An interesting post popped up in the official blogrum today concerning confusion about Mesh objects and prims counts.
While the technical ins-and-outs of the main discussion largely go right over my head, Lightscribe Infinity raises a point that, if accurate, could lead to a lot of additional angst in-world in the coming months:
ARC is high for any av mesh. The default av, simply smoothed and brought inworld, increases ARC dramatically. 91938 versus the 0 of a present av, with everything detached.
ARC – or Avatar Rendering Cost – is a misleading “measure” introduced by Linden Lab some time ago that was designed to provide a loose – one might say entirely arbitrary – measure of the “cost” to the SL Viewer in rendering your avatar, based on the number of textures and primitives (or indeed, sculpties and (soon) Meshes worn by your avatar. The basic value for an unadorned avatar is 100 (and not the 0 mentioned by Ligtscribe, in fairness), but the more prims /textures your avatar wears – and the higher the resolution of those textures – then the higher the ARC count generated.
ARC itself appears as a number floating above avatars’ heads once the option is enabled from Advanced > Rendering > Info Displays > Avatar Render Cost, and it is perhaps one of the most misunderstood information displays available in Second Life, and one of the biggest causes of unnecessary drama. This is for two reasons:
- People look upon ARC is absolute terms. That is, a) the look at it in isolation of other factors; and b) they treat it as if it is an entirely accurate measurement in and of itself.
- People believe ARC lags the simulator, and thus directly impacts everything.
In fairness, the first item – ARC being viewed as an absolute measure – could be said to be partially LL’s fault inasmuch as the ARC count is “colour-coded” when displayed, with “low” lag counts (approximately up to 1,000) being shown in green, “medium” lag counts (between around 1,000-2,000) appearing in yellow and those above 2,048 appearing in red. These colours alone send out entirely the wrong message to the uninformed, given that most us of automatically associate green with “good” and “acceptable” but red with “bad” and “dangerous” – and something that “must” be avoided. However, as the SL wiki entry on the subject itself states:
ARC isn’t the end-all for pinpointing Viewer lag. It’s a useful estimate which should be used in conjunction with other tools
The second point is completely and utterly wrong. ARC is a factor – as the name states – of rendering. Rendering is a function of the Viewer, not the simulator. As such, rendering is dependent wholly on the client end of the equation – what computer you are using, what graphics card / system, the amount of memory it can utilise and so on, and has nothing whatsoever to do with the the simulator itself.
These basic misconceptions have resulted in a wide range of myths sprouting up around high ARC counts – such as:
- a high ARC count is indicative of an avatar carrying a high script load.
- No, it isn’t. Scripts are not, and cannot be, rendered ergo they have absolutely no impact on an overall ARC count
- No, it isn’t. Scripts are not, and cannot be, rendered ergo they have absolutely no impact on an overall ARC count
- a high ARC count directly affects your ability to teleport.
- No, it doesn’t. Teleport failures are largely related to how many scripts your avatar is carrying and/or server-side issues entirely unrelated to ARC)
While it is true that ARC can provide a measure of the overall cost of rendering an avatar by every single Viewer that can “see” that avatar, the point that people forget is that it is not, nor has it ever been – and the SL wiki states – the only measure. More importantly, its not even an accurate measure in any way, shape size or form; it is simply an estimate. If only those who bow down before the Great God ARC would actually understand this!
The fact that LL openly admitted it to be a subjective measure, entirely dependent on a wide range of factors entirely outside of their control, at the time it was introduced went almost entirely unnoticed by those who use now use ARC as a club to batter people with, and who have reached a point where no amount of reasoned argument will sway them from their entrenched view that those displaying a high ARC count deserve to be shouted at, badgered, and / or banned from sims. Not even perfectly obvious demonstrations of the ineffectiveness of ARC as an accurate and reliable measure can break many of these so-called “ARC Police” from their misguided views.
For example, if ARC were truly an accurate measure of an avatar’s impact on things, then one would expect all ARC counts to be constant for any given combination of clothing and attachments worn by an avatar.
