RL is complicating my being at the Content Creation meetings. Fortunately for me, Medhue is live-streaming and recording the meetings. The video is here: Medhue @ Content Creation. My summary follows.
Vir Linden had nothing new on viewers. He often tells us about what he is doing with a specific viewer.
This week Vir has been working on the Animated Objects. Currently the work is in the nature of proof of concept. Next week he will start to look at server side changes that will be needed.
A total trivia bit is: Vir uses a SL Birthday Bear for his animation test subject.
Nothing new has been added to LSL for this testing. Vir still expects to add only start and stop animation commands for objects.
Anticipated load, render cost and server load, is not expected to be high for animated objects. The recent addition of Bento bones made little impact on the viewer and server workloads. But, until they get to a project viewer and supporting they won’t have a clear idea of performance impacts.
I believe they have a good idea of the performance impact. The idea would be based on existing knowledge thus an informed estimate. But, what the actual impact will be is unknown. It may be a surprise or as expected. Only testing will answer the question.
The Lindens seldom venture into any area disclosing what they are planning/doing until they have a high degree of certainty that something can actually be done within acceptable limits of performance and effort. So, I tend to believe they have a high degree of confidence the animated objects feature is going to work as anticipated.
A question: what mesh can be animated? Only new stuff uploaded after the feature is added or will we be able to animate mesh items uploaded since the introduction of mesh? Vir is unsure.
Vir’s thinking for now is all existing mesh could only be used as static attachments and rezzed as static objects. To be a standalone, rezzed, animated object would require either a new upload with new attributes or they may be able to add additional settings (flags) that convert it from an attachable mesh to an animatable mesh. Vir doesn’t anticipate mesh objects being able to be used as both.
Teager would like them to be interchangeable and sites some user cases where attachable-animated mesh would be useful.
My example is imagining a demon with wings, tail, etc. riding a horse. There wouldn’t be enough bones in a single skeleton to make that work. One would need a skeleton for the horse and another skeleton for any complex avatar riding it. This thinking gets nebulous as we consider whether we ‘sit’ on the horse like we would a chair or attach it.
Vir thinks animatable attachments could have a significant impact on performance. But, nothing is set in stone.
Vir’s concern seems to be in the loads of existing code calculating where attachments are in an avatar. This would probably create lots of odd interactions for attached animated mesh. I take this to mean positioning animated mesh is unlikely to work well and be difficult to get right in any reliable way.
You may remember when Hover Height was added to first the Shape and later the avatar right-click. The later was added because there were simply too many factors for the Shape Hover Height to easily handle. For easy of use we have the right-click-avatar Hover Height.
Keeping animated attachments in the correct place is apparently considered to be the same complex calculation of interdependent factors. Just unlikely to work well.
12:05 – Vir describes of how attachment points work and animate. There is a possibility that animated objects could be added to the avatar’s root, the CoG/Center point. They talk about possibilities. But, until Vir can play with them and see how things work together, there will be no decision.
26:30 – Some want to add Shapes, those avatar shapes we use now, to animated objects. Vir says that will NOT be part of this project. The Lindens want to keep the project small so they can quickly iterate through changes to get animated objects working and rolled out.
28:30 – Teager points out user cases for scaling via animations. Consider a kitten that grows into a cat.
Vir points out that scaling opens the door for conflict between sliders and scaling animations. They were running into that with Bento scaling.
I’m thinking if sliders affect animated mesh shape, then aren’t we adding Shapes to the mesh? Thinking in early development stages is often inconsistent… or Vir may be tossing this this out to point out it would be a block to ever getting shapes for animated mesh.
Vir does think a root node scale factor for attached animated mesh might be a workable solution having some control over attachment size for an avatar size. It would scale everything at the same time.
39:00 – The other part of the project is the Supplementary Animations. Vir has not started work on those. From what I am hearing, that is likely to be weeks down the road.
Thinking about why Vir would start with Animated over Supplementary, it would seem the supplementary would be easier. So, why not start with it first? Vir may just be more excited about the animated mesh. Or he may see how animated mesh would have an impact on the supplementary animations. Thus, changing the order in which things need to be developed. No way to know at this point.
Land Impact Cost is a concern for several attending the meetings. Vir says that consideration will wait until they have a simulator working. Then testing can tell them what the viewer/server performance impact will be and what Land Impact Cost should be..
41:00 to 56:00 – Maxwell Graf talks about the problem of building taking people outside SL. So, what can get more people back inside SL? Max is thinking about a simple mesh editor being added to the viewer. He thinks such an editor would being creators back into SL and give prim builders a path to more complex building.
Vir’s answer is that such an editor is not possible with the current system. So, a simple editor would require a huge change.
Discussion on possibilities of a mesh editor or some mesh modification tool continues for about 15 minutes.
Vir points out a prim’s polygons as seen in wireframe are not the real polygons, they are only representations polygons generated from the base prim and its parametric settings. So, it would not be reasonable to try and edit those.
In some math equations, there is no way to move backward from the result to the individual components. For instance, what numbers sum to 12,160? There is a huge set of permutations. So, editing the vertex of a prim presents a similar problem. What parameter has to change to allow movement of a single vertex?
I noticed in the text chat there is mention of a problem where rezzing on mesh items can cause “inventory loss”. Seems an item with a certain type of LoD and Physics Layer will send an object rezzed on it into the valley of lost items. Poof. There is a JIRA somewhere.
As for the mesh baking project, when asked Vir said there has been no work on mesh baking.
It appears work is proceeding on adding Animated Mesh. I can’t decide if we are really in the early conceptual stages of a project that may be dropped if too costly or complicated (think performance impact or development time and manpower), which would be unusual for the Lindens. Or if they have determined this is possible and we are seeing early development stages of work normally hidden in the early stages of an announced project, I think more likely.
Either way, it will be an interesting project. I expect we will get some level of animatable mesh.