Above -- human avatar with Maitreya body, polycount total circled
My report earlier this month that some of the top-selling mesh bodies in Second Life -- such as Maitreya, Belleza and SLINK -- are extremely resource heavy, and degrading SL performance for everyone around them, has caused a lot of conversation. Many readers claim the polycounts being reported can't possibly be that much. So here's an easy way to check for yourself:
Download the acclaimed, graphics-optimized Black Dragon SL viewer, then select Dragon > My Useful Features > Performance Tools > Advanced Complexity Window. As you can see above, the Maitreya body here uses well over 100,000 polygons.
How much is that? By contrast, a character in Red Dead Redemption 2, the latest AAA hit, has polycounts estimated in the 40,000-60,000 polygon range. (Here's a handy site showing character polycounts for other top AAA games with high-end graphics.)
You want to see something insane? Here's a Belleza body with over 500,000 polygons:
Again, these are bodies that require graphics processing that even top AAA games on PS4 or another major console don't require. And they are worn by a majority of regular Second Life users. Many of whom use low-end laptops. Or older desktop PCs. Or worse.
Some readers have only been counting KTris to estimate graphics complexity, but Black Dragon creator NiranV tells me that's a bad metric:
"The KTris display is a bad way of figuring out how much polygons something (especially a single object type) has. It is highly inaccurate and prone to randomness as the rendered amount of triangles can change at any time for any reason; do not use them please. KTris is a debug display that shows you how many triangles are currently rendered in the entire scene. You can imagine that even in a very static scene, this value can jump around a lot randomly since it's the amount for the entire scene, not just a single object. You can try to prevent that by going into the sky and only having the body visible and comparing, but it's extremely inaccurate and random.
"Also triangles and polygons are the same. Polygons are triangles, but polygons can be any shape: quadratic, hexagons etc. In SL, polygons only consist of triangles, it's the only shape SL supports (it converts quads to triangles on import)."
All that said, see for yourself -- download his viewer and report polygon counts in Comments!
Images courtesy NiranV.
Thanks for the update :) I was the one who posted about using KTris. KTris *must* have some bearing on what the viewer is actually doing, unless the stat really is meaningless. However, I went in with the object information on Firestorm and saw that there were many more triangles in the bodies than the KTris usage would suggest.
(I can't use Black Dragon, because I'm on Linux.)
Posted by: Bastilla Loon | Thursday, December 20, 2018 at 03:55 PM
Niran is right about reliability. In the linked post I read Bastilla reports "76 kTris" for Belleza, while to me it was close to an half million triangles. BTW, for my Maitreya Lara, Firestorm reported 100k for the body only and close to 200k with hands and feet. My values were closer to the actual ones, because I rezzed the bodies (in no-script areas, as some item has anti-rez scripts) and I tried to reduce the LOD effect to a minimum, then I used the Firestorm Inspector. BTW, in the new Firestorm 6 beta, I get 521k for Belleza in this way. However some other values weren't accurate and I had even to skip the TMP body, because that body is packed into a tiny ball - which actually it's good when you rez, so you don't appear like a blob of floating parts, but not so good to check it with the Firestom object inspector.
However, Black Dragon Advanced Complexity Window is perfect and really handy!
Here is what I got for the other bodies, with Black Dragon:
Altamura free bodies
To be fair in the comparison, these bodies are a complete avatar: head, body, 2 hands, 6 feet (3 feet pairs/positions). However, I don't know if these free versions have onion layers. For sure Altamura Giselle head has layers and the polycount is 40k, which seems to fit with the below numbers. Anyway, if you walk around with these bodies:
Jenny: 129k. Kim: 134k.
EBody free version (non-bento) v8.5/6
These come with 3 onion layers, although not available in the HUD.
Classic: body: 37k. Hands (2, 1 pose): 18.6k. Feet (6 in 3 positions): 37k.
So the whole stuff it's just more than 90k.
Curvy: body: 75k. Hands and feet are like the above.
Lucybody Afrodite
Body with hands and 3 pairs of feet, like Altamura, minus the head. Layers look like optional separate addons (not a bad idea). The number here is without addons:
184.5k
PsyCorp (demo)
It looks like it has 2 layers.
