Rendering cost - Mesh VS Prims/sculpty - who is better?

[MaxTux Wonder]

As LL changed the render cost algorithm over to the render weight algorithm, which is scaled differently than the previous version. We can now compare the rendering cost of Mesh VS Prims/sculpty - who is better?

https://jira.secondlife.com/browse/SH-2584

 

To check the rendering cost of your avie go to:

Menu advanced -> performance tools -> show draw weights for avatars 

 

My mesh avatar (the tiger) and a standard avie

 

My avatar the tiger with full armor and a full clothed furry avie

 

other prims/sculpy avies

 

My wolf demigod avie in full armor VS other prims/sculpy avies

 

My mesh bull avie VS other prims/sculpy avies

 

why prims/sculpties are more heavy than meshes?

 

for a simple reason, look at those pictures

this is a prim cube:

this is the same prim cube but in wireframe:

as you can see a prim cube has at least 108 faces, "at least" becouse if you modify a prim cube and add an hole you get more faces:

so actually in sl all prims building and avatar in SL have a surplus of not needed faces, for example a mesh cube has only 12 faces (triangles).

Even if a mesh avie will have lots of faces will be anyway better than a prim/sculpty avatar, becouse the mesh topology is optimized.

Something similar happens to sculpties, at maximum resolution, a 32x32 pixel Sculpt Map would describe up to 1024 vertices:

so it's very easy to get a very big render cost on your avie (with prims and sculpties), what it needs is just a fancy couple of boots with many sculpts on it!

 

[Nalates Urriah]

That is a fun comparison. Thanks.

[Ceera Murakami]

Clear message from Linden Lab - They apparently want to see the grid as a place for simple Humans only, and Furries and other non-Human avatars are strongly discouraged.

They designed this system to yield absudrly high numbers, and to red-flag virtually any non-human avatar. What a wonderfully intolerant and creativity-crushing tool.

[Baloo Uriza]

How do you figure?  Mesh furries, as demonstrated above, ranked lower than the humans did.

[Ceera Murakami]

A naked mesh furry was lower, but all the rest, including that same furry with any accessories at all, were higher, and red flagged.

If people start taking that tool seriously, it will result in most currently existing furries and non-humans being designated as "bad", not because they really are, but because they are red-flagging at an arbitrary numeric value in an arbitrary system that most furries won't ever score low enough in to be "acceptable".

Heck, even that Human was already flagged amber, and not green. What do you have to be to get a green rating? The default Ruth avatar?

[Baloo Uriza]

---

Ceera Murakami wrote:

A

naked

mesh furry was lower, but all the rest, including that same furry with any accessories at all, were higher, and red flagged.

If people start taking that tool seriously, it will result in most currently existing furries and non-humans being designated as "bad", not because they really are, 

---

Keep in mind that not everyone has a state-of-the-art GPU, and in crowded locations, it doesn't matter what kind of GPU you have if nobody cares.  That said, most avatars out there really are that bad and could probably go through some simplification without significant change in quality but a substantial reduction in rendering cost, both by the metric and by actual performance gained in high-GPU-load areas.

[Charlar Linden]

Baloo - Exactly!

The number is simply a reflection of the relative difficulty of rendering the avatar. If you have a good card, those high numbers are much easier, if you have a 'less' good card, there is more likely going to be problems; usually showing as lower than expected FPS, feelings of lag and or sadness.

Inefficient use of sculpties, high levels of detail for no reason (prim shoelaces are a classic example), and non-decimated meshes can all contribute to unhappiness among your fellow avatars. 

The number imposes no hard limit, it's merely a way to help residents better understand why things aren't running the way they expect.

For example, a Linden found a coat with fur trim (prim/sculpty/flexi) that tipped the scales at 320,000 points, all by itself. This coat can choke everyone else in the room, but it could be made far less inefficient and we hope that this information will help creators do just that. 

 

Charlar

[Peggy Paperdoll]

Oh yeah, that conspiracy thing again.  LL wants to limit and confine us for some, always undefined, reason.  Second Life was an experiment created  by Philip Rosedale in 2001.  The avatar he chose and the prim building blocks he chose were popular and quite common for the day......they also were not so complex that an average user with average creative abilities could not learn to use them and create with them.  And LL has built on those basic building blocks and avatars.......quite successfully, really.  It's a little late in the game to change now........that would require some substantial server code rewriting, not to mention destroying those billions of textures and millions of objects currently in SL.  Sure LL could have made the avatar more complex with more polygons and the prims more flexible with fewer polygons.  Would it have worked?  Probably.  But SL was an experiment.........something never done before by anyone.  Content creators flocked to SL like ducks to a pond.  Why would they change if the experiment was working?  If LL had changed the avatar and prims early on then maybe it would only have been a minor set back with a relatively few textures and objects being destroyed with the change..........but after a couple years it was too late.

