posters In This Topic

Popular Posts

Being surrounded by Chipmunk-sounding tiny furries is not my idea of fun. I get that it's your entertainment, but it's rather insulting to say that having no desire to be surrounded by Chipmunk-soundi

Or, where a venture capitalist thinks SL-like systems are going. The Metaverse: What It Is, Where to Find it, Who Will Build It, and Fortnite  Is Fortnite the next Metaverse? The Most Imp

Posted Images

3 hours ago, OptimoMaximo said:

I don't think so. First, notice how the tech demo skipped completely the character's specs. UE supports pretty high poly characters no problem, but those have a few limitations in regard of their skeletal deformation. Starting from the practical issue that the higher the polycount over a certain density threshold, the harder is to get good smooth and realistic deformations, considering that influence limit is still 4 joints per vertex exactly like in SL. After the working stage is completed and the rig works, there's something to think about the gpu workload that happens after it. If the polycount itself and the related deformation dataset isn't actually a problem for modern cpu to process in order to send them to the gpu, the actual deformation then happens in the gpu and it's updated every frame, with no trickery possible to skip frames for easier interpolated calculations like it is seen for the GI (see the lighting at the cavern entrance for example, when the sunlight changes angle). Multiply this for 100s of characters and it may become a problem in the cpu as well as in the gpu. Add to all this also the need for dedicated materials and textures, whereas environment scanned assets can be worked with scanned texture data made tileable and reusable, characters can't afford such technique and their shaders include subsurface scattering and transparency in the hair, which are an added resource intensive process when added on top of a regular shader without it, even though their new GI system is able to fake it some other way, it's still an additional computation task that scales up. So no, a character is definitely not simply yet another asset like a cardboard box is, even just under a mere rendering point of view. Then there is all the related and additional data required for them to BE characters, like animations and state machines that have to be loaded, the physics system for the character controllers and the scripts to make everything run, unlike a cardboard box there are many additional factors. 

Ok ill accept that and was a bad analogy on my behalf. I do agree they are more complex to render, animate etc however it all comes down to exactly what the poly count of the characters are as well as their resource intensity. I don't expect them to showcase a engine tech demo on that as that usually comes down to the game itself. For instance the tech demo of unreal 5 could be argued that it is a tomb raider game demo that you will only see 70% of the time only 1 character. Why do epic need to showcase such complexities when games in the past using previous versions of their engine have done such things? That is to say, numerous amounts of people controlled characters in one area all wearing different textures and it has handled it fine.

The tech demo by Epic for UE5 was purely highlighting their downscaling, new GI and animation settings. All tech demos of engines highlight only the improvements over the older systems. There is no scenario where an engine or game will not have optimised content and will not use textures that are already installed and fetched by the engine from the system.

If Molly is looking for a prime example of what he is talking about, take Planetside 2 for instance. Whilst it cant compare with the graphics of modern systems as it was made in 2012, for its day it was leaps and bounds above the competition, it allows for different user selected customisation (not user created like sl) with camo in numerous forms, different weapons from players, vehicle camo, decals, armor selections, effect choices and yet runs perfectly smoothly with no real system drain. Effectively for only 100 people it is entirely possible for the engine to render a different textured and meshed character without any performance hit. Granted it doesn't allow for character customisation to the body shape, face and hair but, like I said, this is a 2012 game where back then it was more often a case of select x body, x head and off you go. What's more games like COD and Battlefield were fixing server numbers to 32 and 64 respectively.

Planetside 2 can handle 250 players per server without any form of player render culling (and all in one area attacking one base). Over this it may start to cull depending on load and will only load those near you that are a threat. Its theoretical maximum character limit per server is 2000 and its record player count is 1158 players on one server at a time. So whilst I will agree with you that characters are entirely more complex than a cardboard box they are still assets and having 100 people in the same area with different looks to others is entirely possible without any issue. If a 2012 game can do it I highly doubt a 2021 engine would suffer from it.

If you want something more comparable to second life, APB Reloaded (a game from 2010) can handle 100 people per playing district and has a far better character customisation system that SL (the equivalent to slightly under black desert online customisation). It also allows for custom user created clothing (though only using in game meshes so designs can be custom), logos, tattoos, camo, music etc. So no, sorry I stand by my point that it is entirely possible for 100+ people in an area with complex render systems in place. It would be a step backwards for Epic to release an engine that cant handle that.

