
Update: See BUG-9962 for issues relating to avatars becoming stick as Jelly Babies when using Avatar complexity.
On Friday, August 21st, the Lab issued their Project Quick Graphics project viewer. Version 3.8.4.304433 brings with it the much-anticipated Avatar Complexity and graphics presets capabilities, both of which are intended to assist in improving viewer performance for those on lower-specification computers.
I provided an overview of the viewer while it was still in an early version, so this is offered as a further update.
Avatar Complexity introduces a new slider to the viewer which can be used to set a level above which avatars requiring a lot of processing will appear as a solid colour (including their attachments), giving them the nickname ” Jelly Babies” after the sweet (candy) of the same name.
As avatars can often be the single biggest impact on the viewer in terms of rendering, particularly in crowded places, using this slider is intended to greatly reduce the load placed on a system compared to having to render them in detail, allowing users to adjust the setting according to circumstance – the setting can be increased, rendering more avatars as solid colours in crowded regions, and turned down for quieter spaces. At the same time, there’s also the ability to set how individual avatars are rendered on-the-fly during the current log-in session.

The Avatar Complexity slider can be found on the Advanced Graphics floater (Preferences > Graphics > Advanced Settings…), The values run from 19,999 to 300,000, above which it switches to No Limit, meaning all avatars in your field of view will fully render, with the default based on the rendering performance of your system. As noted in my last piece on this, the values used by this slider are based on those previously used to determine Avatar Draw Weight / Avatar Render Cost.
It is possible to see the render complexity of all avatars in your field of view (including your own) by enabling Advanced > Performance Tools > Show Avatar Complexity. This displays a series of figures above avatar heads which is updated in real-time. The one likely to be of interest to most users is at the top: the actual render complexity value. This should remain fairly constant, allowing for how people might change their appearance by adding / removing items and changing their appearance.
The viewer also generates information messages in the upper right corner related to Avatar Complexity. One is displayed each time you change your own avatar’s appearance and impact your own rendering complexity. The second acts as an indicator for when you’re over the limit of “too many” of the avatars around you, and are being rendered as a Jelly Baby.

A further element of Avatar Complexity is the ability to selectively alter how individual avatars are rendered on-the-fly. This is achieved via the right-click Avatar context menu, which includes three new options:

- Render normally – the avatar will render as defined by the Avatar Complexity setting. If the avatar’s complexity is lower than the setting in the viewer, it will render correctly; if it is higher, it will render as a Jelly Baby
- Always Render Fully – does exactly what it says – the avatar will always be fully rendered, regardless of it exceeding your set complexity limit
- Do Not Render – renders the avatar as a Jelly Baby (or even not at all save for name tag if already very easy to render) regardless of your Avatar Complexity setting. Note that this setting does not persist across log-ins (so if you re-log, those avatars you’ve used it against will render normally), and it will not block the ability to read their local chat or receive their IMs, etc.
There are a couple of final points worth mentioning with Avatar Complexity. The first is that it is not a replacement for Avatar Imposters, but can be used alongside it. The second is that with this project viewer release, the colours of Jelly Babied avatars has been muted when compared to test versions of the viewer, making them a lot easier on the eye (the image at the top of this article shows the former, more vivid colours).
Graphics Presets (see STORM-2082) allows users to create, save and use their own graphics presets, each designed to meet a specific requirement, and which can be quickly switched between with the overall aim of helping with viewer performance.
For example, one preset may have all the performance hitting items (shadows, projectors, etc.) turned on / up for times when the overall quality and depth of detail in a scene is important for taking photos, another may have them dialled-down for crowded places, and a third might have them adjusted further for “indoor” use (so draw distance is greatly reduced, sky and terrain details are set to low, water reflections turned off, etc.).

