Work on Animated Mesh announced.

[Lucia Nightfire]

I don't know if there will be an official thread from Linden Lab that will allow comments, but it has been stated that they plan on making an official announcement about this feature in the future. If it will be before or after release is anyone's guess, heh.

[Fluffy Sharkfin]

5 hours ago, Medhue Simoni said:

We already have impostors for rezzed meshes. They are called LODs. These are going to be really important to animated meshes. 

Avatar impostors and LODs aren't exactly the same thing though, which is why we have LOD models for rigged attachments to reduce the complexity of the 3D avatars over distance and impostors to limit the number of 3D avatars we see on screen at any one time.  LODs are still simplified 3D models whereas impostors are 2D sprite representations of 3D models which are, one would assume, far less resource intensive.

 As for LODs being really important to animated meshes I'm sure we can all agree that, for responsible content creators, LODs are really important for all meshes not just the animated ones (in fact, depending on how they handle land impact for animated mesh, I'd say they're still more important for worn rigged-mesh since attachments aren't governed by land impact at all).

[HarrisonMcKenzie]

7 hours ago, Love Zhaoying said:

*raises hand* I know! I know! More client-side lag!! 

How exactly will it cause more client-side lag? Most items will have simple animations, and are going to have less script use and fewer and more optimized textures. Instead of a script running every step of the way and causing client lag because it is hogging resources and switching textures constantly, you'll have a lighter script calling an animation that causes far less lag.

Again, we aren't talking about full avatars with hundreds of lengthy and detailed animations. We're talking about objects being able to be animated by a single animation via systems we already have in place. We're not talking about full immersion NPCs with expressions, eye movements, and complex AIs that cycle between hundreds of animations. We're talking about a few frame walking cycle for a dog. That's going to create far less lag.

[Lucia Nightfire]

Personally I don't think land impact should need anything special other than a minimum value, sort of how pathfinding characters are done, but with legacy prim land impact being honored instead.

With pathfinding chars, legacy land impact is not honored, meaning, a 255 link linkset or 25000 land impact, tortured materials legacy prim will become a 15 land impact character with llCreateCharacter(). Mesh land impact is honored, so a 900 land impact mesh pf char will still be 900 land impact with llCreateCharacter().

So with max([32 or 64,actual land impact]) that only increases with prim torture, material changes or size changes like any normal linkset, I don't see anything else special is needed.

A minimum is the only thing needed to curve abuse as far as rezzed animated mesh is concerned. That is to prevent rezzing thousands of 1 land impact meshes for tiny critters like bugs or flowers, etc.

With attached animated mesh, there will have to be something else done, whether it be a hard limit on # of attached animated mesh allowed or a # of slots consumed by a single animated mesh.

Edited May 17, 2017 by Lucia Nightfire

[Fluffy Sharkfin]

15 minutes ago, Lucia Nightfire said:

Personally I don't think land impact should need anything special other than a minimum value, sort of how pathfinding characters are done, but with legacy prim land impact being honored instead.

How exactly would legacy prim land impact work with animated mesh when its mesh, not prims?

57 minutes ago, Lucia Nightfire said:

So with min([32 or 64,actual land impact]) that only increases with prim torture, material changes or size changes like any normal linkset, I don't see anything else special is needed.

A minimum is the only thing needed to curve abuse as far as rezzed animated mesh is concerned. That is to prevent rezzing thousands of 1 land impact meshes for tiny critters like bugs or flowers, etc.

I think keeping the existing formula for calculating mesh land impact and just increasing it based on the number of bones the mesh contains would be a good solution, so that simple animated objects that only require a couple of bones to perform smooth animations like doors, curtains, rocking chairs, etc. would still have relatively low land impact compared to say a fully animated mesh character or creature.

There are a lot of potential uses for animated mesh that have nothing to do with NPCs, it would be a shame if we lost sight of that and nerfed other forms of content with unnecessarily high land impact, etc because we were focusing solely on the most advanced possible application (especially since we don't have any idea what sort of impact animated mesh has on performance yet).

[Lucia Nightfire]

4 hours ago, Fluffy Sharkfin said:

How exactly would legacy prim land impact work with animated mesh when its mesh, not prims?

Animated mesh is still an object linkset that has a root link and will allow linking and delinking.