If you're after something what streams textures and meshes made by people like SL well I cant provide such an example as it doesn't exist but that wasn't what Molly was implying.

UE5 is entirely around downscaling so even if they were to replicate any of the 2 above games in their system and times the avatar complexity by 10, its downscaling would still, I would dare say, make such things possible without any issue.

EDIT

Just as an aside and building on character customisation mentioned above this shows how outdated SL's UI is not to mention how badly implemented appearance editing in sl is. This vid below is what the outfit manager in Second life should be like. Simple menu on the right where it access your outfit folder and only requires 1 click to add the clothing. If they could replicate this system as well as the clothing creation system to something like APB Reloaded SL would be improved 1000x and make it be able to weather a lot more from competition, user retention, fun scale 1000 etc. This is the level of inworld customisation that is needed in SL for average users that cant create custom content out of world. Keep in mind this is a 2010 game that is still playable.

Without this level of UI friendliness, clean simple look and editing capabilities SL wont go anywhere. For those that are going to say but its user created stuff etc, why cant it still be. Have the user created stuff in the outfit section that perhaps isn't modifiable but allow others to create and sell mesh that is textureless and can add bought textures onto it.

To be honest when I first heard LL were doing BoM I was envisaging the above video regarding texture being able to be overlayed, not the mess they made. That is to say THIS is baked one mesh done properly. The same level of customisation with overlays can be applied to objects like cars etc as well so isn't limited to just avatars.

Edited by Drayke Newall
added vid
  • Like 1
  • Haha 1
Link to post
Share on other sites
12 hours ago, Drayke Newall said:

What LL do is akin to a gaming studio saying they are going to release a brand new game but make it so that it only runs on the original unreal engine and consequently hampering any optimisation or feature possibilities that the new versions offer all because some people might have a windows 98 PC. These gaming companies or really any company just wouldn't do that.

A better example would be WoW where, they introduced later expansions that completely DESTROYED the entire previous expansions for end game players. This meant that the original game, quests, etc are now impossible to experience. They didn't listen to their current userbase saying but if you do that it will break x or mean I wont be able to get that particular piece of gear. They just did it and ended up increasing their playerbase at the loss of a minority.

Business have to break things for things to work in some cases and unfortunately the one platform that absolutely needs it is Second Life for it to progress but they just wont do it.

is a pretty interesting topic for sure

i think virtual worlds are different from games in how they are approached by the provider companies. Game companies are like Apple, they have limited/restricted design form factors - built-in obsolescence. The longterm successful game companies produce new experiences (and/or titles) to fit the new form factors on relatively short cycles

whereas the virtual world companies that survive over the long term, don't produce new titles (whole other new virtual worlds). Their survivability is sustained because of the Microsoft-like approach to their customer base - incremental improvement, backward compatibility, don't break stuff

mostly because virtual worlders are not like gamers who jump from title to title (experience to experience). Been there, done that, and on to the next. Gamers move fast, and game companies respond to this and incorporate this into their business model. Providing new limited form factor games (experiences) that are broken and obsoleted on relatively short cycles
 
virtual worlders don't move fast, not in the same numbers that gamers do, and have shown over the years that they have little to no inclination to do so. Jumping from title to title (experience to experience). OpenSim, Hi Fidelity, Sansar, etc. Is just not in the make up of virtual worlders to do this

this said I think there are major things that Linden can do, on the long cycle incremental path. Which I think they will do. Some are:

1) Water physics. Linden ocean waves which physical objects can interact with. A UX switch: enable/disable water. Enable and a water vehicle is going to get buffeted by the waves. Disable and the vehicle will not get buffeted and run flat as it does now
2) Flex mesh objects. Again a UX switch. Enable/disable. When enabled the flex will interact/move with the environment
3) Linked object hierarchy. Might not be possible with existing objects. Would tho be doable I think with a new flex mesh object type. To simplify development this object type can't be linked to the old types (prim, sculpt, or old mesh)

Link to post
Share on other sites
5 hours ago, Mollymews said:

is a pretty interesting topic for sure

Snip

Oh I agree 100% with what you have said dont get me wrong. The problem is that by keeping backwards compatibility LL are shooting themselves in their own foot. Even using your example of Microsoft being more similar to SL; Microsoft, when they release a new platform, dont make their OS backward compatible from the get go. It requires you to then download those additional compatibility patches i.e. you will download a DirectX9 file to play an older game and even then sometimes it just wont work. There are numerous old games and programs that don't work on Windows 10 because they get broken and the only way to fix them is either patches by the game/software maker or a third party.

