The Effects of Mesh

Tateru Nino writes the Dwell On It blog, a favorite of mine because she does numbers. Today she has an article titled, The effects of the Mesh content-type on Second Life. The article is an accurate assessment of the current state of ‘mesh’ in Second Life. The comments show the user reaction to mesh and are interesting to read. I have a very different opinion than I see in the comments there.

It is All Mesh

Opinions

In psychology there are the concepts of transference and projection. If you have a new age belief system, you know those concepts as mirroring. The concepts are a mix of ideas that basically results in the idea that what we see in others is ourselves. So, if we see someone being dumb, we are probably seeing our own stupidity. If we are smart our first thought would be to ask of our self if we are being dumb and missing something. At the least it can save us considerable embarrassment.

So, with mirroring as the top cliché read through the comments on Tateru’s post.

Defining Mesh

I read lots of stuff. Somewhere in my reading I came across an idea Einstein expressed in a conversation about language. It contained the idea that people cannot communicate or understand each other if the language does not have words that precisely describe concepts. If we use ambiguous terms we will have ambiguous discussions. Ambiguity is a disaster in technical conversations and any conversation where one hopes to achieve some consensus of opinion.

 To talk about the effects of mesh on Second Life we have to get away from using the word MESH with the definition we have assigned it in SL. We seem to have generally defined it as a special new feature just added to SL. It comes across as a THING. It is not a thing as the feature is implemented in SL.

We can see the confusion in the SL forum. People are frequently asking, what is mesh? The answers are surprising.

Many existing residents not already making sculpties have no idea what is meant by ‘mesh,’ vertices, UV Maps, and Level of Detail (LoD). SL is and always has been mesh, has used vertices, UV Maps, and LoD. Most answers miss that point.

The current MESH FEATURE is an IMPORT feature that has been added. That’s all. The import feature allows us to use professional tools outside SL to create free style 3D models and bring them into SL. It is all about being able to edit mesh vertex by vertex with full control over textures and LoD using professional level modeling tools.

The SL Viewer has undergone major changes to handle objects made from free style mesh. Prior to this change the viewer could only render predefined 3D objects, primitives or prims. The upgrade make SL much more like other 3D games.

My Take on the State of Mesh

Having free style 3D objects creates a number of complex issues the Lab had to solve. Lag is always a major complaint of SL users. To reduce lag, building in SL needs to be efficient. Until the release of viewer 3, efficiency was imposed by limiting builders to a predefined set of relatively efficient objects, primatives. However, some of them were not very efficient. A cube and a torus have very different impacts on the rendering process and servers. But, they have the same cost to the builder. If you want to see a render scene in SL slow down, put a thousand twisted tori in it. You can’t do the same thing with cubes. That inequity is being corrected now.

The possibilities for horribly inefficient builds were introduced with free style 3D models (mesh). The Lab had to handle that problem and do what they could to push SL users to build efficient models. That is done with Land Impact (LI) costs, previously named disastrously misleading as Prim Equivalence (PE).

Learning to make efficient 3D models is complex. It won’t be for the majority of residents. Land Impact is a way to put the need for efficient modeling in users’ faces. If you insist on building inefficient models and lagging us, it will, at least, be expensive.

Where We are Coming From

SL is a creation environment. They did an excellent job of creating an easy and fun to learn modeling system. But, it was for novice 3D modelers. It limited everyone to the predefined objects, prims. Professional and advanced modelers did nice things with the simple system. But, they were limited. For many it was too restrictive. By adding what we call ‘mesh’  (free style 3D models) they added the advanced levels of modeling for professionals. It makes SL attractive to a new group of people.

It is like a small town adding a college. The addition makes no difference to most of the residents. It also takes time for the influence of the college to show. The same is true for the new free style models.

Where We Are Going

We know the Lab needs and wants to increase user retention and add more new users. By adding free style models they have made SL much more attractive to advanced and professional modelers.

We know other features are being added to make SL easier to use. They are tackling both ends of the user levels, novice to professional. Direct Delivery of purchased items is an effort to make buying and wearing clothes and stuff easier. Free style models are for the advanced users. Basic and Advanced mode viewers are for new and experienced users.

It has been difficult to use SL for RL work. To move models from AutoCAD, a professional RL design and drafting system for engineers and architects, to SL required a US$3,000 addition to AutoCAD (I think I have the price right). That has changed with the import feature for models in Collada files. This should make SL much more desirable in the professions.

For now the Lab is optimizing and stabilizing the viewers for mesh. Viewer 3 is likely still running a lot of debugging code. The Development viewer is even slower than the main viewer. Lots of the Lab’s effort and manpower is shifting to performance and stability issues. It takes lots of data to find the bottle necks and understand what has to change. Thus additional tracking code is needed in the servers and viewers.

Many residents are learning to make efficient 3D models, some just models – forget inefficiency. Neither a simple task. But, the knowledge is usable in other virtual worlds. I’m learning Blender and working with a small team to move things from Myst games (fan made content) to SL and the Unreal Game Engine, which is awesome and a little less new and probably less buggy than CryEngine3. All that learning takes time and keeps people out of SL.

The Effect

In the four weeks since the ‘Mesh Release’ I am not surprised that we are seeing little, if any, statistically visible change from the addition of free style models. I think it will take months and will be indirect. It won’t be like the change in the sign up process where we see a dramatic and now persistent improvement.

Also the graphics quality of games only has a small effect on player retention. The Facebook 2D or isometric games are a good example. So, I can’t see a prettier world changing retention in any major way. It has possibilities, but we have massive stats and numerous examples that graphics quality is not a significant factor in retention. Graphics are a major attractor, not a retainer.

Summing It Up

In many ways we are wondering what the Lab will do to improve SL player retention. Rod tells us there is a plan. We know the generalities. We know the sign up process change was a part of that plan. We know the age verification process is changing. We know purchasing things in SL is to change, at least the delivery side. We know there is a plan to be the virtual worlds’ market center. There are other tidbits of information that we have; premium membership, premier sandboxes, more anti-griefing stuff, future marketing campaigns, and more. By year’s end we are supposed to see more changes implemented that help with player retention.

Part of the plan is obviously to change the financial equities of building in SL to force more efficiency and less lag. The manifestation of that is in Land Impact costs. We have heard about script limits for some time. But, counting scripts was too simple and ineffective. Using the idea of server, network, and render load is more effective. Rather than setting limits they are assigning costs. As we go forward I expect to see more Land Impact metrics exposed and costs implemented.

But, I think it is safe to assume we are missing parts of the puzzle. We know the Lab needs to shift the financial model of SL and do something to convert a significant percentage of the sign ups to active users that stay with SL. So, what is their vision and what are the aspects of it we are not seeing? Will learn more over the coming months.

Leave a Reply

Your email address will not be published.