SecondLife "roadmap"

[Rohacan Hirons]

Hi,

I'm a bit curios, where SecondLife is going on ? Why the hell technical improvements for creations are often throw to the trash ?

• New mesh LI calculation to improve LODs creation and rendering cost ?
• Animesh position/rotation offset ?
• Animesh attachment points ?
• Animesh resizing ?
• Grid wide experiences ?
• Ability to click on animesh to finally remove those invisible "helper" prims ?
• Animations syncing allowing obviously to sync animations but also to start animations with a time offset ?
• Hierarchy of links instead of root/children only ?
• ... and a lot of other one maybe even more important than these ones (feel free to complete the list)

I know there are works in progress in many different aspect of SecondLife (rendering, better UI, marketplace, viewer perfs, ....) but right now when I look at what's coming next I'm thinking "Ok, soon even poo in SecondLife will be shiny". While I agree that visual improvements are nice and not useless at all, I would like to know when will we be able to create better stuff, when devs will have time to come back on those kind of abandoned / never started improvements? I'm tired waiting for new features in a random other new project viewers/servers that finally disappear from the roadmap everytime someone at the lab has a new idea.

What do you think ? What is for you the better future for SecondLife ?

1. A beautiful world where you can do a few things
2. A world where you can do a lot of things even if sometime it's awful

[animats]

4 hours ago, Rohacan Hirons said:

• New mesh LI calculation to improve LODs creation and rendering cost? (Nobody could come up with an LI charging scheme which both improved things and didn't cause existing objects to increase in LI. If any rezzed object increases in LI, some parcels will be over-filled and items will be deleted. Long discussions at Creator User Group last year. )
• Animesh position/rotation offset/Animesh attachment points/Animesh resizing? (All that is for followers, I expect. Animesh were originally intended as standalone characters, not body parts. That you can attach them was an unplanned feature.)
• Grid wide experiences? (Works, but not allowed to users. Too much griefing potential.)
• Ability to click on animesh to finally remove those invisible "helper" prims? (Not clear on the problem.)
• Animations syncing allowing obviously to sync animations but also to start animations with a time offset? (More LSL control over animation has been proposed. The trouble is, the info that tells viewers to run animations is in UDP messages, and changing that format, even a little, breaks all existing viewers. Everybody has to upgrade simultaneously. But maybe you could do something via the new puppetering system, which has a completely different data channel.)
• Hierarchy of links instead of root/children only? (Recently "accepted" in JIRA, might happen someday.)

 

[Rohacan Hirons]

Hi, thanks your reply.

Quote

Nobody could come up with an LI charging scheme which both improved things and didn't cause existing objects to increase in LI

There is a feature actually that can update LI on a single click and works very nice... animesh. We only have to check on "Animesh" and voila, the LI changes. Does it sound crazy to think about one more option "Use LI calculation v2" ?

Quote

The trouble is, the info that tells viewers to run animations is in UDP messages, and changing that format, even a little, breaks all existing viewers

I'm not sure the hardest part is the way messages reach the viewer, everything (almost everything?) in SL was in UDP, now we have HTTP messages, CDN, ... add new LSL commands, clone the code related to animations, update it to use the time variable (maybe the hardest part, no?) Sure viewers have to be updated to execute new feature. Was it an issue when sculpt prim arrives ? Was it an issue with mesh ? Animesh ? Bake on mesh ?

Quote

Ability to click on animesh to finally remove those invisible "helper" prims? (Not clear on the problem.)

Sorry, I was talking about the invisible prims that creators have to add to rigged mesh (not only animesh) to allow users to click on their items and alt-click to zoom on it instead of clicking on whatever is behind the object. The viewer code contains the necessary lines to allow users to click on rigged mesh but this make FPS to drop drastically. I wrote some code modification to work around this FPS drop (my compiled viewer is still functional and I had no issue with this since late 2020 / early 2021, I wish all viewers can have this, and we can finally remove these invisible prims)

 

Edit :

Quote

All that is for followers, I expect. Animesh were originally intended as standalone characters, not body parts. That you can attach them was an unplanned feature

Ok, then I can only hope this will finally be a thing. Some people worked hard to imagine and suggest how these features could be implemented, what was the good or bad points of those ideas, ..

Quote

Hierarchy of links instead of root/children only? (Recently "accepted" in JIRA, might happen someday.)

Yay, one point! Hope it's not a "possible future work for a future project"

Edited September 5, 2022 by Rohacan Hirons

[Henri Beauchamp]

2 hours ago, Rohacan Hirons said:

There is a feature actually that can update LI on a single click and works very nice... animesh. We only have to check on ”Animesh” and voila, the LI changes. Does it sound crazy to think about one more option ”Use LI calculation v2” ?

This is a different mechanism. The switch method you cite is from the legacy (primitives count based) accounting to the Land Impact (LI) accounting (and it can as well be achieved by setting the prim as ”convex hull” instead of ”primitive”, or adding a material to one of its faces, or changing the Alpha method of a textured face for alpha masking, for example): be careful however, for with some old builds with tortured prims, this could cause the LI of the whole link-set to skyrocket and the object to get returned because of a parcel prim allocation overshoot (so better doing this in a sandbox sim).

Suddenly recalculating the LI of all existing objects would be both intricate (only rezzed objects could be recalculated, not the ones in the inventory database) and cause spurious auto-returns that would ruin entire region builds.

Switching to a second LI calculation based on a manual action by a mod-ok permission holding user, while doable, would also be tricky as it would require a new flag to be set on each primitive (also meaning issues with compatibility for old viewers build tools), and I do not see why a user or creator would voluntarily degrade the LI of their builds, if the old LI system gives them a better ”value”...

