This is an interesting meeting. Vir Linden announces some of the projects the Lindens are going to be working on. I have some pre-meeting coverage and then some post-meeting footage. The video is long. I did not time compress this video. There is quite a bit going on in chat. At normal speed, it is hard enough to keep up with. So, I decided not to speed it up.
I also provided a topic index/summary of the video. You’ll find time marks in the text. [Maybe a few minutes before the video finishes uploading and becomes visible here.]
Animating Breasts – You can animate breasts. But, the animations suffer from the buoyancy and cleavage appearance sliders getting messed up.
Medhue may have a workaround. He thinks we might be able to animate the attachment points to animate breasts without breaking the sliders.
Adding More Bones – (TM 03:00-05:00) Not going to happen. The viewer is at the limit for what will work with older computers. So, no new avatar bones for at least a couple of years… assuming tech will improve enough in that time.
New Mesh Baking – Vir Linden announces the Lindens are starting work on Server Side Mesh Baking. (See TM 5:15 for announcement)
The point of this project is to remove the need for onion skin mesh bodies… A body for the skin, another for the underwear, and another for the clothes. Possibly another for tattoos. Onion skinning essentially triples or quadruples the polygon count.
With the Classic avatar body, the Server Side Baking combines the skin, tats, underwear, shirt, and jacket layers into a composite texture. The idea is to extend baking to handle mesh bodies the same way.
One of the problems is mesh bodies are starting to use materials. The baking engine does not currently handle materials. As the Classic body does not make use of materials, there is no need for the bake engine to handle materials. It is unclear whether the bake engine will be revised to handle materials or whether it will need to.
Also, the bake engine currently renders 512×512 pixel textures. That will be revised to bake 1024×1024 textures. Vir considers that a first step. Then work will start on baking mesh objects.
There are a lot of questions as to how this will be done. You’ll find then scattered throughout the meeting.
AssetHTTP – (TM 08:00) Vir says this viewer version is in place to promote to the Default Viewer. This is the viewer that will load more assets from the Content Delivery Network (CDN). Promotion depends on how well it does in Release Candidate (RC) testing.
(TM 08:37) A question is asked if the new bake-system will work with Omega appliers. I’ll tell you, yes. But, Vir seems to have an odd idea of what Omega Appliers do, as evidenced by his comments. He is clear that the bake-engine is going to only bake diffuse textures.
If that is the final case, this may not be anywhere as useful as the Lindens think. Consider. We have 4 layers with materials, meaning they can have a diffuse texture, specular map, and normal map. Those maps are not handled by the bake-engine.
Since there is only ONE mesh body (one mesh, one layer) to put the maps on, someone will have to bake those maps into a composite layer. So, we would need specular and normal composites.
We already have a number of people agitating for Modify-OK mesh bodies. But, baking specular and normal map composites is not going to help the novice or intermediate users. So… you may want to point your favorite mesh makers to this article and get them participating in the Content Creation UG meetings.
I am seeing some exotic bodies that use materials and other object properties. Limiting these bodies so that all effects must apply to the single body seems restrictive. Consider wearing tattoos that I want to glow. If I have clothes over some part of them, what will the glow do the to the clothing not intended to glow?
(TM 10:36) End of official announcements. Vir talks about starting work on the server side of animated mesh.
Number of Allowed Animations? (12:30) You may not have considered, but we are starting to load up on animations. There could be walk, face, wings, ears, eyes, and tail animations playing at the same time. Then we could add hand animations, which we have seen recently. Maybe ev3en separate left and right-hand animations.
Next, we add some animated mesh items as attachments and they have 3 or 4 animations. Think of a purse swinging over your shoulder with tassels.
So, how many animations can we hang on an avatar? 64 …that seems like a lot.
Vir thinks they will have to do some testing and see what the viewer can handle. They may have to lower that limit.
(TM 16:00) Discussion about scaling avatars down to tiny’s and up to giants. Seems there is a problem with face bones distorting the face at the extreme ends of size scaling.
