Content Creation Improvement UG

The new group had its first meeting. I’m bummed to have missed it, but RL fun was calling. They posted their meeting minutes so I can have an idea of what was happening. Some of the interesting things from the meeting follow.

Geenz Spad is chairing the meeting as the facilitator. He states the purpose of the meeting as: the idea is for people to discuss their ideas in a collaborative environment, and if everyone can agree on an idea, that idea will hopefully be made into a proposal at some point for Linden Lab, and prototypes demonstrating that functionality will also eventually be submitted as well.

So, this meeting is an opportunity for fashion and content designers to contribute ideas to enable better content creation and get problems in the design and creation work flow smoothed out.

Education

Stickman Ingmann had an item on the agenda about educating people to the limits of 3D modeling systems. Part of the challenge is in getting those limits from Linden Lab. Most 3D games have a polygon and texture limits guideline set by the game/platform designers. When those limits are exceeded the game’s performance degrades as the client console (think viewer in our case) and game servers are overwhelmed.

There is doubt amount those attending the Lab is going to set a standard. However, the Lab has published limits for various aspects of Second Life®. See: Second Life Limits. This page is edited from time to time by various Lindens.

In the Wiki’s mesh pages various polygon …quantities… are discussed to give SL users a idea of what works. The Land Impact information in the Wiki also provides information.

Stickman would like to see more information published on what computer hardware and software is in use by SL users. I take it that would be data on which CPU’s, GPU’s, and operating systems are most common and which viewers are most popular. I will guess that we MIGHT get the hardware information out of the Lab. But, probably not. The viewer information is already considered sensitive and is not published except under NDA (Non-Disclosure Agreement). So, we probably aren’t going to see any new information published.

What we have is:

One can spend a load of time reading through information trying to figure out what criteria to build for. But, all 3D games have the exact same criteria. Use the lowest number of polygons and the smallest textures possible to produce the desired result.

Consider the SL Avatar as a bit of a guide. The body, omitting hair, skirt, and eyes has about 7,000 polygons, counting everything in the avatar about 9,000. The Head 1,844, upper body 3,688, and the lower body 1,654.

It is interesting to note the full length skirt only uses 294 polygons. Using ratios a long sleeve shirt would have about 600 polygons. These numbers can help one get a sense of what their poly counts should be.

Stickman has a JIRA feature request filed for collecting the information he wants. See JIRA: WEB-2596Collect And Publish Opt-In Hardware and Compatibility Information on Users. I won’t be holding my breath, but do join the watch list if you think there is a possibility the Lab will publish the information.

Oz Linden points out, “The problem is that the range is too wide for documentation to be of much use to you. If you can imagine it, someone connects to SL with it (and some you probably would never guess). No one system type is even close to more than a low single digit percentage.”

What he is saying is that no single computer type or hardware configuration is common to more than 5% of the users. If you build for a target hardware group you are building for less than 5% of the users.

The Dance

I’m probably over educated on Second Life and how various parts are fitting together. So, a follow up comment by Oz highlighted the dance the group was doing. Oz asked, “Think about what metrics you could expose in the viewer to help people understand what contributes to costs.

Discussion was swimming with thoughts of how to encourage better and more efficient use of mesh and prims. Impose this, build that manual, collect data, write best practices… it sounds like the ideas the UN puts out… more bureaucracy and centralized control.

The Lab has the most effective free market tools already in place, Land Impact Costs. It’s in the designers face at each upload. Good design is a cheap upload. That will alert and push more designers toward good design than anything we can imagine and place in the Wiki, which no one reads anyway.

Land Impact hits people as tier. Good design has a low prim equivalence and reduced tier cost. These are huge incentives to build things efficiently and reduce the LI and prim equivalency as much as possible. Even in the old system we saw low-prim-furniture advertised and out selling high prim-furniture. Given some time we will see the same advertising incentives showing up in mesh furniture.

I think it will take longer for us to see this drive for efficiency show up in mesh clothes. The Lab has placed render cost displays in the viewer. See Advanced (Ctrl-Alt-D) -> Performance Tools -> Show Draw Weight for Avatars. For most people this is a new feature and they are unaware of it. The old Avatar Render Cost is gone. But, if you have been around awhile, you remember the ARC Nazis. They will come back and awareness of well designed mesh clothes will become an issue. People will start to pay more attention to the quality of what they buy and quality will improve.

