Kitely Starter Avatars world

[Natasha roy]

hi,
I recently discovered the Kitely Starter Avatars universe, and I must say that the range and quality of avatars on offer are remarkable! It's an excellent resource for anyone wishing to swiftly customize their online persona. Strongly recommended.Fm whatsapp

[Michael Timeless]

If you haven't considered Avatar Complexity before, here is some information. Sim owners who don't want visitors to crash are interested in this, especially educators who have students with older or less powerful computers.

This is probably the most comprehensive writeup on the subject I've seen. Not just noobs but some of us who've been around a while can really benefit from reading - and rereading this. Thank you, so much, Ada.

[Tess Juel]

This is probably the most comprehensive writeup on the subject I've seen.

It is but avatar complexity as it is calculated today is a meaningless number. Without going into details ther are at least six known factors that skew the way ARC is calculated and that means the "jellydoll" function is very unlikely to catch the avatars that actually cause the most lag.

Beq Janus did some tests when she was working on Firestorm's new performance measuring tool and she posted the results here: http://beqsother.blogspot.com/2022/03/h ... mance.html

The seventh illustration on the page is perhaps the most telling. I hope Beq doesn't mind that I repost it here:

image.png (74.65 KiB) Viewed 155914 times

As you see, the avatar with the lowest ARC during her test was the one that actually was most costly to render. The most telling comparasion is perhaps between the first and 14th avatar on the list: A third of the ARC = three times as high actual render cost!

It's bad enough the "jellydoll" function might as well pick random avatars to derender as to rely on the calculated AC.

At least that's how it i with LL's official viewer code. It's possible Beq was able to change it to use the actual render cost rather than that bogus ARC. I doubt it though, it doesn't sound like something LL would allow. I'll ask her but it may take a while to get an answer. I haven't seen her for a while so she may be on vacation.

[Ada Radius]

That said, the easiest way I can increase my frame rate is to go into Preferences/Graphics and set Complexity to around 100K.
It's a highly flawed number, yes, but we don't have a better one. If someone comes up with a better one, please please write it and hand it to Beq. Please.

[Tess Juel]

It's a highly flawed number, yes, but we don't have a better one. If someone comes up with a better one, please please write it and hand it to Beq. Please.

I got a reply from Beq and I'm afraid it's not a good one. Firestorm has to comply with standards set by Linden Lab for Second Life and they're not going to fix their mistakes here.

LL does have a bad but still understandable reason for not doing anything. Most of the changes that need to be done to fix the issues would effectively break most existing fitted mesh and all fitted mesh creators would also have a hard time learning almost from scratch how to make render efficient meshes. That would be a disaster for Second Life since its economy is so dependent on sales of fitted mesh by now.

That said, the easiest way I can increase my frame rate is to go into Preferences/Graphics and set Complexity to around 100K.

Derendering a few random avatars will always improve the frame rate of course but I think it's still important to know that the avatars the "jellydoll" function targets are effectively chosen at random and not in any way based on their actual render cost.

[Ada Radius]

Derendering a few random avatars will always improve the frame rate of course but I think it's still important to know that the avatars the "jellydoll" function targets are effectively chosen at random and not in any way based on their actual render cost.

Yup, LL suks on this, and Beq is certainly tired of hearing about it. The algorithm needs work duh. And so do the testing displays inworld - what we get from running them isn't reliable either. Kitely has the website displays, and that IS helpful. I know from working with avatar Character data that LL started out with great ideas in 2003 and got sloppy after that.

And it's worse than that - in a club, for example, where we tend to see more overloaded avatars, derendering the high AC avatars (whether or not that AC number is any good) significantly increases my frame rate. But an avatar with bad attachments, even jellied, can crash my viewer and the viewers around me, or make moving around impossible. The last time this happened, was within the last week, at a Blender class. Someone arrived with an AC over 800K, lagging everyone in the sim. When asked to fix it, she set off a giggle gesture and added enough crap that her AC number was nearly a million. I crashed and decided that a nap was a better idea.

Because I'm an avatar dev and maker, I keep the AC numbers displayed and run the tests - I want to see the patterns, maybe figure out WTF, as we certainly are not getting accurate information from LL. If someone is jellied and I know he or she has no idea what they've done, I might render them individually and see if I can figure it out and help, if the person wants help.

I think it's up to those of us with even a little knowledge to teach and try to change behavior in both designers and users. Doesn't matter in SL, no reward for making or buying better avatars and accessories there. In Opensim, grid and sim owners are definitely motivated to encourage better behavior.