Once a preset has been set-up, using the revised Advanced Graphics Preference panel, it can be uniquely saved, and then applied at will using the either via Preferences > Graphics > Load Preset, or more directly by the Graphic Presets icon located in the top right of the viewer.
When the mouse is hovered over this icon (shown right), a list of all saved presets is displayed, a tick appearing alongside the one currently being used. Clicking on any other preset will immediately apply it.
In addition, this panel also has a button which will open the viewer’s graphics settings in Preferences.
As noted in my previous article on these updates, the Advanced Graphics Preferences panel has been seen as less-than-optimal due to its size; the Lab have acknowledged the feedback, but have not made any significant changes to the layout as yet with this project viewer release. Whether they do or not may depend on feedback they receive directly from users, and what they feel can be done to improve clear deficiencies.
The ability to create and save graphics presets is a welcome addition to the viewer – these are not the same as backing-up and restoring viewer settings as seen in other viewers, but do provide a fast and efficient way to adjust graphics settings according to situation, if needs be.
Avatar Complexity is liable to be an interesting addition to the viewer. While there is a risk of seeing a return of ADW / ARC drama, it also provides the means for people to accurately judge the impact their avatar might be having on others – and their own, given their avatar must be rendered by their own computer as well – SL experience. It also potentially offers content creators to better understand how the use of mesh and textures can impact other people’s SL experience, allowing them to further improve their products.
Those wishing to try the viewer for themselves can find it here. Keep in mind, that is it s project viewer, prone to possible bugs and to further changes from the Lab, and issues should be reported via the JIRA.
RIP Graphics preferences.
LikeLike
Exactly my thought.
LikeLiked by 1 person
Reblogged this on Windlight Magazine and commented:
Inara reviews the official Second Life viewer from Linden Labs, which has some cool graphics updates:
LikeLike
My usual Avatar has a remarkably low render weight. People have been able to find out their own number easily, High numbers are very common, maybe people don’t bother.
When this gets commonplace the reactions will be interesting, if people bother to use it.
LikeLike
I think the notices people get will draw attention to it, and as such, reactions will be interesting.
LikeLike
Useless crap added to viewer instead of a draw distance slider.
LikeLike
You really have no idea do you? Where avatars are concerned, as the heaviest load on graphics processing, a DD slider will have minimal impact on people’s performance compared to something like Avatar Complexity, simply because while at most busy venues, even at 32 metres, people will still be rendering a lot of potentially complex avatars which are affecting their local performance.
LikeLike
I fear you don’t use LL official viewer as you day to day one. Cause a draw distance slider is s0me real useful and used, be when sailing, flying and so on.
But well, what is the point of pointing some that reg daily sl users know and feel ll viewer lacks. Firestorm is the official viewer, right!
LikeLike
I’m not naysaying the usefulness of a DD slider. I’m commenting on the idea that you dismiss AC as “useless crap” and compare it to a DD slider when the use cases are totally different. A DD slider will have little impact for the use-case AC is intended to fill. Period.
LikeLike
Draw distance is one of the settings that’s saved when you create a preset (all graphics settings are included). While that doesn’t make it quite as easy to fine tune, if you’ve got a few settings you normally use, creating a preset for each works very well.
LikeLike
Thanks Oz, I’d totally forgotten to point to graphics presets as the solution for DD changes, despite discussing it previously!
LikeLike
tks for the info, guess i can emulate a sort of draw distance slider by saving diff presets with the only change on it.
But i still wonder, is to so hard to implement it as a slider on the top bar of the viewer?
LikeLike
Btw, if ll realy wishes to support users with old systems, they would not stop Xp and old mac to be used, they would still keep at least at a minimum level some official control over linux, but saying any against the Lab here is useless, worse then be blinded is to not wish to see.
LikeLike
Like it or not, XP and versions of OSX prior to 10.7 are no longer supported by their manufacturers. As such, there is the potential that continuing to try to support them places an additional burden on developers that small companies like LL aren’t in a position to afford – hence why they are far from alone in the decision.
As to Linux, that is an unfortunate decision. What’s more, I may understand the business economics behind the decision when looked at purely in terms of the volume of reported users running Linux (around 1%) – but that doesn’t mean I necessarily agree with it.
In terms of being blinded, not at all. I’ll critique the Lab as and when there is proper cause – as I often have. However, I long ago gave up blindly blasting them at every turn simply for the sake of blasting them. Instead, I’ve taken the time to understanding them, their operations and their business model, then weighing my feedback, thoughts and opinion accordingly.
LikeLike
Unsupported systems that are over 13pct of total, with users that can not upgrade, just ignore and let them fade?
Like the end of supported foreign languages that mean more then 20 pct of total users.
Sl exists not due to the company that owns it but despite it…
LikeLike
Talking to the Firestorm folk, representing the most widely-used viewer – and having their permission to quote.
They indicate their own stats suggest around 0.6% of their users are on Windows XP (32 or 64-bit). Even allowing for the numbers of XP users dropping as a result of things like Win 10 encouraging upgrades, a fall-off due to the initial blocking of XP without SP 3 installed, etc., this figure would still suggest the total number of people using Firestorm on XP to be a relatively small percentage of their overall user base.
When you add this to the fact that Mac users – supported or otherwise – have generally represented around 5% of the user community and Linux around 1-2%, it is really, really hard to see how you arrive at 13% of the total user base now being “unsupported”.
Also, it’s worth pointing out that the Lab are offering – for as along as they reasonably can – the means for those on XP / pre- OSX 10.7 who use the official viewer a means to keep accessing SL.
LikeLike
Girls, girls, calm down. We all know you both mean well. Let’s see it like this, very honest and very brutal and very easy: All SL residents are grown ups, or at least at an age where you can make conscious decisions. Everybody should know that SL is a graphically pretty intense application, even if it doesn’t show for the most part. So it’s logical you need a computer with strong graphics to achieve the best inworld experience.
That’s it. That’s all. The better your computer the better your SL. Easy as that.
So now all every individual SL customer can and has to decide on how much they value their time in world and how much they are willing to spend for the necessary hardware; or have their parents spend on it. I know ZZ is on a very powerful PC, so of course she couldn’t care less about LL’s graphics trickery. I’m on a fairly powerful system myself + using Singularity 64-bit Linux viewer, so I’m mostly ok as well. Only suffering a bad connection. So, yes, this Avatar Complexity stuff is fairly boring for people on good hardware, splendid OSes and good viewers. But ZZ must see that Inara wasn’t talking about people like us but about »those on lower-specification computers.«
LikeLike
Hm. The comments below are another example of people basically considering things to be a step below magic. As someone who makes avatars who are often quite high on the ARC scale, I can assure you it makes a big difference and draw distance is simply not the dominant factor for most SL rendering. There’s times when it helps, and I’m spoiled by Firestorm’s quick-access, but graphical presets will render that need almost entirely irrelevant *and* make it less of a hassle for the everyday user.
So, yeah, even if I’ll be rendered as a jelly baby more often than previous, thumbs up from me on these changes. 🙂
LikeLike