This week the CC UG met in ADITI, the Preview Grid. The regions with the name Animesh(?) have the Beta server software supporting animesh running. So, you get to see some actual animesh in action.
You’ll also see the problem of animation appearing to speed up when the camera is far away.
This week Medhue was having problems with the Internet and could not stream. So, his usual video is not available. I was hoping to catch some footage of animesh, so I was recording. So, you get mine.
I will try a bit different format this week. Let me know in the comments if it is better, more useful, or not.
Vir is working on getting Animesh ready for release. Currently solving the problem of getting animesh to work correctly when another type of prim is the root prim.
Silent Mole, she, is the one to answer questions about the content. I have no idea how to email her or any mole. I suppose a note card via IM.
Question The only way a cat can wear a shirt or hat is if it is in a linkset with the animesh. So, a permissions problem comes up. I’m wondering off Carrie’s wording of the question so it makes sense to me.
If a cat-animesh is No-Mod, how does one add or remove an accessory? As it is now a hat or shirt would have to be part of the original No-Mod animesh. So, we are into an ATTACHEMENTS problem, as animesh attachments are planned to be only via linkset.
Carrrie brings up the point if animesh accessories have to be in the primary linkset (think the cat), the 20k polygon limit for the entire animesh linkset (cat root, added hat, shirt, etc.) is a severe limit. But, that brings up more questions, like is the 20k poly limit enforced at link time? Rez time?
Several more questions are being asked in the Animesh thread: Project Animesh Feedback Thread
I think Piscine Mackenzie and this post are what is talked about in the video. Piscine explains well why being able to edit the linkset via script is so important for animesh use. It is the practical way to add something like a hat to the cat.
Vir explains the intent is to use linksets to avoid the overhead added to support agents (avatars) and to avoid the multiple mesh layers we see in avatar bodies now.
Providing the attach and detach ability will add a lot of overhead to animesh. Vir thinks it unlikely that the feature will be added. Certainly not in this phase one. But, he acknowledges they may end up doing that.
He hopes there are enough features to manage linksets via scripting.
Vir reiterates that the current 20k triangle limit is a beginning limit while data is collected. Vir he says they will not revisit the limit until AFTER the Lab has collected performance information.
Piscine goes on to explain how limits in what we can do script-wise with linksets are limited by the No-Mod permission. To be able to modify the linkset the item has to OK-Mod. This creates a number of security issues for content theft.
Once an item is OK-Mod people can add scripts into it. So, scripts could be added to hack the pet management system and circumvent a creator’s food system.
This conversation has spun off into a JIRA BUG-139168 – Feature Request – Mod Keys. An environment where secure, remote no-mod object manipulation can be achieved. This is about adding a feature to prims that would prevent modifying via the viewer and ALLOW modifying by a script. The idea seems to be the creator with full-perm could set the state and modify the linkset. The creator’s scripts could modify the linkset. The end users would not be able to modify the item except via use of the creator’s scripts.
This seems to be a practical solution with less overhead than adding attachment abilities to animesh. But, we’ll have to see how the eggheads see this and find out next week.
There is a problem of animesh not casting shadows. You’ll see that in my video. Avatars are casting shadows but not the animesh prims. Vir explains it is a known bug. They just have to dig into the render engine to find out why the shadow maker is ignoring animesh. The same process used for avatar shadows should work for animesh.
Once shadows are working we will see how much of a performance impact they make.
You’ll see the little dog being crazy in my video. It is Aki Shichiroji’s dog. She tossed the raptor script and animations in it just to play with animesh and pathfinding.
Vir talks about the 200LI limit. He is not changing it until they have good data. If there is a problem with it and testing, then he is willing to talk about it. Otherwise, it stays until they know what they are doing.
Medhue related his experience running 100 animesh objects with pathfinding. Very light load.
Vir talks about how animesh will provide way better performance than bots. He expects most of the cost to be from the poly count.
At 23:00 Carrie takes the discussion back to animesh/linkset attachments. As a reader here you got more information above than she had when asking the question.
You may have noticed that right-click-drop doesn’t work with mesh. So, a follow-on would be will it work with animesh? The question came up in Vir’s mind and he has looked into it. Seems there is no server-side code to support dropping mesh objects.
Since this is part of an old system no one has messed with in a long time, it is unlikely to change. As it is NOT a quick fix it won’t be part of the animesh project. Vir isn’t saying it will never be fixed… but don’t hold your breath.
Bakes On, as sometimes called, is in progress. But, it won’t be ready by the time Animesh I releases.
It will allow adding a shirt to a cat, like shirts can be added to the classic avatar as a texture.
It will be Animesh II or Animesh NPC that will address the Bake On features. It appears this is a bigger project than I thought. Meaning it is some ways out in the future. In the video, Vir explains the problems they are having to handle. Many of the design answers are dependent on how animesh works and as of now parts of that are yet to be decided. It may be fair to think of animesh as it is now as more alpha than beta project.
Another interesting question is whether or not we can sit on animesh?