Try for instance the vid of APB Reloaded I posted above. It uses a baked on mesh and outfit (wardrobe) system that is far superior to Second Life. If LL were to try and now improve on the baked on mesh system to something comparable to that APB Reloaded system they will either break their newly released BoM system as content for their 'stage 1' version has already been made and therefore wont work, or they would have to spend countless years trying to adapt the existing code to it (making it even harder and longer to impliment) or the avenue LL usually take, keep the status qou and just don't updated it.

That is more how I see things, in that there are times that to improve the system it is 100% a necessity to break the current content even in a perpetual VW so as the platform can improve to a modern standard. I hope that if any LL employee looks at my post here and above that the seriously download APB Reloaded and look at their UI for a clothing wardrobe and BoM system. It is a simple UI interface that is user friendly and allows for (i.e. more powerful) far better outfit customisation than anything SL offers.

Their major problem for years has been that the UI and outfit change system is just to user unfriendly and that needs to change. But alas I wont hold my breath that they would even look at the above vid or change it to something better. It is these sorts of changes that will keep SL modern and able to compete with any other VW or MV coming.

5 hours ago, Mollymews said:

this said I think there are major things that Linden can do, on the long cycle incremental path. Which I think they will do. Some are:

1) Water physics. Linden ocean waves which physical objects can interact with. A UX switch: enable/disable water. Enable and a water vehicle is going to get buffeted by the waves. Disable and the vehicle will not get buffeted and run flat as it does now
2) Flex mesh objects. Again a UX switch. Enable/disable. When enabled the flex will interact/move with the environment
3) Linked object hierarchy. Might not be possible with existing objects. Would tho be doable I think with a new flex mesh object type. To simplify development this object type can't be linked to the old types (prim, sculpt, or old mesh)

It would be wonderful but there are still major things that would need to be solved/implemented along with all your points.

For the water issue, they still haven't made it possible to remove water from a boat when it is sailing on the water (apart from the old invisiprim system which is not optimised), for people to swim without needing a hud (not new user friendly currently), create voids under water for buildings without windlight tricks etc.

For the mesh, they would just say use Animesh of which we know doesn't do what you are suggesting. Animesh should have included the system you state however instead of looking at all possibilities of needs or wants, they did one thing then released it (similar to when they updated mesh and then bento and didn't even think of updating their system avatar to their own detriment, further reducing user retention). The problem with doing this is it just further delays things and shows to the press etc when they talk of the metaverse, that they are just to slow and not forward thinkers.

I think linked object hierarchy would work and even know how it could be implemented UI wise however lets be a little real here. LL still haven't even added to their default viewer any way to cycle properly through individual linked objects. Firestorm has 2 little arrow icons to cycle through them when edit linked parts is on, not sure on other viewers. In addition to what you suggest they also need to fully update the edit menu to be far more user friendly.

Why don't they have another tab, next to the features tab in the edit/build menu that lists all the linked objects in the build, by name if they have one? This simple UI addition would be of HUGE benefit to creators where they could see the parts of the linked object by name and click on them to edit directly. The bonus of this is, it would also make your hierarchy request possible. All that would be needed would be for them to make it so that in that tab the root prim is at the top and in descending order wherever the object is placed in that list in that tab denotes their hierarchy. Then if a creator wants to change the hierarchy it would be a matter of dragging in that tab list one linked object up or down and dropping it where you want it in that hierarchy. The problem is, it has been a known fact for years that even if you want to add a feature request for a UI change LL will refuse it unless you literally show them how to do it and include UI design for them.

Link to post
Share on other sites

I just tested a viewer that apparently utilizes or put the load on the GPU more and it gave me about 2X performance increase, and if it's really because of the GPU being used more, apparently the reason why LL / most TPVs don't do it is because most people who use SL are not the gamer type who have strong GPU and not because it's not possible, I'm not sure though, I posted it in the machinima section.