[Michael Timeless]

"...The last time this happened, was within the last week, at a Blender class. Someone arrived with an AC over 800K, lagging everyone in the sim. When asked to fix it, she set off a giggle gesture and added enough crap that her AC number was nearly a million. I crashed and decided that a nap was a better idea."

That's one advantage I have over you. That type of behavior on my grid would earn an immediate ban. To me that is simply a griefer and no different from the Priscilla Kleenex's of the world, especially in a classroom setting. You are correct that users need to understand how their complexity damages the experience for everyone, but to me rude behavior is just that, rude behavior. I'm lucky in that regard, most of my users are people I've met personally with just a few exceptions. There are times when I really feel for Ilan and the things you have to put up with here and on SL.

Take care.

[Snoots Dwagon]

Tess Juel wrote: "It is but avatar complexity as it is calculated today is a meaningless number. Without going into details there are at least six known factors that skew the way ARC is calculated and that means the "jellydoll" function is very unlikely to catch the avatars that actually cause the most lag."

You're right; we noticed this at the very beginning of the ARC con game (pardon my bluntness), and have tracked it through the years. But as Ada says, there's not much we can do about it.

I strongly suspected when Linden Lab first came out with ARC that it was primarily an attempt to shift the source-blame of lag from the system to the users (as they had often done before. That was just their way at the time.)

But we ran a test that by now many people know of: We ran an ARC test on a dwagon avatar (total prims head to toe, 451 at the time), a Warhammer 40k mega-avatar of 1,500+ prims, and a standard female avatar wearing a simple flexi dress, flexi hair, shoes, no jewelry.

The Dwagon clocked in lowest at about 1000 ARC (if I remember correctly after all these years. ARC was measured differently in those days). The WH40k hyper-armor avatar came in at about 2500. The highest ARC of all was the standard female avatar, registering at over 6,500 ARC!!!

We knew at that time the ARC system was messed up, and while it has improved somewhat since then, it is still flawed as your diagram shows. Does anyone want to tell a female avatar she can't wear a nice dress and hair because her ARC is too high? Get real. (Same goes for males, though they generally tend to not be as fancy as females.)

So yeah, I've never been a fan of the whole ARC scam. The concept is valid: Let's keep avatar complexity reasonable. But I'm especially not a fan since they've imposed silly limitations these days. (You're wearing too many prim items or too many animesh pets, for example). They're not measuring the IMPACT of those items; they're just counting pieces. It's getting to the point one almost HAS to wear a mesh avatar to keep under their prim limit. So I've moved on to other things.

(We did once run extensive dual-verified tests on both animesh and NPCs... and discovered they create close to zero lag; they're basically an animated prim.)

I know Kitely is a lot smarter about a LOT of things than the standard Opensim release... and even Second Life itself. But due to lag issues caused by all sorts of SL bottlenecks... millions of people have checked out SL and left. At least on Opensim we're not paying $250 a month for the privilege of lag. Admittedly avatars do lag. In truth they are probably the laggiest thing on the grids. But I'm not sure the ARC system as it now stands is the solution to that issue. Telling users they can't wear their animesh pets just alienates users.

All that is why I now build guitars and flutes in real life and watch Netflix for entertainment. After 20 years I just got tired of the mess. Much of that has been fixed on Kitely, due to Ilan's and Oren's hard work... which is why I live on Kitely instead of elsewhere ;D

[alexia gomes]

"The universe of Kitely Starter Avatars has an amazing selection of avatars, all of them are bursting with personality. It is a must-visit location for anybody wishing to improve their virtual identity because of the attention to detail and variety of alternatives. Bravo Kitely for creating such a captivating experience!"top follow

[Vixy Sharpfang]

"The universe of Kitely Starter Avatars has an amazing selection of avatars, all of them are bursting with personality. It is a must-visit location for anybody wishing to improve their virtual identity because of the attention to detail and variety of alternatives. Bravo Kitely for creating such a captivating experience!"top follow

By the way: Hello, I'm Vixy, I'm new here, aaaand I'm a member of a group making new avatars for OpenSim from the ground up.
We use a simple approach, like, on Second Life, known from manufacturers such as Luskwood, DSD, Kinzart, Aventity, ... :
Add-On Components on a human system default body.
That way one can also use any mesh body one wishes to use and thus can use that mesh bodies clothing.

So far we have only one Avatar model ready: An anthropomorphised Shark, male and female, several different species.
Currently we are assembling the v2.0 packages. We distribute them as full-perm freebies.
We could put them into the Kitely Market - As of writing this I haven't understood how to upload items to there - but we'd also be happy to donate them - all or in part - to the Starter Avatar location.

Whom would I have to talk to for that?

Regards
Vixy