Legacy prims will perform just as they do with any linkset. They just won't be apart of any animations since those are only rendered with mesh links that have rigging data.

This will allow ease of building scenes, settings, buildings, furniture, vehicles, etc. and combining them with animated mesh, especially with existing content. It will allow packaging them for rezzers as well.

Animated mesh is not an only-rigged-links exclusive.

As far as land impact based on bones actually being used, wouldn't the animation itself dictate what bones are "dead" or "freewheeling" and not the skeleton. As far as we know there will be no skeleton customization offered.

[Fluffy Sharkfin]

1 hour ago, Lucia Nightfire said:

Animated mesh is still an object linkset that has a root link and will allow linking and delinking.

Legacy prims will perform just as they do with any linkset. They just won't be apart of any animations since those are only rendered with mesh links that have rigging data.

This will allow ease of building scenes, settings, buildings, furniture, vehicles, etc. and combining them with animated mesh, especially with existing content. It will allow packaging them for rezzers as well.

Animated mesh is not an only-rigged-links exclusive.

Ah, you were referring to having actual legacy prims as part of the linkset, I understand now, sorry.

1 hour ago, Lucia Nightfire said:

As far as land impact based on bones actually being used, wouldn't the animation itself dictate what bones are "dead" or "freewheeling" and not the skeleton. As far as we know there will be no skeleton customization offered.

Since the introduction of bento the number of bones in the rig exceeded the max limit for mesh uploads so they changed things to allow mesh to be uploaded with only the bones to which it's rigged, afaik you still have to include the entire skeleton when uploading bento animations (animations without the "Hip" bone as the root don't seem to work), it's just that any bone which doesn't have transformations applied to it by the animation isn't affected when it's running so is "freewheeling".  I'd assume that it would be easier to determine how many bones are being used in an animated mesh by looking at how many bones a mesh is rigged to rather than how many bones in the animation have transformations applied to them.  

The other reason for using the mesh to calculate the land impact rather than the animation would be that since we'll be able to add multiple animations to animated meshes and trigger them via script it would mean the LI for those meshes would change depending on which animation is currently playing and I'd imagine that would cause all sorts of problems.

Since they were also recently discussing the ability to trigger multiple animations on avatars, it will be interesting to see if they do the same for animated mesh so we can have one animation playing to control one part of the mesh and trigger a different animation to move other bones which are rigged to a different area of the same mesh (or perhaps another mesh in the same linkset).

Edited May 18, 2017 by Fluffy Sharkfin

[Lucia Nightfire]

6 hours ago, Fluffy Sharkfin said:

Since they were also recently discussing the ability to trigger multiple animations on avatars, it will be interesting to see if they do the same for animated mesh so we can have one animation playing to control one part of the mesh and trigger a different animation to move other bones which are rigged to a different area of the same mesh (or perhaps another mesh in the same linkset).

Afaik, supplemental animations are only going to be used with new animation override functions. You might want to confirm that at the next Content Creation meeting.

[Fluffy Sharkfin]

33 minutes ago, Lucia Nightfire said:

Afaik, supplemental animations are only going to be used with new animation override functions. You might want to confirm that at the next Content Creation meeting.

Ah, thanks for the clarification.  While I'd heard mention of it at a couple of the meetings I didn't realize it was only going to be implemented for use in the new AO function.  It's a pity though since being able to simultaneously run multiple animations on different bones within the same rig would allow us to add a much wider range of movement to a single animated mesh using fewer animations instead of having to create a unique animation for every possible combination of motions the object could perform

[Medhue Simoni]

 

[Macrocosm Draegonne]

This is exciting 

Could this help make elevators, escalators,  moving platforms, obstacles, etc, more smoothly animated too?

A bit off topic, but still in this video... The officially supported windlight/environment per-parcel controls is sweet!   I love firestorm supports this already, LL doing this means we'll have a ui, and they could plug in more features, like time-cycle setups, trade, market use, ect. I hope they still allow the settings per elevation (as it is in firestorm), but smaller x-y-z sized zones than the parcel boundary could be interesting too.    Definitely remove the 5 second delay, with system default to auto-windlight transitions on.  Controlling the env lighting in smaller spaces with a zero LI impact is just too useful.  It can then be scripted and animated too, right?  >.<

[Medhue Simoni]

2 hours ago, Macrocosm Draegonne said:

Could this help make elevators, escalators,  moving platforms, obstacles, etc, more smoothly animated too?