 

Had LL started with a more complex platform and insisted on creators optimizing the content before they were allowed to upload to SL then SL would certainly be less laggy and avatar weight would not matter as much.  Blue Moon tried that and look were they are today......and some very smart people developed Blue Moon.  Had LL started out with something different SL probably wouldn't be here today.

 

It's not a conspiracy.  It's the way SL was built.  You can't change it now without destroying what it is in the process.  I doubt SL is ever going to evolve into that wonderful world of the perfect (or even better) avatar or the basic prims in the build tools ever getting better either.  It just can't happen without destroying the entire virtual world.  Something entirely new will have to be built from scratch.....maybe even built by LL (who knows).  But it was fun reading your conspiracy theory anyway.

[DanielRavenNest Noe]

---

Peggy Paperdoll wrote:

Had LL started with a more complex platform and insisted on creators optimizing the content before they were allowed to upload to SL then SL would certainly be less laggy and avatar weight would not matter as much.  Blue Moon tried that and look were they are today......and some very smart people developed Blue Moon.  Had LL started out with something different SL probably wouldn't be here today.

-------------------------------------------------------------

Umm, Peggy?  The name is "

Blue Mars

".  Blue Moon was a detective agency in the "Moonlighting" TV series, and a number of other things, but not a virtual world.  As far as the optimizing content issue, we were limited to 512KB compressed data per clothing item, and 5000 triangles.  That is actually not restrictive.  I was able to do some very nice items within those limits.  In a few cases I needed to upload it in two parts, but usually it could be done as one.  By comparison, the base SL avatar is around 7000 triangles.

 

 

[Peggy Paperdoll]

Blue Moon is also a small (but expensive) restaurant chain here where I live.  Thanks for correcting me........I did mean Blue Mars. 

[Medhue Simoni]

You can be a furry, you just need to be a mesh furry.

[Alisha Matova]

I've always called it Blue Moon too. Freudian slip, I suppose. =P

(as I logged in there only once in a blue moon)

[Jennifer Boyle]

---

Peggy Paperdoll wrote:

... If LL had changed the avatar and prims early on then maybe it would only have been a minor set back with a relatively few textures and objects being destroyed with the change..........but after a couple years it was too late....

---

Can you help me understand why it would not be possible to have two, or more, classes of avatars that are constructed according to different sets of rules?  If that were done, newer, and better, ways of making avatars could be introduced without breaking anything, couldn't they?  It seems to me that server and viewer code could be written to recognize more than one type and handle each appropriately, but I am not knowledgable about such things.

[Baloo Uriza]

---

Jennifer Boyle wrote:

If that were done, newer, and better, ways of making avatars could be introduced without breaking anything, couldn't they?  It seems to me that server and viewer code could be written to recognize more than one type and handle each appropriately, but I am not knowledgable about such things.

---

 

I'm just happy to see that it degrades as well as it does on a viewer that doesn't support it yet (it could do better, but hey, at least it doesn't die horribly).  Sometimes, it's just a matter of what's being rendered hadn't been done that way in previous viewers, because they're introducing support for a new method while supporting older methods (as opposed to breaking everything by immediately cutting off older methods).

[Siddean Munro]

There is no human v furry bias to be found here. I have seen human avvies with much higher rendering cost than these furry avvies, thanks to their 255 torus blingy bumpy jewellery, flexy prim alpha textured skirts that use all 1024 textures, and their masses of twisted torii and flexy and sculpty hair. It's efficiency vs non efficiency where the bias lies, not in what flavour of avatar you prefer.

[Peggy Paperdoll]

Oh, I'm sure the code could be written........a lot of it too.  But then comes the viewer........it has to be re-written to see both (or all) the sets of avatars.  What about the resources needed from both the servers and the users?  How many megs or gigs would the viewer be?  Isn't it about 25 megs already.........50 megs is a huge download (getting like that Blue Mars viewer download ).  And all the confusion it would cause........."Why can't I see my friend's avatar?"  Complicated programs run customers off faster than buggy programs.  Jamming code together like that just won't work..........something's likely to break.  Like the entire grid.