However, it looks to me like LL already modified the LI calculation for newly uploaded meshes (i.e. the LI calculated at upload time for any given mesh object): I think that the lowest LOD needs to be much simpler, now, or the LI quickly increases... A Linden reading this could perhaps confirm it or not.

[Wulfie Reanimator]

2 hours ago, Rohacan Hirons said:

I wrote some code modification to work around this FPS drop (my compiled viewer is still functional and I had no issue with this since late 2020 / early 2021

Do you have a commit?

[Henri Beauchamp]

3 hours ago, Rohacan Hirons said:

The viewer code contains the necessary lines to allow users to click on rigged mesh but this make FPS to drop drastically. I wrote some code modification to work around this FPS drop (my compiled viewer is still functional and I had no issue with this since late 2020 / early 2021, I wish all viewers can have this, and we can finally remove these invisible prims)

I think this has been largely addressed in the ”performance viewer” code changes (there are way less bounding boxes recalculations for rigged meshes when ”picking” them)... I certainly do not see any FPS drop change, here (with the Cool VL Viewer) when I enable the ”Pick rigged meshes” option.

10 minutes ago, Wulfie Reanimator said:

Do you have a commit?

I am curious about those changes too...

[Rohacan Hirons]

About clicking on rigged items here is the JIRA : https://jira.secondlife.com/browse/BUG-229845

LLab was about to include changes into a viewer, then suddenly went backwards

Quote

I certainly do not see any FPS drop change, here (with the Cool VL Viewer) when I enable the ”Pick rigged meshes” option.

If I remember well, this setting is shunted in multiple places in the code, then no matter if you set it on/false in your viewer option it has no effect. That's why in first place I had to compile my own viewer, then I learned about the FPS drop issue and the fact that someone decides it was better to shunt it (why... ? I don't know. Setting it to false by default whould have been enough I think). When I started to work with RezzRoom I was really needing this feature because we often have to manipulate animesh, click on them, alt-zoom on it, etc.. often without those "helper" prims, and even with them it was painful. Then I wrote a workaround, shared the code and add some fixes when I found some special cases where there was still FPS drop.

Quote

I do not see why a user or creator would voluntarily degrade the LI of their builds, if the old LI system gives them a better ”value”...

I don't understand where this can be an issue if you have the choice. If I'm a creator and upload a new mesh I have two options :

1. I do as I'm used to, create my LODs (or let the viewer do it, probably set them to High/High/0/0 xD), upload the mesh.
2. I know there is another way to create meshes with another way to calculate the LI, I create LODs thinking about this fact, upload my mesh, drop it without carrying about the big LI I will see because I know I only have to activate an option and voila, my LI is OK.

Fun (or not) fact: I made an ambulance few years ago, when LLab started to talk about these new calculation rules that will apply to animesh in the first place then for all meshes, I made it with this in mind, I was so proud of myself, talking about this to everyone, talking to creators around me about what was coming, about how they will have to change their methods.... 2, 3 years (?) later this never happened

 

To come back on the main purpose of this thread (at least for me xD), I'm curious, what people think about the latest and incoming improvements of SecondLife ? Am I the only one thinking the "game" will be a bit more optimized and looks better (for those with a decent graphic card) but that's all, no way to improve our creations ? WE WANT MORE POWER !!

Edited September 5, 2022 by Rohacan Hirons

[Quistess Alpha]

5 minutes ago, Rohacan Hirons said:

I don't understand where this can be an issue if you have the choice. If I'm a creator and upload a new mesh I have two options :

The more complex/high-poly/"bad" a mesh is, the more Li it "should" have. The "point" of the Li system is to incentivize creators to make simpler/lower-poly/"good" mesh, which is not as simple as just pushing the "make my mesh good" button, it requires extra skilled labor.

The 'problem' is that it's possible to 'abuse' the current system by making the lowest LOD (or in some cases lowest 2) a single triangle, then tell their customers to "fix" their LOD settings so their fancy stuff doesn't disappear when viewed from more than a few meters away.

You can't have your cake and eat it too, If you 'cut out the loophole' creators and their customers will be angry the Li went up. If you leave the loophole in, no creators are incentivized to change their workflow, which is the whole point of making an new Li system.

[Rohacan Hirons]

Quote

The more complex/high-poly/"bad" a mesh is, the more Li it "should" have. The "point" of the Li system is to incentivize creators to make simpler/lower-poly/"good" mesh, which is not as simple as just pushing the "make my mesh good" button, it requires extra skilled labor.

The 'problem' is that it's possible to 'abuse' the current system by making the lowest LOD (or in some cases lowest 2) a single triangle, then tell their customers to "fix" their LOD settings so their fancy stuff doesn't disappear when viewed from more than a few meters away.

I totally understand this agree, that's why I was really happy with this incoming change and so sad that a few years later this looks abandoned.

Quote

You can't have your cake and eat it too, If you 'cut out the loophole' creators and their customers will be angry the Li went up. If you leave the loophole in, no creators are incentivized to change their workflow, which is the whole point of making an new Li system.

I still think if more and more creators adapt their workflow to create better mesh and make use of this to promote their brand like "Build for new LI, optimize the world, better rendering, save your CG, ..." a kind of "new" business will arrive, forcing all creators to do better, then players will by themself drop their "old ***** stuff" to the trash. For sure creators will have to work hard to create nice LODs with limited LI. And if we're a dreamer, we can even think that if everything goes well, this massive optimization may allow a higher prim count limit in the future. Yes, I'm a dreamer

Edit: Of course, I don't think we need a higher prim count limit, people often rezz too many stuff already, everything is overcharged, region are laggy, ... but I think it can help players to prefer a 3 LI nicely built chair instead of a table + 4 chairs + 1 shelf and 1 carpet for only 1 LI with poor LODs.

Edited September 5, 2022 by Rohacan Hirons