Hmmm, probably could make all those things with animated mesh, but you'd have to work out the pitfalls. I'm not sure about the physics of it all tho. Will have to see when we get our hands on it.

[Lucia Nightfire]

Other than the visual for the escalator, which you could use Animesh for. you already can use keyframe motion to have smooth movement with platforms and elevators plus when an agent is standing on it, they will move along with the movement.

http://wiki.secondlife.com/wiki/LlSetKeyframedMotion

[Morgan Rosenstar]

On 16/05/2017 at 5:02 AM, Medhue Simoni said:

I remember talking about animated objects back when bento was released. I was saying that Bento is HUGE, but if LL did animated objects, it would be even more HUGE. The info is trickling thru SL as we speak. That content creation video has 3 times more views than any of the past meetings. Even as the meeting was going on, people were hearing about it and passing the live stream around. By the end of the meeting, there were 35 people watching on the live stream, with more coming as I was ending it. Animated objects is HUGE!

Free donations to upgrade my PC and keep logging in after this update will be huge too.

[Macrocosm Draegonne]

On 8/27/2017 at 3:20 AM, Lucia Nightfire said:

Other than the visual for the escalator, which you could use Animesh for. you already can use keyframe motion to have smooth movement with platforms and elevators plus when an agent is standing on it, they will move along with the movement.

http://wiki.secondlife.com/wiki/LlSetKeyframedMotion

Yeah, the keyframed animation is a lot better than the old school elevators I remember, but its still jumpy for me, and has a fairly low speed limit before its crazy unstable, but even at a slow speed its still somewhat jarring, is it not for you?  I have a fairly beefy system too.   Plus, I do have the best elevator in MP, it uses the keyframe method, and its actually really great, but i have to keep it set at a slow speed.

On 8/26/2017 at 3:09 PM, Medhue Simoni said:

Hmmm, probably could make all those things with animated mesh, but you'd have to work out the pitfalls. I'm not sure about the physics of it all tho. Will have to see when we get our hands on it.

Yeah its so intriguing to me bento animates so smoothly, is it operating on the same physics setup as general world objects?

Smooth fast non-seated elevators would be so sweet, and then all the obstacle course action type objects/enemies, architectural stuff, generally any interactive moving things right?   I'm now thinking terms of bones to use for an interactive space, lol, strange headspace to be in, and probably wrongheaded too? lols... BTW I have little understanding of how this all works yet, but im digging in anyway!  >.<

[Medhue Simoni]

22 hours ago, Macrocosm Draegonne said:

Yeah its so intriguing to me bento animates so smoothly, is it operating on the same physics setup as general world objects?

That is the part that I'm not all that sure about. I'm pretty sure that a rigged mesh is not going to have the right physics to make something like an elevator. Maybe, an attachment could do it. Not sure we'll even get attachment on animesh tho. Won't really know until LL releases it. The animesh will have some kind of physics, but how malleable that physics will be I have no idea.

Edited August 29, 2017 by Medhue Simoni

[Macrocosm Draegonne]

4 hours ago, Medhue Simoni said:

That is the part that I'm not all that sure about. I'm pretty sure that a rigged mesh is not going to have the right physics to make something like an elevator. Maybe, an attachment could do it. Not sure we'll even get attachment on animesh tho. Won't really know until LL releases it. The animesh will have some kind of physics, but how malleable that physics will be I have no idea.

Yeah, it will be a nice surprise when we get more info, i'm sure it will be amazing no matter what, I was only curious about the potentials really, not making any demands or anything. 

[MoonHowler Snowpaw]

I just guess, that this will work similarly to regular avatar animations. Like with dances or gestures, where avatar offsets far from the center locally in viewer, but the true location is still the center of avatar's coordinates. So i think the objects will usually move separately, using a script, like before, animating their parts with animations. Just a guess though, based on how it's done now.

[Medhue Simoni]

16 hours ago, MoonHowler Snowpaw said:

Just a guess though, based on how it's done now.

Imagine taking an avatar, and just rezzing it on the ground, using all the same animations. That is animesh in a nutshell. 

[MoonHowler Snowpaw]

7 hours ago, Medhue Simoni said:

Imagine taking an avatar, and just rezzing it on the ground, using all the same animations. That is animesh in a nutshell. 