Just thought it would be relevant here since it would slightly negate the, "Second Life's engine is inherently very unoptimized" bit and kinda show a better outlook in term of optimization or performance.

Edited by lucagrabacr
making stuff more clear
  • Like 1
Link to post
Share on other sites
1 hour ago, Drayke Newall said:

 LL still haven't even added to their default viewer any way to cycle properly through individual linked objects. Firestorm has 2 little arrow icons to cycle through them when edit linked parts is on, not sure on other viewers. In addition to what you suggest they also need to fully update the edit menu to be far more user friendly.

Hmmm, LL's viewer has had keyboard access to this for a long time.  Try CONTROL PERIOD and  CONTROL COMMA.  Add shift to select ranges.  Buttons on the edit tool panel would be more discoverable.  I have been using the key combinations described in 

and http://wiki.secondlife.com/wiki/Shortcut.  Just putting this here so people see it, not to refute anything you have said.

  • Like 1
  • Thanks 1
Link to post
Share on other sites
1 minute ago, Ardy Lay said:

Hmmm, LL's viewer has had keyboard access to this for a long time.  Try CONTROL PERIOD and  CONTROL COMMA.  Add shift to select ranges.  Buttons on the edit tool panel would be more discoverable.  I have been using the key combinations described in 

and http://wiki.secondlife.com/wiki/Shortcut.  Just putting this here so people see it, not to refute anything you have said.

Hmm didn't know that short cut thanks. Though as you say and I suggested buttons or the new tab would be far better and easier.

Link to post
Share on other sites
22 hours ago, Mollymews said:

with every new tech demo (engines, games and worlds) then I am still waiting for a multi-player demo with more than 100 characters independently outfitted and operated. Even the latest demo of Unreal 5 only has one character in it

Me too. I'm always looking for systems which can pull off dynamic region resizing with reasonable avatar/avatar interactions. That's SL's fundamental limitation - can't have crowds or the server chokes. SL can do maybe 150-200 for performances at a region corner of four regions, with careful seating and performer layout. Improbable's Spatial OS and Dual Universe claim to be able to do large numbers, but neither has a convincing demo where they're all close together. There are space shooters which do it, but space is easy compared to ground level with walls and rooms.

 

Link to post
Share on other sites
4 hours ago, Drayke Newall said:

 linked object hierarchy

we are probably thinking about this the same way, as a linked hierarchy in terms of a tree structure for positioning and rotational purposes. Example:

      A (root)
  B   C   D
 E F  G   
   H
  I J

BCD are linked to A (root). G to C. EF to B. H to F. IJ to H

rotate A then the whole object rotates as it does now

rotate C and G moves with it

B moves EFHIJ. H moves IJ

 

  • Like 1
Link to post
Share on other sites
On 6/4/2020 at 7:39 PM, Ardy Lay said:

After seeing somebody else do it, I made a map marker avatar.  It's a green cone pointing downward under a white sphere.  I think that would be less weird than a blob of jelly.  Apparently the render cost is quite low.

image.thumb.png.e16d7717b670628a92ccc0c119ee6655.png

FOOOOORRRRRRE! 🏑

  • Like 1
  • Haha 5
Link to post
Share on other sites
15 hours ago, lucagrabacr said:

I just tested a viewer that apparently utilizes or put the load on the GPU more and it gave me about 2X performance increase, and if it's really because of the GPU being used more, apparently the reason why LL / most TPVs don't do it is because most people who use SL are not the gamer type who have strong GPU and not because it's not possible, I'm not sure though, I posted it in the machinima section.

Just thought it would be relevant here since it would slightly negate the, "Second Life's engine is inherently very unoptimized" bit and kinda show a better outlook in term of optimization or performance.

which viewer was that?

Link to post
Share on other sites
5 hours ago, Selene Gregoire said:

FOOOOORRRRRRE! 🏑

Hahah!  I could probably use path cut to make a one-prim golf tee and might be able to improve the resemblance to a golf ball.  That would be funny.  Then I will need to see if sitting drives the tee into the ground.

It will probably still work as a low-impact avatar for when I want one and offer a little more amusement at the same time.