As Dictatorshop (avatar name) points out demonstrating furniture often requires two people, a partner or victim O.O If we could use an animesh avatar it would be great. The way furniture is made, it would be easy if we could sit an animesh on a pose ball or similar.
Vir explains that sitting is a special case in the SL system that only works for avatars. So, adding that ability adds overhead and programming work.
Most of the animation visualization for animesh in viewer side. There is nothing server side that would let the server handle ‘sitting’ animesh.
So, the current solution would be to have a root object that positions the animesh so it looks right for the animation. While the animesh can use the same animation as an avatar, sitting changes the avatar’s parent. There is no way to do that for animesh objects. Meaning no way now in the system. So, the furniture’s sit-position will have no effect on animesh. The ‘position’ will have to be handled manually, scripted.
We may well use a basic prim at the animesh root and position and rotate it and then play an animation. This would likely be the simplest way to fake a sit.
Cathy Foil suggested making the demo-dummy part of the furniture. So, the dummy would be animated by the couch’s controls. This too is a possibility. I think it adds a huge amount of work for building demos. But, it may be the best overall solution as the change of ever sitting an animesh is remote.
A big part of the problem is in how AV Sitter and NPOSE work. I expect the heavy lifting for animesh-demo-dummies is going to fall into the laps of these animation control people.
There was a question about whether attachments for animesh would be considered if linksets don’t work out. Vir provides the answer I think one would expect. ‘Anything is possible.’ But, his goal is to figure out what people want to do, the easiest way to do that, and understand why some things don’t work and figure out if they can be fixed.
You may have found that non-rigged mesh is not visible as part of an animesh object. Vir is working on correcting that. So, at some point, the non-rigged mesh will be visible.
Will llCastRay be updated to handle animesh? Vir thinks it a reasonable idea to have an animesh filter for llCastRay. But, as yet has not worked on it.
llCastRay is the scripting function we use for a script to detect things in SL. It is used to replace object bullets and sometimes in place of Pathfinding to allow an object to ‘see’.
Alexa Linden spoke about using JIRA feature requests to get more eyes on a subject/issue. It is the best way to reach the maintenance team.
Rider Linden updated us on EEP, the changing Windlight feature. He hasn’t done much this week as he got pulled to work on something else. I suspect an exploit patch. But, the Lindens never say.
However, he has almost completed the Day-Cycle integration.
Example Animesh Rezzed, see video at 51 minutes in.
The idea is animesh will teleport with the avatar. I presume it will have t be attached to the avatar. As it is right now, teleporting from a region with animesh support to one without support is a problem. Of course, that problem goes away when all regions have animesh support.
If you move out of an animesh region the script will error and stop. If you move back into an animesh region you have to manually restart the script.
I am guessing there won’t be a special case scenario for animesh roll out. As in some other rollouts, the viewers without animesh support will likely be updated to not crash when seeing animesh. This can happen long before animesh is finalized. It isn’t that hard to say ignore this thing.
Firestorm and Linden viewers are reported to crash in certain instances when animesh is in view. For now, this can only happen in ADITI.
You may find you cannot comment on feature requests and bug reports. You can email LetMeIn@lindenLab.com and request access. You can also mention access to the Wiki editing.
There is no meeting November 2nd. Conflicts with an internal Linden staff meeting.
The video after 60:00 is a discussion after the Lindens leave. I think about the only thing of interest in the section not in the main meeting is how important something being able to sit animesh will be. I think it would make a lot less work for users. I think it would make a LOT more work for the Lindens to program and delay animesh release. Plus driving up the cost of animesh.
But, it is really way too early to be making those calls.
00:00 – Vir updates status of the project.
01:15 – Silent Mole commended for creating the animesh content being provided in Animesh4.
01:45 – Meeting open for questions.
01:50 – Carrie Tatsu question on animesh linksets.
03:00 – Reference to the post on linkset accessories problem.
04:00 – Vir explaining the intent of their development.
05:30 – Vir talks about 20k polygon limit.
07:26 – Part of post placed in chat.
08:50 – Vir asks if No-trans is good enough. Response follows. TL:DR – No.
16:00 – Discussion of linkset problems ends here.
16:36 – Problem of shadows… lack of.
18:00 – Doggy rapture & 200 LI thing – Cost of animesh shadows…
20:00 – Medhue talks about running 100 animesh with pathfinding.
23:00 – Carrie asked how the Lab is planning for animesh to be used with accessories.
24:45 – Will right-click-drop be added to animesh?
27:40 – Is baking animesh textures still in progress?
31:30 – Can we sit on animesh discussion starts.
37:40 – Is there any possibility we can have attachments if linkset don’t work out?
39:50 – Cathy Foil, add test demo dummy as part of the furniture.
41:30 – Will non-rigged mesh be visible in animesh linksets?
43:35 – Will llCastRay get a filter type for animesh?
45:00 – Alexa Linden on the benefit of JIRA
47:30 – Rider Linden on EEP
51:35 – Elephant animesh being rezzed.
52:20 – Animesh and teleport.
60:00 – Meeting ends