Plus, scaling is limited in how small or big it can go. According to Vir that is just a factor of the slider range used. So, it could be extended. But, no one is sure what other problems that may create. A certain one is that if the existing slider were changed, it would break existing content.
JessicaAnne points out that once the avatar size passes a certain point, tiny or giant, the faces rigged to Bento bones start to distort. This is generally because the bones cannot be scaled to the desired size, the have to be shrunk/grown by an animation. (I think I got that right.)
Adding a slider for more control and a greater range of size would be easy and possible. It would PROBABLY reduce the distortion problem. (20:00) But, without resigning the UI only 2 more will fit… Really? Well, not a big problem. But, there is a 255 slider-limit because of the way sliders are implemented. Going past that limit would be a big project, I suppose because of the complexity needed to avoid breaking existing content.
Cathy Foil points out that scaling has another problem. (23:00) Motion is not affected by scale. So, a tiny takes the same size step as a human-sized avatar and a giant. A tiny zips across terrain looking like the Flash. A giant appears to be stuck in slow motion. The animations play at the right speed but lose relation to the ground and thus appear to skate or moonwalk across the ground.
The distance covered by a walking human was just programmed in without regard to scale. The height slider that scales the avatar pretty much keeps it in the range of a size that looks right with the current walking speed.
Baking Arbitrary Textures (30:00) Getting an LSL function to bake arbitrary textures it out of scope for this baking project. Baked textures are not like asset textures. Baked textures are temporary and apparently do not live in the asset database.
Baked avatar textures are short lived and expected to be. But, an arbitrary texture for an animated mesh might need to last indefinitely.
Arbitrary texture is a maybe someday project. But, discussion later in the meeting shows it has possibilities.
The idea is interesting. One could specify a set of textures to bake into a composite and place on a given face of a specific mesh. This has possibilities.
Which Textures Appear on an animated mesh NPC? (39:40) There is the possibility that any mesh could have textures baked. For an NPC there is the question of whether to use the textures from a real avatar’s mesh-body-bake on the NPC. Or be able to specify textures to bake to the NPC. Confused yet?
3rd Thing (39:40) Have the bake-service look in the animated object’s contents and use textures found there to bake a composite for the object. Speculation only. Beyond current scope.
Know the baking is for animated objects gives the system an idea of expected life. (43:00)
State of animated objects (5/25 – TM 45:00) – Prototyping viewer and server to animate an object in place. The current work is getting information to the shared experience. No limits decided. But, some limits are expected; Land Impact, ACI, or number in a region or per avatar.
BUG-20027 – Allow joint-offset-relative translations in animations. (46:50) This is an entry for user input. Check out the JIRA entry if you are into animation and custom avatars.
Changing Avatar Limits (Population – 47:25) Vir doubts there will be a need to reduce population limits based on animated objects in a region.
Limiting the number of animations played (48:50) It may be necessary to change the limit on how many animation can be played simultaneously.
There are tools that play one animation per bone… posing tools…
Medhue says the cat we see in the video is using 7 animations. You’ll find the cat for sell by Zooby in about a month. (52:00) talking about cat…
Animating with Flexi verses bones (56:11) Vir doesn’t have good numbers. The general experience in the Lab is flexis are very inefficient.
HUD linked rigged mesh (1:00:00) In the chat a question is asked about Rigged mesh that is linked to a HUD. Oddly, IMO, the mesh falls into place on the avatar just as you would expect. HOWEVER… no one can see it, other than the viewer with the HUD.
Vir explains that HUD’s live in screen space, not world space.
Vir ending meeting (1:05:00) No meeting next week due to conflict with Linden internal meeting.
Post meeting talking about baked mesh avatars.
Medhue thinks these features will happen quickly. (1:09:00) Plus others responses and thoughts about why these features will happen more quickly.
Unforeseen Problems (1:10:00) Accounting and object scale…
Interesting Trivia (1:15:00) With Bento the Lab hired a testing company to test the new bones with 300+ video cards to evaluate the impact on rendering.
Animated Mesh Limits (1:16:00)
LoD for Animation (1:21:00) This would be something like not animating fingers at a certain distance.
Medhue and Cats Take Over World (1:25:00)