Things that some Second Life content creators do that are naughty

When creating in Second Life i always try to balance what i want to achieve with the limitations i set myself on the Second Life platform. Because there are not limitations, you could create something that rivals stuff on a PS4, but that would only nuke performance on the Second Life Platform. The ability to set limitations for your creations is hard to do if you are unaware of how stuff works under the SL hood, and sometimes i see creators building in ways that they think is prim saving but in fact could be making performance worst.



A few years ago Linden Lab started being more aggressive on the Level Of Detail settings. RenderVolumeLOD debug setting decides how far away from an object you should be before it starts to lower detail of an object. It’s default setting is a measly 1.2. There are two naughty things some creators do with RenderVolumeLOD. The first is that some have their RenderVolumeLOD set to 4.0 or above and build optimised creations at this setting not realising that any new person coming into Second Life will not see their creations as intended. The second is some creators purposely create stuff with REALLY low LOD import settings to save land impact then expect people to up their RenderVolumeLOD debug settings to see their creations properly. This sacrifice of performance for land impact gain is a creators biggest sin.



Some people like Sculpts but i personally hate them with a passion. They lie about their massive resource eating, they look ugly, take for ever to load and spend most of their time as giant blobs. To me their is NO excuse for any more sculpts to be made EVER. If you have the skills to make a sculpt, then you have the skills to make a mesh instead. Ok so there are ‘some’ instances where sculpts rule, such as region surrounding landscapes but most cases creators use them because they register as a single prim. They do in fact use a lot more resources than 1 prim, it’s just not telling you that because its on the old inaccurate prim calculation system. Its another case of sacrificing performance for land impact gain.

Dont trust a Sculpt, they lie!


Mesh of a Gazzilion polygons

One of the reasons Materials Mapping was so welcomed was because creators had got to a stage where they were packing more detail onto a mesh than ever before. Every wrinkle on a piece of clothing, every marble relief on a wall carefully crafted by 1000’s of polygons. The gaming industry has known for decades the trick of using material mapping to give fake 3D details to mesh models and so it was very exciting to finally get the feature in Second Life. Unfortunately many users do not see the benefits because their computers don’t switch advanced lighting on. What is worst is that some creators flat out ignore the reasoning behind material mapping and still import huge detailed mesh at terrifying poly counts. I bought a demo shirt earlier this year and while i admired the great detail of the wrinkles in the clothe i was shocked to find it was over 200 land impact and pushed my Avatar Render Cost (ARC) so far into the red that i felt like a collapsing star. In fact thats how i would describe wearing extreme poly mesh. Walk around with a regions worth of land impact attached to your avatar then you are practically a Black Hole sucking the resources from fellow SL residents causing everything to slow down and in some cases cause people to crash on lower spec systems. Its a case of sacrificing performance for detail.

Easy to get carried away in blender making a model look super smooth, but we ain’t playing a pixar movie.


Packed Regions

If there is one thing I’m guilty of its packing to much into a region. A full region has 15,000 land impact allowance and this has not changed in for ever. The problem here is that for the high cost of setting up then renting the server you’d be forgiven for wanting to cram as much in as possible. It’s not like you can just whip up another region at a cost of approximately USD$5500 for the first year just to spread the content when your first region performance starts to degrade (cost based on UK resident). So i find myself filling up my one region with items that do a lot of scripted physical things and this can bring performance down especially if the objects are constantly chattering, and by that i mean sending packets of info about position, colour, what they detect etc.  Its a case of sacrificing performance for getting your monies worth.

Managing a region is often a case of what can i improve and what can i remove.


It can appear there is no limitations to what we can create in Second Life, but lag and low frame rates show there are limitations and creators need to know and understand the limitations in order to create content suitable for the platform. (i must also add that I’m not an expert so if you do know more than me about whats going on under the SL hood feel free to correct me in the comments :-P).

13 thoughts on “Things that some Second Life content creators do that are naughty

  1. Mesh gives you the opportunity to go haywire with textures, and some creators do exactly that. Just know that each time you fly that helicopter with at least 12 full 1024×1024 textures into a sim (one for each ‘face’), you’re slowing down the already problematic loading of textures consi…de……ra……..bly. And if the thing’s got Materials, it might have as much as 36 unique fullsize textures to load.

    • I was playing Disney Infinity recently, specifically their ToyBox mode where you basically make a region to play in with buildings, toys and stuff. They had this Performance Barometer telling me how well the region would perform with the content i had placed out. Is something like that possible in SL? We have the ARC for avatars(which most ignore), do we have something like that for Regions?

      • Many years ago, there was at one point a test viewer that had a rendering mode that highlighted things in colour codes in terms of texture and complexity. Unfortunately, it was never released for public consumption.

