First meeting in a couple of weeks.
Alexa Linden announced that she will add a description to the Aditi regions (Animesh1 to 4) saying they are for animesh only. They Lindens will be returning things that are not animesh and setting an auto-return time cycle for the region.
Medhue made a video, Thank you, Medhue. My Thursday was spent fighting with Microsoft while they ran me in circles.
I’ll provide short summaries of the various topics discussed in the meeting and provide a time mark for the beginning. If you want more info than I provide check the video.
02:40 – Vir Linden tells us the Lab has hired a programmer specializing in rendering. He has worked for the Lab on Second Life™ before, Graham Linden. He is full time. So, things in the rendering pipeline will be changing. There is already a Project Viewer for a branch of viewer changes to the render pipeline. See: Second Life Project Render Viewer version 184.108.40.2062446.
He is looking first at stability and bugs. As there is no timeline for other work Vir wouldn’t go into what might be in the plan.
ARC tan? This term popped a number of times. I can’t decide if Vir is saying ARC-tan or ARC-ten…
Graham will be working on the render cost at some point. Vir started but, his priorities have been focused on animesh and ARC/ACI work is sort of only on his back, back burner. In the short term, animesh costs will have to be added in. In the longer term, Vir says they will do a deeper dive looking into all render costs.
05:15 – Anchor Linden talks about his work on the Bakes On Mesh changes. The work is planned to complete the end of first-quarter 2018. He says the 1024 textures is still on his To Do list… ?
It sounds like the 1024 part of the project will be done soon. But, I did not get the impression baking to anything other than the classic avatar is going to happen by then. But, Medhue’s specific question brought out that it will work with the mesh bodies we are using. Classic and attachments. Supposedly the need for union skin bodies will disappear.
Vir clarifies that this new service will ONLY be for avatars. As an avatar is not associated with animesh it won’t bake animesh textures. Baking for animesh is not part of the current Bakes on Mesh project.
I get the feeling Anchor is designing without really understanding how we use layers in the mesh bodies. Community feedback may be at a minimum in this project.
As best I can tell, the Bake Service has been changed to output 1024 textures in place of the 512 textures previous used. That is out… I’m unclear, but I think grid wide, but definitely on Aditi.
Anchor is adding the part that will allow Baking for attachments, like avatar mesh bodies. I get that the idea is to apply the system clothing, tats, etc. we use for classic avatars to attachments. There is a missing part for alpha layers… They haven’t looked at that or maybe haven’t figured out exactly how they do will it… Whatever, I am pretty sure we aren’t there yet.
18:00 – Good explanation of some problems with alpha and mesh bodies.
A key understanding is demonstrated by tattoos. The alpha in the tattoo makes the layer(s) below visible. An alpha layer punches a hole in the avatar and nothing is seen where the alpha is active.
23:15 – Vir explains how the Baking Service works. Too long for here. Watch.
27:15 – The problem of how users will know what works with what comes up. This is confusing. It is hard to know who, even Linden, understand what is being said. It seems the new system layers could see the same item being applied to a human, snake, or elephant. That is not going to work. Putting the same texture on different UV Maps generally never works. So, buyers have to know what goes with what.
That brings up a problem for the marketplace design in that the branding field only allows one brand. So, if I make something for Maitreya, Belleza, and Slink… I have to pick one. Sucks.
28:00 – Standard Sizes replaced by brands. Omega getting designers together to have items fit multiple brands. Good summary of mesh body attachment problems.
31:15 – An example of the high-density mesh the revised ARC/ACI hopes to eliminate. The exact same appearance could have been created with WAY fewer polygons.
Vir talks about the data collection features currently in the viewer to collect rendering cost data. Part of the effort is to get a sense of how the different video cards are handling the render.
32:00 – Oz Linden explains the Lab’s intention, to make the render cost assigned to more accurately represent actual render cost. The current costs were designed to reduce the amount of data the simulators had to transfer as a first priority. Now that the Lab has taken the simulator out of the data transfer business, the priority changes to the actual render cost.
The old equation gave an incentive to build without LoD to reduce data transfer. They plan to change that. When asked for details Oz explained they have not decided on the details as they are measuring to see what actually needs to be considered.
36:15 – When Animesh releases, it will release with a temporary cost. Later, when they do the deep dive they’ll do a more precise and representative cost.
The plan for revised Land Impact cost is gone over again 37:30. Oz explains how they will handle it to avoid massive region item returns when it goes into effect.
40:00 – Summarizes how they plan to handle the animesh land impact. Primarily based on polygon count in the detailed LoD. Low LoD models with less than half the highest and preceding LoD will not count. There will be a set cost for adding the skeleton. The idea is to encourage good use of LoD.
The base cost will not be 200 LI. No info on what it will be… but, less.
48:00 – Some confusion on uploading .anim files.
51:00 – Will there be a hard limit for worn mesh? Vir say no, as there will be a check at upload.
Will we be able to attach 2 animeshes to an avatar? For now, only one. But, they’ll look at it.
53:30 – An interesting idea of having particle-attachment-points in animesh. As it is now, we can only emit particles from the origin of a prim.
…and the end of the interesting stuff.