I have a script weight scale next to my pose stand. Once dressed I check my script weight. I do occasional check my render cost. For now mesh anything tends to drive it through the roof. But, mesh is new and people are learning.

I think a large part of the criteria the Lab is using to decide the system limits is oriented toward getting new people into SL. Making the system as forgiving as possible and making it robust enough to handle poorly designed builds is part of the plan. New people are not going to be experts on their first day… year?

Bumps and Normals

We have a few select bump maps available for use in SL. You can find them in the Build Panel on the Texture tab. But we do NOT have a means of adding our own custom bump maps. Being able to do so would require adding a materials system to Second Life. There are rumors about such a system being in the works. Whatever the subject came up again in the meeting.

Geenz says a system has been proposed to Linden Lab. The proposal is joint effort by Geenz and Ash Qin. It includes custom bump maps and specular maps.

Bump maps create surface texture on objects that affect the lighting and shading effects of the object. They create the illusion of the face of a brick sticking out farther than the mortar between the bricks. The difference between bumps and normals is the illusionary offsets created by a bump can only perpendicular to the surface. Normal maps don’t have that limit. Normal maps are capable of more interesting illusions.

Geenz says there are a couple of more features in the proposal but Geenz is not willing to talk about them. Geenz says progress is being made and hopes to hear something soon and be able to make an announcement. That sounds good.

While Geenz was pressed on the issue, he didn’t say any more than progress was being made and work was being done. I know people will speculate on that. But, from different sources we know the LAB is not working on a materials system. Remember. The Lab makes a clear and specific distinction as to what working on a project means. When they have decided to do something and have assigned people to the project, they are working on it. As recently as last week (27) a Linden has said they are NOT working on it. The Lab considers many things but will usually not talk about what they are considering.

This means any work is being done by third party developers. OZ described it as, “They’ve been working on a proposal, and we’ve been looking at it very closely.” I think that confirms the status of the materials system pretty well.

Geenz speculates that we could see something from the current proposal early next year. But, maybe sooner. I am hoping the materials system in Cloud Party ups the priority in SL.

Morph Targets

If you have not worked in Blender, or whatever 3D modeling program, you have not seen the shape keys and morph targets. These are the avatar shapes use by the appearance sliders to deform the avatar. It is thought if we are allowed to connect our weight painting to the morph targets we will have a better deformer for clothes than may be available via Karl’s Mesh Deformer.

To be able to use the ‘morph targets’ some viewer changes have to be made. Also, some data specifications will need to change on the server side. The server side will need to be done by the Lindens.

I’m not sure exactly how these targets and shape keys work. But, it seems we can sculpt the mesh to make shapes keys rather than weight painting things. If that is true, we would have far more control of how things deform to shape keys than we get with weight painting.

Saying this another way… when we plan to use the Mesh Deformer we design how far above the avatar skin our mesh will float. The deformer then adjusts our mesh to keep the same distance off the skin with the appearance sliders considered. The shape and morph deforms would let us have more control over how things deform. We could change the distance our mesh is from the skin as it deforms. We could also change the direction of the deform. The mesh deformer projects along the normals extending from the avatar. A shape deform could move in additional directions.

Being able to move vertices in more directions may let us handle the skirt stretch we see. We could start sliding vertices around the legs to reduce the distortion in the limited number of skirt polys that make up the space between the legs. So, this process has possibilities.

What First

The group is going to try and decide which improvements are most desired by the SL designers and builders. With such a small group we can’t expect the user group’s decision to be representative of the entire SL Community or even the subset of builders. But, on the other hand the SL community as a whole does not understand the issues or technicalities.

Summary

The group has gotten started. We got some interesting news on possible new features. We don’t have any solid decisions or new methods we can start using. But, that may come with time. The possibilities give us beginning places.

I am of the opinion that this will be an excellent meeting place for content designers.

2 thoughts on “Content Creation Improvement UG

  1. Pingback: 2012 Reviewed – July To September » Ciaran Laval

  2. Pingback: 2012 Reviewed – The Full Monty » Ciaran Laval

Leave a Reply

Your email address will not be published. Required fields are marked *