        Most responsible content creators are going to know about how to check texture size and wireframe for density, or to check RenderInfo via the Develop menu to check actual triangle count (be it for a whole scene or for individual pieces and their LODs) – unfortunately, there really isn’t a decent tool for the common consumer… and the fact is most don’t know or care. Worse, there are many content creators who should know better and actively choose to cater to that demographic rather than attempt to make something both look great and be more efficient.

  2. I would love to see a list or brand that creators could use if they maker lower poly count stuff and their own LODs (rather than auto lods). I spend hours looking for clothing and building stuff that won’t kill performance because of triangle count or explode into ugly triangles at the bottom few lods, but it is very very hard to find anything that fits that bill. I spend a lot of time making sure that my stuff looks the best it can up close and at a distance while keeping things fairly efficient, but there is no way to easily explain it to consumers either.

  3. I saw a wire railing on a house this week constructed from mesh. It aliased to a pixel wide line less than 5 meters away. Could have got a better result using alpha. Unfortunately a lot of content creators seems to think high poly detail = quality. In other words they’re not creating game ready optimized assets. This is probably down to a lack of training in many instances. I’m always inspecting stuff in wire-frame. As well as 1 pixel wide mesh railings, this week I’ve also seen a poster with a slightly bent corner that was probably around 512tris, it could have been done with 5 tris; and a lamp shade so dense in in wire-frame I couldn’t see air. I’ve been in buildings so poorly constructed that they LOD from less than half way across the floor space. All this bad craftsmanship is of course creating a lot of lag.

    Personally I really enjoy the process of creating LOD meshes. Here’s a few pointers:

    To avoid nasty inflation/deflation of the mesh when the LODs switch, ensure that during construction all the LOD sub meshes retain identical bounding boxes. These can be inspected in blender by switching view to bounding box. It’s not always possible to stop the meshes slightly changing shape when reducing loops etc, so I came up with a neat trick. I define a frame for the mesh by using two triangles. So for example I will have 2 triangles making a bounding box of 2m and build everything within those confines.

    Silhouette – it’s surprising how much you can reduce for the very bottom LODs, I’ve changed rocks into 4 faced black boxes and it’s been sufficient to trick the eye when panning in and out.

    Cruciform alphas – I often take a photo of my mesh using orthographic mode in blender. This is then used for my bottom 2 LODs. I keep the size small, usually 256 pixels. So you can imagine the up-loader is loving me when it’s going to generate 150 triangles and I’m supplying a mesh with 8.

    If anyone would like to see examples, hit me up in world and I’ll send some.

  4. Sculpts are terrible to make and especially to texture.
    With mesh i have a hate love. it’s pulling the fun out of building and creating things, blender is to complex for creating things. especially the materials section and texture alignment cannot beat the SL inworld tool. Worse that new things seems to be based on mesh.

    About textures, 1024 is maney cases not needed, but you forgot so quick to make a copy and resize it to a lower resolution. some advise very low resolutions but that gives a poor quality inworld.

    For now, still looking for a good and easy to use mesh tool with readable interface. other words font need to be scaleable. Blender have things that really could be better.

    Mesh is not soemthing to create on the fly, and that is how you build the best.

  5. I hate sculpts too, horrendous.

    I am guilty of a few of these things but am learning and bettering my ways.

    I try and make everything look reasonable for high and medium LOD settings and will eventually re-build those earliest meshes that were made just for low impact.

    I do use large textures and probably too many but I just can’t resist them, they make everything look so sharp and crisp
    But now I only use the big ones when I really need them and have been decreasing the others.

    I am however very guilty of packing a sim.
    I must.
    I want my city, I want it busy, I want it crowded, even though that means extra lag.
    Luckily it only has an impact on the area of the sim where the poor working classes live 😉

    • Yeh its hard to not use big 1024 textures when they give obvious better quality visuals. Im a bit sceptical about how the viewer handles textures and cache. On the one hand yes there is only so much texture memory to fill, but on the other hand isn’t the viewer built to selectively display content intelligently? In many ways SL is like a video game environment, but in many ways it’s also not. Do video games have interest lists and display only what you see? maybe some of the texture issues see are to do with how the cache works or how little texture memory the viewers are permitted to use rather than to many textures on content.

      I’ve tried to use one big texture instead of 10 little textures and i’ve tried to learn UV mapping techniques that allow you to use bits of a texture in multiple places. I’ve also tried to using big detailed textures for stuff at your avatars level and smaller low detailed textures for tops of buildings where your focus is never likely to be. I’ve tried lots of little ideas but still seem to get the same texture issues as always.

      • Yep!
        And just when your sim is perfectly balanced and carefully textured… you let in 80 tenants who all rez their 5 year old sculpties with mega textures 😉

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: