limit avatars to 1000 prims

OK, an option then: a system similar to land tiers. By default, an AV can equip up to 256 prims (or similar amount) ... past this point, there's fixed fees, each fee buying you right to wear 128 more prims for a week, or 30 days or something similar.

One can then be still as selfish as they want... but at their own, actual expense. Difference from telling others to shell out their own money to 'buy a better computer' so they can put up with one's selfishness, if you will.

I think we should limit some shapes rather then prims... a torus for example is lot more hard on the system then a box

As proved by the guy who attached 7600 of them to his av all at once and went from 45 fps to 2 fps and promptly crashed his system.

Have you tried turning down avatar detail? This is something that LL actually lets you control.
Prims in homes exist on the sim, the sim has to keep track of them, all the time. Prims on an avatar only exist in the client application.

I think my definition of reason is pretty valid. Others here seem to agree. (or disagree lol). Other than "I want prims!" what valid reasons are there for allowing unlimited prims? We've presented all sorts of valid reasons for limiting such.
Making murder against the law also takes away options. And it's absolutely necessary for a safe society. "I don't want to" has never been a valid reason for lack of regulation. Someone shouting "I don't want avatar prim limits!" doesn't cancel the validity of the concept. Because frankly, "I don't want lag!"

Refusing to compromise and establish reasonable limits seems a bit one-sided. A reasonable solution would be to establish avatar limits at an amount that would encompass the majority of the avatars on the system. From what I've seen, 256 would be a reasonable number.

But I am not proposing that. Why? Simple: because even 256 prims would lag under current conditions... espcially if they were tortured prims. Applying an av prim limit at this time would be an exercise in futility. But if they set it at say, 500 or so... at least that would stop the griefers and clowns. If someone is wearing more than 500 prim, I frankly don't even want 'em near my land... because I know they are going to lag like murder.

Good point.

That's exactly why av prim limits makes sense. Because not only did he lag and crash himself... what about anyone standing within visual distance of him?

I saw one guy walking around wearing 200 prim wings refuse to remove them when asked to do so by people he was lagging to death. I saw another girl who was just absolutely going to wear her hoochie hair, she didn't care what it did to everyone else. Another guy came to a packed event wearing 500-prim armor.

These people are why regulations exist and ard needed, because they're self-centered [expletive deleted]'s who are probably the same way in RL. They don't have the grace to moderate themselves, so they can only be moderated with enforced regulations. Unfortunately, on the internet, such people aren't all that rare.

But again, limiting avatar prims without taking care of all the other deep-core lag factors would be a futile effort. So I wouldn't really be willing to advocate av-prim-limits unless the other things were taken care of first.

Have you tried turning down avatar detail? Do you like watching bald girls walking around?

I have a new, duo-core computer with a gig of ram and a high level graphics card. I paid for that. Why should I be forced to sacrifice graphics detail just because some self-centered user who has no regard for anyone else decides to parade around wearing 1000 unnecessary prims?

We've discussed this before in another thread: server side or client application... it's still Second Life software and it still causes end-user lag. If someone comes to an event wearing a 300-prim avatar... he's going to lag everyone around him. Whether that is server side or client side is really irrelevant to the people who are being lagged to syrup.

Quoted merely to reference who I'm replying to, I'm feeling a tad lazy to snip the non-quoted material.


Avatar prim limits should exist, but they need to be of a state that actually reduces lag while still allowing from high-non-lag prim counts. The idea of a prim-value (cube is 1, totured toruse is 20-bazillion or what have you) is something I can agree with.
The av that I use on occation that causes some lag, but not as much as hoochie hair--Daryth's dragons--has very few toruses, the avs are almost entirely spheres--which while more laggy than a cube, is not unreasonable.
If a prim-count limit of any method went into effect that I could agree with and still killed Daryth's dragons, I'm sure she would reduce them in prim complexity enough (fewer toruses*) to make it work, but saying that 200 prims is max would mean that the avatar is worthless--I don't think it can be done with half as many prims and still have the regal look that it does now.




*By estimation, the dragon I have may have as many toruses as hoochie hair, HOWEVER, that encompases all 14-odd attachments (tail is probably 3 times 2 positions, 'beard' has 5 (I don't like it anyway and if the face didn't look stupid without it I'd delete them), wings have 2 or 3 each times 2 positions and two for the neck, and probably a few I've forgotten) for the entire av and isn't "just hair" where the person would then have bling and 40 prim diamon rings, 14 prim dresses, shoes, and god knows what else too. When wearing it I don't wear anything else.

Yeah, that all prety much makes sense. I agree that limiting avs to 200 prims would be useless (and said so above) for 3 reasons:

1) It would kill out a few popular avs... like the dragons,
2) It wouldn't really reduce lag all that much... because even 200 prims can currently cause significant lag, and
3) Since there is no weighted-prim measure, even a 100-prim av using tortured prims could cause significant problems.

So while the concept itself is valid-- it would be useless to implement it unless the other issues causing severe lag are fixed first (such as asset servers, data handling, transfer protocols, graphics routine streamlining and simplification, etc etc). Since avatars wearing over 500 prims aren't all that common, such limitation would, with rare exception, likely be an exercise in futility.

Now, if they fix the core problems and get SL running well, then we can see if the avatar-prim lag issue is still a problem. If it is, then limiting avatar prims might be a valid last step. Because for sure.... avatars are the #1 client-created lag source on the planet.

Sorry, I refuse to agree to anything that forms a limit you can't circumvent.

If I understand this corrrectly... we should only set up rules that can be bypassed at the whim of any individual?

So... should we remove land prim restrictions as well?

Look, it's this simple:

Prims add to lag.
It is therefore necessary to restrict the number of prims allowed per area of land.
No one questions this necessity.
Using exactly the same logic... it is reasonable that avatars should also have prim limitations, so that they do not contribute excessively to lag.

No offense intended, and this is not directed at you personally, but if this seems unreasonable to anyone, I would have to question whether that person is protesting from a standpoint of sound logic--- or "I want" without regard to anything else.

The detail slider only seems to control the level of detail part render, i.e. you can control with it how soon the prims are being rendered using less faces. The catch is, the actual process of detail reduction that constantly goes on as your camera moves and turns around can be more processor-intensive than drawing full load of triangles that rarely need to be replaced with "low detail" version. Which in the end means you'll get less 'lag' with sliders cranked all way up, than with reduced details o.O

Hmmm... I don't know how the detail reduction process works, but I have been curious about something:

Every prim is undergoing a 20-times-per-second render/don't render decision. Whenever something is present and a decision is made to not render it... how long does it take to make that decision compared to just going ahead and rendering the item? In other words, does it take longer to make a render/nonrender decision for each item, then render it if the answer is YES.... or just to go ahead and render it in the first place?

I'm sure the answer is that the decision process is shorter than the actual rendering process... but one does wonder.

I would like to say that the most scriptiest avatar i would ever consider using has 2 scripts in it and therefore the entire world of second life must be absolutely limited to no more than 2 scripts per avatar because anyone who is not just like me is a blithering idiot.

Furthermore i believe the color red is annoying but only on tuesdays and therefore all prims attached to avatars on tuesdays must not be colored red because anyone who would wear red on a tuesday obviously does not count for anything.

Seriously, if you don't like prims on avatars, set your avatar detail slider down, it won't render them... if yer client is *STILL* not running smooth then its something *ELSE*... remember SL is exceptionally cpu/network bound so things like leaving a big inventory open, or color changing updating objects can actually cause significantly more work per frame to render.

Asking for a global limit to break tens to hundreds of thousands of OTHER people's SL experience because you think you know what you are talking about is the worst kind of self interested biggotry.

If you really are annoyed, ask for *BETTER CLIENT CONTROLS* not for hard line world limits.

While computer hardware and technology constantly evolves, hard coded system limits rarely do.

What do you set as a system 'target'? i can get decent framerates even with 20+ avatars on screen with my system, and its now almost 2 years old. (athlon 64 3500+, first gen) Its not as slick as a high end fps, but then SL is not twitch/reflex game, so it doesn't need to be.

So according to my own experience, the current setup seems fine, where is the limitation need?

is it for people running p4's at/below 2.4ghz? Is it for people running older athlons? god forbid p3's?

will the population of people running p4's at/below 2.4ghz as a percent of the total SL population remain static over the next 2 years? if not... why would any proposed hard coded limitation?

what about alpha? an alpha polygon costs > 10x more than a non alpha polygon to render, especially on older hardware... and not all prims are polygonally similar... someone using 10 transparent tortured torii will render more polygons and put more workload on a system than 1000 opaque cubes.

The simple answer is to adjust your own settings more properly to your own system, not to try and adjust everyone *ELSE* in sl to make you feel better. We have almost all of the controls we need for this already... you (the general you for everyone always pushing for hard-line limits in SL) just may have to accept that you can't 'will' someone else to do what makes *YOU* feel good. You don't have a right to tell your neighbor what to build, what you *DO* have a right to do tho is ask for better CONTROLS so that sl can be a better experience, for everyone. Each of us can enjoy the SL we wish that way.

The graphics data is typically broken down/sorted into some sort of spatial structure (octree/quadtree/ABtree/whathaveyou) ... and the culling is then performed by doing visibility tests of as large chunks as possible. Say, your camera is facing north, that means you can discard roughly half the data that's south of you with just 2-4 checks if the bounding box/sphere fits camera frustum. The tests are very fast themselves, just a few multiplications with the spheres e.g., and that's something computers are quite good at -.o

I think SL client has a switch in debug menu to draw its octree structure, so you can see how it's all arranged.

Yeah. Not what I mean however. It stands to reason that anything behind you or out of camera range will be automatically culled. What I was referring to were items within visible range. For example: You have 2000 prims directly in front of you. Would it take more time to simply render those prims... or to examine each individual one and calculate whether it can or cannot be seen by the avatar? If it can, render it. If not, move on to the next prim. That's a decision process that rather than just being yes/no, involves line-of-sight graphics calculations. Are those faster... or slower... than simply rendering the prim to start with?

I've always wondered.

i doesn't change the problem that is you where in RL presenting to your game studio lead artist models like we have in sl (avatar+ attachments) you would be promptly kicked out of the studio

It doesn't do that either heh, hence why you want to look at the octree as Joanah Cramer suggested
In occlusion culling it isolates cubes that can't be seen and removes all primitive safely within them. Individual tests would just be horrible and probably wouldn't be worth it as you seem to be getting it/wondering about. If the majority of prims are visible, and you check ALL of them, then you would be wasting a ton of comparisons for no gain

Got it. Makes sense. Thanks!

Compared to girls wearing chichi hair that causes more lag in 10 toruses than a nice set of wings cause with 200 spheres?

I understand the problem. The thing is, your problem basically comes down to the inability to limit the *kinds* of detail that get reduced when you turn down Avatar detail.



It doesn't matter to the problem.

It does matter to the solution.

Since the lag is only client-side, then a client-side limit will solve the problem. Without causing problems for people who want to see all the chichi hair and gool dragin wings in infinite detail...

One thing they had pre 1.10 was an option to limit draw distance based on FPS, so you could set your draw distance high and have it drop down when your FPS dropped. I really wish they'd bring that back and apply it to other limits.

At this point you've already oversimplified things too much for the rest of your argument to be meaningful. Because there's more than one kind of lag and, yes, it does matter.

It's not an either/or Argent. Both are issues. I'm not saying the guy with wings is wrong in wearing them; in fact, my statement to this point is that a good limit is 256 prims. But... when either of them refused to de-prim at the request of others (ie, insisting on their own wishes rather than the general welfare)... then both were in the wrong. And that's why limits need to be applied.. because they're not the only people on the net with such attitudes. Face it: if people were self-controlled and self-moderating, we would need no laws or regulations-- either on SL or RL.


Argent, when people running heavy-duty machines experience the same problems as people with lighter machines... then is the problem "only client-side" or is it a matter of LL policy and system design?

On the one hand, you're saying that we should not restrict people from wearing as many durn prims as they want to, no matter how much it lags everyone around them. You're standing up for their theoretical (and debatable) "rights"... but what about the consequenses of their actions upon others around them?

So while standing up for those "rights"... at the same time, what's being deinied is the basic right of everyone to have as good a graphics detail as they can get. By suggesting that they have to choose between either less lag or lousy graphics... you're removing their rights to an enjoyable SL experience-- because some people are inconsiderate enough to insist on wearing 2000 prims when they know such is detrimental to the environment.

That's the issue. When someone has a fast system and fast graphics card and still are lagged by excessive avatar prims... then that is obviously a LL issue. To try and put the burden of blame on the customer is an old and oft-repeated mistake.

Whatever works. Because in truth, for what we're paying to use this system... we deserve good graphics and low lag.

Someone once mentioned that they should grant a little more to paying customers. Grant freebie users one level of performance, paying users a higher level of performance. That seems like a viable solution, and reasonable (ie, you get what you pay for). Now, a lot of people might scream at that idea as well, but hey... someone is always going to scream no matter what's done. Someone has to be willing to make the hard calls.

There is nothing wrong with the basic concept of reasonably limiting the number of prims an avatar can wear at one time. And people crying "But I want! I want!" really isn't addressing the issue... that of grid-wide serious lag that affects Second Life every day.

Have you ever heard of Occam's Razor?

I did not say that prims are the only cause of lag. Excessive prims cause significant lag. That is why there are land limits. And I've already pointed out-- three times I think, that limiting avatar prims will do little good unless they get rid of the other lag-causing problems as well.

That's pretty simple, all right.



Now... if you want to get more complex...

It has been pointed out by many that simply limiting prims on an avatar will not work. A person can wear 200 simple cube prims and lag far less than someone wearing 50 tortured torus. So the suggestion has often been made that rather than limiting the number of prims only... set an avatar "weight" based on the type and complexity of prims involved. That's the only true measure (to my knowledge) of tendency of an avatar to lag. But then, we get into other areas such as scripting. Are they wearing non-scripted items... or the old super-laggy AO devices that carried about 20 main-channel listeners?

There's a lot involved. But to do nothing because one can't do everything won't solve the problems. We deal with the things we can now... and worry about dealing with the rest later.

That's about all I'm going to say on this issue... 'cause I think the point's been made, by me and several others. Field seems pretty divided. Been a very good discussion. Your turn.

A single sphere gets broken down to ~600 triangles for rendering.
Single piece of torus as typically used in hair strands, ~700 triangles on average.

So in your example that's 7 k faces to render the hair, and 120 k to render 'nice set of wings'... really, no point in demonizing the tori or hair in general, it's not as clear-cut as public opinion would lead one to believe -.o;