Edited by Ardy Lay
  • Like 2
Link to post
Share on other sites

i still like torturing prims myself

prim golf tee

Type: Torus
Size: X = 1.0, Y = 0.5, Z = 0.5
Rotation: X = 0.0, Y = 90.0, Z = 0.0
Path Cut: B = 0.0, E = 1.0
Hollow: 95.0
Skew: 0.0
Hollow Shape: Default
Twist: B = 0.0, E = 0.0
Hole Size: X = 1.0, Y = 0.45
Top Shear: X = 0.0, Y = 0.0
Profile Cut: B = 0.0, E = 0.3
Taper: X = 0.0, Y = 0.0
Radius: 0.0
Revolutions: 0.0

for people reading who want to learn more about what is possible with prims then Ayumi Cassini documents some methods here:

https://ayumicassini.blogspot.com/2009/07/ultimate-guide-to-prim-twisting.html

  • Like 1
Link to post
Share on other sites
On 6/5/2020 at 3:16 PM, Drayke Newall said:

Most people that are interested in virtual worlds (or really VR) are gamers and therefore they don't buy stock comps.

How do we know that? I'm not saying you're wrong but you are making a baseless assumption there. Nobody has ever made a bona fide attempt to create a virtual world for "regular" grown ups yet and that means it's impossible to tell if there can  be a market for something like that.

Besides, gamers may have expensive top notch hardware but that doesn't necessarily make them computer savvy. Most gamers I know would struggle if they were asked to do some basic text formatting in a word document. Give them anything beyond "Se Big Red Button - Push Big Red Button" and they're out of their comfort zone. How are people like that supposed to handle the convoluted trainwreck the SL viewers have as a user interface substitute?

  • Like 1
Link to post
Share on other sites
31 minutes ago, ChinRey said:

 

Besides, gamers may have expensive top notch hardware but that doesn't necessarily make them computer savvy. Most gamers I know would struggle if they were asked to do some basic text formatting in a word document. Give them anything beyond "Se Big Red Button - Push Big Red Button" and they're out of their comfort zone. How are people like that supposed to handle the convoluted trainwreck the SL viewers have as a user interface substitute?

That is a bit of generalization of gamers. A lot of gamers are tech savy, it fact most true hardcore gamers built their own systems and know the ins and outs of how it works and how to repair it if needed. Most do know a lot about the os and how to optimize it for their gaming experience so would have little problem learning how to use sl at all.

Dont lump in normal console gamers with pc gamers, there is a big difference between the two groups.

As someone who is both a pc gamer and a console gamer I find this a bit insulting.

Edited by Drakonadrgora Darkfold
  • Haha 1
Link to post
Share on other sites
1 hour ago, ChinRey said:

How do we know that? I'm not saying you're wrong but you are making a baseless assumption there. Nobody has ever made a bona fide attempt to create a virtual world for "regular" grown ups yet and that means it's impossible to tell if there can  be a market for something like that.

Besides, gamers may have expensive top notch hardware but that doesn't necessarily make them computer savvy. Most gamers I know would struggle if they were asked to do some basic text formatting in a word document. Give them anything beyond "Se Big Red Button - Push Big Red Button" and they're out of their comfort zone. How are people like that supposed to handle the convoluted trainwreck the SL viewers have as a user interface substitute?

Well, there is a market of at least one. Make that two. Neither of us would be considered a gamer, we both build our own PCs and buy stock ones, we both use SL and before SL there was AW, where we met over 20 years ago.

There is a market, albeit a small one. Possibly minute.

Link to post
Share on other sites
20 minutes ago, Kurshie Muromachi said:

In regards to some of the discussion on gamers, VR and stuff, the New Paradigm Strategy Group did survey on gamers. It surveyed over 3000 people from the US in Q4 2019. You can check out the stats at the following link.

https://research.npstrategygroup.com/gaming/

 

a lot of information but a really small playerbase so not really a good representation of gamers in general. 3000 out of millions = not very accurate statistics.

plus you cant really trust most studies for they are biased in what they are searching about to begin with.

Link to post
Share on other sites
11 minutes ago, Drakonadrgora Darkfold said:

a lot of information but a really small playerbase so not really a good representation of gamers in general. 3000 out of millions = not very accurate statistics.

plus you cant really trust most studies for they are biased in what they are searching about to begin with.

Still, just a small bit of data to think about even if it is a small group.

Edited by Kurshie Muromachi
  • Like 1
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×
×
  • Create New...