Core body Female 2.1 Type A: 49k. 1 Fem hand: 12.5k, so 25k for both. Feet Fem A: 21k.
So the whole thing would be 95k; but I saw only 1 feet pair, had it 3 like others, the count would be higher (unless I missed something, as this body is different than the typical human body).
Tonic Fine v5 (demo)
Full body with hands and 3 pairs of feet and onion layers.
267k
TMP, The Mesh Project, female body
Body: 57k. Hands: 445k(!!). Feet: 129k.
Polycount-wise, it looks like we have something even worse than Belleza!
How so? TMP is not bento: instead of a pair of bento hands with moving fingers, it comes with a bunch of different hands in various poses. Moreover TMP comes with 5 pairs of feet, not 3.
Belleza, comparison note
To have a fair comparison, I need to add a note for Belleza too.
Belleza bodies are a single item comprising of body, bento hands, feet and onion layers. Unlike others, the feet pairs/positions are 4. Furthermore they actually have 3 pair of breasts (pushup, etc). As if that wasn't enough, they have an extra pair of arms and legs (and accompanying layers): Belleza feet and hands are positioned differently then SLink and compatible brands, thus you can't simply hide the feet and put SLink feet on; so Belleza bodies have those extra arms and legs that are positioned differently to be compatible and the HUD flips them visible or invisible when you need them.
These things work like those pre-animesh animations, where you have all the frames in a single item, all of them invisible, but one. They were complex of course. This surely contributes to Belleza heavy polycount.
However, the result is an half million polygons that you carry around.
Posted by: Pulsar | Thursday, December 20, 2018 at 08:22 PM
Black Dragon viewer is unreliable for determining arc on avatars. When you download the viewer using the third-party webpage at lindens labs there is a note about the conflict between actual arc and that viewer. Don't get me wrong. Yhe viewer has its uses but determining your arc is not one of them.
Posted by: Lizzy | Friday, December 21, 2018 at 05:40 AM
While bodies are a problem I agree, the switch to dense mesh in the Home and Garden arena is even worse. I mean we each wear only one body at a time. We can fill a room with many many decor items.
While not quite near tears, I was VERY disappointed to find that most all of the houseware and decor products that I picked up as gifts this season were over the top in tri count -- as well as so very many textures. One small item (about six inches by a foot in RL terms) was well over 200K tris with 37 textures (not small ones). I was tough though; I deleted it along with its other high count buddies.
Happily along with about a dozen discards, I found two items that I felt I could keep and use. That isn't a good ratio. You CAN have lovely and optimized at the same time, but we certainly aren't seeing that these days.
These are not unknown "newbie" creators, but likely more folks going over to the darkside and downloading mesh made for RENDERS instead of game assets and passing them off as "original" creations. They are gorgeous, but tremendously laggy. Food items seemed to be the biggest offenders.
Posted by: Chic Aeon | Friday, December 21, 2018 at 08:11 AM
well, are seem good on inspect others, but what you do on yourself?
Posted by: bufje | Friday, December 21, 2018 at 10:34 AM
@Lizzy
Ignore the note from LL.
BD's arc calculation is much better at determining your actual avatar complexity than anything LL has to offer right now. LL's arc calc is a joke. I'm not going into details yet again, you can read the reasoning for changing it on my blog.
But to give you a short TL:DR
LL's ARC:
Optimized Avatar vs Unoptimized
80,000 ARC vs 40,000 ARC
150,000 polygons vs ~3,000,000 polygons
BD's ARC:
Optimized Avatar vs Unoptimized
250,000 ARC vs 3,000,000+ ARC
150,000 polygons vs 3,000,000 polygons
The numbers are arguably big yes but the calc does a much better job at getting rid of bad avatars (unless you set it way too high or disable it of course).
Posted by: NiranV | Friday, December 21, 2018 at 04:58 PM
Even though I agree that the formula BD uses for ARC is more representative of the true load an avatar puts on the system than LL's, they are different and I can even see why people can get hung up about it. But the bottom line is that it's irrelevant to the points raised in this article. There is no "formula" involved in a raw polygon number. That's part of the input data, not a result. It is merely counted, not calculated.