I know, it's obvious. Maybe i wasn't clear enough in my previous message. I was talking about elevators. That technically they'll be local and offsetting visually, if done fully by animation without some script really moving a thing. And if at all an mpc will be able to move an avatar, while it's sitting on it. In my guess an avatar will sit in air in the center of mpc's real coordinates, while the mpc will locally-visually move away from it, while the animation is playing, that's why i think it will be important to move the mpc itself with a script like a regular prim, animating only its decorative movement, so avatar will appear in the center sitting on it.

[Macrocosm Draegonne]

9 hours ago, MoonHowler Snowpaw said:

I know, it's obvious. Maybe i wasn't clear enough in my previous message. I was talking about elevators. That technically they'll be local and offsetting visually, if done fully by animation without some script really moving a thing. And if at all an mpc will be able to move an avatar, while it's sitting on it. In my guess an avatar will sit in air in the center of mpc's real coordinates, while the mpc will locally-visually move away from it, while the animation is playing, that's why i think it will be important to move the mpc itself with a script like a regular prim, animating only its decorative movement, so avatar will appear in the center sitting on it.

Yea I think I understand, that must be why seated elevators are generally smoother, the physics ones can be a bit jumpy, especially if set too fast, or there's lag.  I still prefer the non seated ones, but I'll put the added teleports in my buildings anyhow, some people hate elevators lol.  

I suppose thats a lot for the server to calculate physics on a moving object, thats also moving multiple avatars, which aren't seated on it.  I guess there is a potential here that an animated mesh attachment might be usable, but maybe not.    Perhaps elevators, trains, subways, and other such non seated movement, might need their own unique class of physics, seems like a big task (and sort of low priority), but it could be useful someday none the less. 

[Medhue Simoni]

1 hour ago, Macrocosm Draegonne said:

I suppose thats a lot for the server to calculate physics on a moving object, thats also moving multiple avatars, which aren't seated on it.

Why not have the avatar "seated"? You don't actually have to override their AO with the seated animation. You then also eliminate the need for any physics and the negative results of some aspects of it. I imagine a skeleton that is scaled up super large, to not have issues with how far the elevator bone travels. How the sitting will work tho, is unknown. Like, could we really be specific and sit on a bone? Or attachment of a bone?

[Macrocosm Draegonne]

41 minutes ago, Medhue Simoni said:

Why not have the avatar "seated"? You don't actually have to override their AO with the seated animation. You then also eliminate the need for any physics and the negative results of some aspects of it. I imagine a skeleton that is scaled up super large, to not have issues with how far the elevator bone travels. How the sitting will work tho, is unknown. Like, could we really be specific and sit on a bone? Or attachment of a bone?

In all the elevators I tested "seated" usually meant having the avatar stuck to the side of the car like a magnet on a fridge lol, and if you unsit you fall down the shaft to certain doom.   I suppose that could be any sort of "sit"  tho right?  ...they could be visually standing probably, even accounting for a max capacity and offering several standing spots.  

In the elevator I chose (the gentek dev edition), my favorite by far, you just walk in and press the button, and it goes, even with 2, 3, or more avatars standing inside, and you can move around in there too.  There's a demo in their mainstore you can try inworld, its actually excellent.  The trick I found (suggested by the maker) was to use a slower speed setting to reduce the jumpiness and glitches, but that obviously reduces the distance potential and peoples patients for ride duration.   He said it was a limitation of the physics engine and the keyframed animation technique, still it is the best elevator IMO, he really did a wonderful job on it.

Still, I think you may be correct that a fixed position would be best, performance-wise, and more flexible in distance/speed.  ....and, now I'm also really curious if we could sit on the animated mesh bits, that feature could allow for a lot of different interesting creations... now my creative mind is swirling again! lol

Edited August 31, 2017 by Macrocosm Draegonne

[Fionalein]

On 8/31/2017 at 7:45 PM, Macrocosm Draegonne said:

In all the elevators I tested "seated" usually meant having the avatar stuck to the side of the car like a magnet on a fridge lol, and if you unsit you fall down the shaft to certain doom. 

You have not been around SL a lot, have you? There are almost no sitting elevators around.

Edited September 17, 2018 by Fionalein

[Phil Deakins]

He's been around a year longer now