Mesh Body ACI Values Compared

Mesh Body Addicts is a fun and informative site. The recent article AC and Jellydolls: Tutorial and Popular Bodies list took some effort. The list gives the ACI for each of the mesh bodies being marketed in Second Life™. There are some brands I was unaware of. So, that is a plus too.

JellyDoll and Fully Rendered Avatar.

JellyDoll and Fully Rendered Avatar.

The list at Mesh Body Addict’s site is alphabetical. Handy for finding your body. I wanted the list arranged in order of ACI so I could see the best, lowest ACI, and worst, highest ACI, for lag. So, I resorted the list: 

1.       Belleza Isis – 7,462 (full version); 3940 (Lite version)  (Including hands and high feet)
2.       Belleza Venus – 7,750 (full version); 4,02  (Including hands and high feet)
3.       Belleza Freya – 7,880 (full version); 4168 (Lite version) (Including hands and high feet)
4.       eBody v8.2 – 8,066 (including hands and classic feet)
5.       Altamura Sofia – 8,452 (Including hands and feet)
6.       Tonic Curvy 3.02 – 8,688 (Includes hands and high feet)
7.       Tonic Fine 3.02b – 9,518 (Includes hands and high feet)
8.       Maitreya – 9,606 (Including hands and feet)
9.       Adam – 12,336 (Including hands and feet)
10.   Slink HG v2.9– 14,074 (includes casual hands, deluxe high feet)
11.   Slink Physique v2.9.2 – 14,496 (includes casual hands, deluxe high feet)
12.   Slink Physique Male v1.3 – 15,056 (including casual hands & flat feet)
13.   TMP (beta) – 14,745 (includes ouch feet and deluxe hands)
14.   SelinA Petite v2.1 – 17,746 (including hands, flat feet and nails)
15.   SelinA Busty v2.1 – 18,010 (including hands, flat feet and nails)
16.   Eve (Slim) v8.4 – 27,370 (Including hands and feet)
17.   Eve (Pulpy) v8.4 – 24,794 (Including hands and feet)
18.   Signature Gianni v2.7 (old version) – 41,956 (including hands and flat feet)

Belleza wins hands down. Of course ACI isn’t everything. How well the HUD worlds and easy it is to use is a significant factor. How the body looks is another factor. Our personal preferences will decide what we think is the best body for us. So, don’t let my saying best and worst for lag be the deciding factor.

I do find it interesting that ACI spears across such a wide range of ACI. If I take Belleza Full as a base line (7462 + 7750 + 7880)/3 = 7697 then Signature is 5.4 times more costly. My Slink, which I love, is 1.9 times more costly.

The SL classic avatar comes in at about 1,000. We could take it as the base line but, that’s no fun. I seldom wear my classic body these days. I still use the classic head. That is because I hate the cast-in-place expressions. Of course I seldom use facial emotes so why I am that way makes no sense. Somethings we just choose not to be rational about. I like chocolate, preferably in 1 ton bars.

It does seem designers are reacting to ACI. So, will we see those ACI numbers coming down? I expect we will. That is a good thing.

I’ve been using Memhistory, a free program that shows how much computer memory is being used. It will give me lists of how much memory each running program is using. Mostly I am just interested in what the computer is doing when my viewer slows to a crawl and mini freezes set in and I eventually crash, if I don’t relog. It has been years since I have been on a relog to avoid crashing process. I remember when Emerald and the Linden viewers had to be restarted every hour, if you were lucky, to avoid crashing.

What I find is my viewers do pretty well until I enter a crowd. Then the viewers start eating memory. On an 8GB machine the viewer will use up 6+GB of RAM. I can use a memory cleaner to recover some memory and keep going. The Linden viewer will let me do that. The Firestorm viewer will too but, if after cleaning memory I don’t teleport or restart soon Firestorm will crash. I can’t say it is perfect cause and effect but, I am treating it as if the memory clean will lead to a crash.

In areas with only one or two avatars, memory use doesn’t seem to be that much of a problem. The viewers do very well. Even in areas with lots of mesh buildings, memory use doesn’t seem to be much of a problem unless a lot of avatars are around. So, I believe ACI is a big deal.

What is the Limit?

The Firestorm Team and Linden Lab have set some default limits that vary depending on your video card. I don’t see where the Lab has published that information. They are tweaking the values so it makes sense they wouldn’t. Firestorm published this: (reference)

  • Low: 35,000
  • Low-Mid: 100,000
  • Mid: 200,000
  • Mid-High: 250,000
  • High: 300,000
  • High-Ultra: 350,000
  • Ultra: 350,000

Personally I am working to keep my outfits below 150k and have mostly kept them below 100k and that is with my 14k Slink body. I hasn’t been hard. Sadly I don’t wear the hair I love that is 147k on its own. But, I am past that loss.

There are people with their limit set pretty low. I got a couple of notices that I wouldn’t be rendered by everyone in the region when I was wearing an 86k outfit. I assume they have a low end computer.

Your computer and preference will decide what limit you set. For my GTX 560 I use 120k to 150k. I am having a problem with my new machine, memory issues, so I don’t know what setting I’ll use with its GTX 1060.

Others I’m talking with seem to be settling on values between 100k and 200k. Others I read about are setting theirs to UNLIMITED. I think that is foolish and indicative of ignorance of the setting’s relation to video griefing protection.

I use the Firestorm setting to show me ACI for JellyDolls, those that exceed my ACI limit setting. I have seen some with >600k ACI.

I think we will start to see ACI numbers coming down and designers doing a better job of optimizing their stuff. We may see the lag in crowds start to decrease. I am hoping.

Warning

Be aware that these ACI numbers are not completely guaranteed. Some people are questioning them and asking for changes. So, forget getting all righteous and ripping on some one with 1,000k ACI. See Avatar Complexity Oddities.

Leave a Reply

Your email address will not be published.