But it is not. For proof of this, simply get a group of friend together, or go to a shop or mall where there are a number of avatars and turn on ARC. You’ll get a range of counts appearing above people’s heads – but if you cam around, zoom in or out – the counts will change! Simply teleporting from sim to sim can increase / decrease an individual’s ARC, simply because the need for the client to render everything else in the view is impacting on the count ascribed to the avatar itself; and before anyone starts telling me the error of my way, yes, every single object you see in your world view also has it’s own rendering cost as well.
So how does this relate to worn Meshes? Well, if Lightscribe is right, and Mesh attachments end up sending ARC skyrocketing, then doubtless the ARC Police (which is perhaps the politest name for them compared to others I’ve seen) are going to be bludgeoning a lot more people with there ignorance and creating even more unnecessary drama and angst.
Which is a shame, because one gets the feeling that Mesh is going to have more than its share of drama, upset and misunderstandings once it arrives on the main grid, without anyone stirring the pot even more.
Inara Pey: Living in a Modemworld 
Excellent article!
To make things even more confusing, ARC’s formula has changed over the years. For instance, it did include the cost of scripted attachments (vs. non-scripted ones). But it certainly doesn’t have any impact on the sim, unlike what most people fear. In fact, it’s really quite hard to get most residents to understand that very few things actually impact the sim — or, rather, very few things under the user’s control.
Displaying your friends list — specially if it’s huge! — will have a far, far greater impact on the sim than 250-prim-hair, for instance. That’s because of some poor implementations done by LL on the communications protocol…
On the other hand, I remember Babbage Linden mentioning that a lot of avatars tend to use up to 64 MBytes of RAM with all their scripted attachments. While the scripts themselves have little impact (most might even be “asleep” and not using CPU), with 100 avatars like that in the same sim, that’s over 6 GBytes, and the sims have far less than that available — which will push the virtual server running the simulation software to its limits, consuming way over the available memory. So, well, there are ways to create some impact just wearing too many attachments… but that has little, or nothing, to do with the impact of meshes, attachments, etc. which is supposed to be “calculated” by ARC or any measure of that kind — which only impacts viewer performance, and that is all local.
LikeLike
The amount of confusion between ARC and script counts is mind boggling.
Again, whether or not an attachment has scripts in it makes little difference to the ARC count – it is the complexity of the attachment in terms of the number of prims, the number of unique textures on each of those prims and the resolution of those textures (are they 32×32, 64×64, 256×256 or – God help us, 1024×1024 or greater?).
Your point on the use of MONO scripts for absolutely EVERYTHING is well made as far as scripts and there impact is concerned. Until such time as SL can accurately measure the real memory use of such scripts, rather than simply assigning each of them 64Kb at a time *regardless* of how much memory they actually use – then scripts are going to be one of the few user-controlled items that continue to run riot in terms of overall resource utilisation. And even then, we’re still going to need the proposed script controls to effectively manage things.
But this is, as we agree, irrelevant where ARC is concerned. Nevertheless, we’ll doubtless see people getting walloped with the ARC ban hammer simply because they are wearing a cool avatar Mesh. Robots and furries, take care!
LikeLike
That’s really what saddens me. I certainly understand that a lot of people have slow computers (I have a very slow one too), and that’s why high-ARC avatars tend to ruin their own personal SL experience. Sadly, so many believe that if their own computers run slowly, everybody else’s will…
LikeLike
Or, at the risk of sounding cruel, if their own computer is lagging, and they have a moderate ARC count – they take the view that the fault must be with *everyone else*. The lack of performance within their own computer just isn’t a factor in their thinking, nor is any consideration given to the environment their in – and whether their computer is struggling to handle dozens of 1024×1024 textures on things like walls, floors, ceilings, vendor boards or whatever else is around them.
They simply see red ARC numbers and start hitting people.
And it saddens me that many commercial parcel owners react the same way. They’ll happily slap high-resolution images across dozens of vendor boards (and even have vendors “paging” through images of their wares) and not give any thought at all to the rendering impact all this is having on their own computer – or anyone else’s – but as soon as someone with a red ARC count enters their property, they are all over the poor unfortunate like a bad rash.
LikeLike