The polycount is the polycount, whatever the viewer does with it afterwards to calculate a "complexity" metric. And most mesh avatars in SL are stupidly, insanely, polygon-dense. I've seen sets of FINGERNAILS for Bento hands that use nearly 20k polygons!
Posted by: DaveB63 | Sunday, December 23, 2018 at 06:20 AM
Well it is all too late now. Unless someone can break the stranglehold that the top 2 or three bodies have upon the lower end of creation in SL.
For a long time now I have been buying things in SL to use in my builds, and in the end, having to re make versions of them myself in order that they make sense to be used in world.
And what did LL do to help this problem of crazy poly count addiction by creators? they upped the prim limit in Sims, so that people can go even more crazy.
Long time back I came from an Unreal Engine level designer background where one of the first meaningful comments I ever got about one of my creations was from the once Legendary Cliff B. It was “Love the levels you have created, but your polys are out the ass”.
When I first came to SL way back in 2004 it was from that heavy critical background where poly count was all you heard about.
Right from the start in SL it seemed crazy to me, but you find yourself getting sucked in and no longer bothering to care. Except now I build on Homesteads, and in a way I am glad to have my restrictions back.
Back in Unreal in the early 2000's of course everyone would have loved to just go nuts and say, sod the poly count I am going to build what I want. But it was ingrained right from the start that all that did was kill the game in the long term.
I have heard many times from top designers who work with the top mesh bodies how seemingly irresponsibly, from a performance perspective they are constructed, and how the creators seem deaf to any calls for improvement.
Of course I do not blame them, they dug themselves a hole right from the start. Because it is from the start that you have to get it right. As LL themselves know only too well.
Their hole is so deep and almost impossible to get out of that they have attempted to dig a new one.
There is no doubt at all that the creators of the top bodies have worked extremely hard and well deserve their positions at the top of the pile.
But that should not make them free of, or able to be deaf to, serious criticism about their products. In fact people should always be made aware of the problems they are buying into.
My personal gripe is not so much with the construction of the base bodies, but with the monopoly that the top 2 or 3 bodies now enjoy, and how this translates to their total control of mesh clothes creation in SL.
Anyone with any talent at all in this area, or even those who just wish to maybe have some fun trying it out, as we all did when the standard SL body was still the dominant base, will find a pretty much impenetrable wall of hurdles to get over in order to get the hallowed mesh dev kits.
Unless of course you can beat the seeming catch 22 situation wherein you have to be a top creator in order to get the kit that top creators use.
This has to be wrong somehow.
Of course a creator is free to do as they wish with their product.
But when they are almost the only show in town, they must surly take on a kind of higher responsibility to the community as a whole.
They cannot behave with quite so much self interest as they once did, or at least they should not be allowed to do so by the owners of the world they have come to dominate.
Posted by: JohnC | Sunday, December 23, 2018 at 07:46 PM
I agree that it's bad that mesh bodies are dominated by 2 or 3 creators, and that the mesh dev kits aren't just freely available as they would be for the system body. However, it's not quite right that you have to be an established designer to get one. When I was getting started a year ago, and had only made one or two pieces of mesh clothing, I was able to get the Slink mesh dev kits just by asking. Belleza and Maitreya didn't respond.
6-9 months later I tried again, and this time got Belleza. Still no response from Maitreya. (I filled out the form a third time a couple of weeks ago; nothing yet.
So my experience does seem to match the idea that there is an impenetrable barrier against getting a kit for the dominant body (if Strawberry Singh's survey is representative, it had something like 50% of the female mesh bodies out there), but Slink was very responsive to requests.
It is very unfortunate that you have to get individual permission and software from body creators to make clothes that work well with those bodies. It would be analogous to needing a special license from Microsoft to be able to write Windows software. (That situation *does* apply for the Xbox, I believe. The world is filled with these unfortunate walled gardens.)
Posted by: Bastilla Loon | Wednesday, December 26, 2018 at 05:10 AM