[DanielRavenNest Noe]

The SL character folder is 3.5 MB zipped, which is how it is delivered.  In case you have not noticed, when you install Second Life, the program folder grows to around 100 MB on your hard drive when it's unpacked.  A new and better avatar set added to the existing one might add twice that to the download, say 7 MB.

If you think 50 MB is a huge download, then you are still living in the 20th century.  Netflix standard definition movies run 170 MB per hour, and my AppData/Local/SecondLife folder (where textures are cached) is 300 MB.  Just idling in SL I use 9 MB/hour.  If I am actually doing anything it's a lot more as textures download etc.

Some comparison 3D program sizes:

Autodesk XSI Mod Tool - 607 MB

DAZ Studio 4 - 446 MB

Vue R10 (digital nature software) - 2.59 GB

CryEngine 3.3.7 SDK - 754 MB

Autodesk 3ds Max - 3.5 GB

Crysis 2 Game (as installed) - 10.7 GB

 

 

 

[Charlar Linden]

Exactly Siddean.

It's about efficiency, and in the case of avatars, it's all about giving residents the knowledge to understand why performance is different in different circumstances. You might have a super-powerful rig, but the other people in the room may not and when their FPS suddenly slows to a crawl, it's helpful to everyone that they know why. We don't want to limit anything because there are plenty of circumstances where a high cost avatar is desired (photography, machinima, groups where everyone has good rigs).

Rigged mesh seems to me like a really great step forward for furries, as much as for any other avatar type. We're excited about improving support for all types of avatars; human, furry, mech, tiny, etc.

 

To add context to the discussion about adding a second class of avatar, one that takes advantage of the advances in the tech: Yes, we could, but doing so is similar in many respects to our addition of mesh support. There's a bunch of code to add it at all, but then there's also a host of ancillary issues (do old anims work on new avs; can new avs sit on old chairs? What about the reverse? Clothing support? etc. etc.), and social issues (e.g. if features are made for new avs that aren't backward compatible, does that mean old avs become second class citizens?). Solving each one adds more work, or creates compromises. That's a long winded way of saying "it would require significant resources and work to do correctly", and we have to balance any undertaking against other priorities, which is something we're constantly doing. The forums are a great place for us to get a feel for overall resident interest, so please keep talking (not like I have to encourage that with you all).

 

Charlar

 

[leliel Mirihi]

---

Charlar Linden wrote:

[...] but then there's also a host of ancillary issues (do old anims work on new avs; can new avs sit on old chairs? What about the reverse? Clothing support? etc. etc.), and social issues (e.g. if features are made for new avs that aren't backward compatible, does that mean old avs become second class citizens?).

---

 

I think there are a few some what easy answers to those questions. Current animations are tied to the current skeleton, making them work with a new skeleton would limit what that new skeleton could be. I think it's best to just say no to backwards compatibility and give us custom skeletons. As for old chairs, well current avatars have problems sitting on them too . Making old clothes compatible with a new avatar is doable, but once again would limit what that avatar could be. And mapping the new mesh onto the old UV map could end up looking worse than it does on the current mesh. I think it's best to say no to backwards compatibility again.

All this is leading up to my grand proposal of make the avatar a wearable asset like everything else, the old avatar would be the default. As for the question of second class citizens I think you should let the users figure that one out.

---

Solving each one adds more work, or creates compromises.

---

Personally I'm a little tired of compromises. There is way too much of sl that still looks like it's from the 90s and I think that needs to change sooner rather than later.

[Baloo Uriza]

---

leliel Mirihi wrote:

 I think it's best to just say no to backwards compatibility and give us custom skeletons.

 

---

 

You know there's a whole orphanage full of babies in that bathwater you're trying to throw out, right?

[leliel Mirihi]

---

Baloo Uriza wrote:

---

leliel Mirihi wrote:

 I think it's best to just say no to backwards compatibility and give us custom skeletons.

 

---

 

You know there's a whole orphanage full of babies in that bathwater you're trying to throw out, right?

---

There's an easy solution that kills two birds with one stone. Make avatars a wearable asset so we can have custom whatever avatars, and also make an official avatar 2.0 that is improved but also backwards compatible. A lot of people want quadruped avatars but the current skeleton really doesn't work with them, without custom skeletons it's never going to happen.