[11:02 AM PT] Andrew Hellershanks: Once more for the recording. Hello, everyone. :)
[11:02 AM PT] Motoko.Karu @login.digiworldz.com:8002: *waves*
[11:04 AM PT] Arielle Popstar: heyas
[11:04 AM PT] Tess JL: Are we waiting for Ubit?
[11:04 AM PT] Vincent.Sylvester @hg.zetaworlds.com: No commits this week, the summer is having everyone either overheat or on vacation
[11:04 AM PT] Tess JL: Hello Arielle
[11:04 AM PT] Arielle Popstar: every1 navel gazing?
[11:05 AM PT] Andrew Hellershanks: Let me ping him.
[11:05 AM PT] Arielle Popstar: Hi Tess
[11:05 AM PT] Andrew Hellershanks: Here he comes.
[11:05 AM PT] Lyr Lobo laughs
[11:05 AM PT] Lyr Lobo: sorry, new term for my AI class
[11:05 AM PT] Andrew Hellershanks: Hello, Ubit.
[11:05 AM PT] Ubit Umarov: hi
[11:05 AM PT] Lyr Lobo: Hi Ubit
[11:06 AM PT] Andrew Hellershanks: Order of business for today, Vincent and Ubit can give us an update then Tess has something to discuss.
[11:07 AM PT] Ubit Umarov: no code changes last week :(
[11:07 AM PT] Andrew Hellershanks: Ok. That was a short update. :)
[11:07 AM PT] Arielle Popstar: so no new bugs?
[11:07 AM PT] Tess JL: No need for a sad face. It's summer and RL is important!
[11:07 AM PT] Andrew Hellershanks: I'll let Jamie get settle and then I can turn it over to Tess.
[11:07 AM PT] Andrew Hellershanks: Hello, Jamie.
[11:07 AM PT] Ubit Umarov: all the old good bugs yes
[11:08 AM PT] Lyr Lobo: Heya Jamie
[11:08 AM PT] Ubit Umarov: well some 1 week old
[11:08 AM PT] Jamie.Jordan @grid.kitely.com:8002: Hi everybody
[11:08 AM PT] Andrew Hellershanks: They aren't bugs. Just unexpected features.
[11:08 AM PT] Arielle Popstar: if only we had a mobile viewer so we could go outside and not be tethered to the desktop!
[11:09 AM PT] Andrew Hellershanks: Gavin is unable to attend today. His viewer is blocked so he is unable to enter this region. He thinks it is due to a bump in the minor version number.
[11:09 AM PT] Tess JL: I go outside to enjoy RL, the last thing I want is to birng opensim with me!
[11:09 AM PT] Arielle Popstar: pokes Joe for a viewer update
[11:09 AM PT] Andrew Hellershanks: I hope he gets it sorted soon so he can be with us again next week.
[11:09 AM PT] Joe Magarac: Hello.
[11:10 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Heh he broke that again huh
[11:10 AM PT] Orbert.Tatham @hg.zetaworlds.com: I am with Tess
[11:10 AM PT] Andrew Hellershanks: Hello, Joe.
[11:10 AM PT] Ubit Umarov: what you mean blocked?
[11:10 AM PT] Ubit Umarov: he had fixed that
[11:10 AM PT] Arielle Popstar: Osgrid blocking viewers
[11:10 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Probably just manually fixed it
[11:11 AM PT] Andrew Hellershanks: He had it fixed. He said it is probably because the minor version got bumped.
[11:11 AM PT] Ubit Umarov: he had fxed that
[11:11 AM PT] Ubit Umarov: dont' think dan made any changes
[11:11 AM PT] Joe Magarac: How does that blocking work?
[11:11 AM PT] Joe Magarac: Is that a server-side check? Of what?
[11:11 AM PT] Ubit Umarov: its is a silly string maths Joe, usind viewer channel version strings
[11:11 AM PT] Ubit Umarov: match
[11:12 AM PT] Andrew Hellershanks: I'm not sure how blocking works. Gavin can log in to the grid so it isn't a grid wide block.
[11:12 AM PT] Ubit Umarov: grid side at login
[11:12 AM PT] Arielle Popstar: [2023/05/26 05:07] dan banner: (Saved Thu May 25 22:12:56 2023)gavin used a generic name like "viewer" or something his viewer isnt banned
[2023/05/26 05:07] dan banner: (Saved Thu May 25 22:23:37 2023)but the tldr; is if he used 'Dayturn' in the name it wouldnt get blocked
[11:12 AM PT] Lyr Lobo grins at Arielle, Tess, and Orbert
[11:12 AM PT] Andrew Hellershanks: I would have thought Dan would have told Gavin that.
[11:13 AM PT] Joe Magarac: Hm, let me see what I am using. Which field in the login message?
[11:13 AM PT] Ubit Umarov: yes that was the problem a few weeks ago and gavin was here since that
[11:13 AM PT] Ubit Umarov: i just told joe, channel and version
[11:13 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Probably compiled from the git, sounded like he just edited the file last time so as to not compile again
[11:13 AM PT] Vincent.Sylvester @hg.zetaworlds.com: We match via regex now I think, for the names at least
[11:14 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Joe, yours reports as libclient 1.0 from what I can tell in my logs
[11:15 AM PT] Joe Magarac: You mean right now? This is Firestorm.
[11:15 AM PT] Vincent.Sylvester @hg.zetaworlds.com: When you had Orbert test it
[11:16 AM PT] Andrew Hellershanks: I tried to relay the comment about viewer name to Gavin. I'm not sure the message was sent to him. If he reads the log he will see that and I hope that helps.
[11:16 AM PT] Ubit Umarov: gavin was here last week after that ban issue
[11:16 AM PT] Joe Magarac: I'm sending the UUID for "viewer digest" borrowed from the Python demo.
[11:16 AM PT] Ubit Umarov: joe is not udp is LOGIN
[11:17 AM PT] Andrew Hellershanks: I know. But the version number was changed so the unban must still be tied to a specific version of viewer.
[11:17 AM PT] Ubit Umarov: where you also send MAC, ID0 etc
[11:17 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Mac you should also hash, not send in plaintext, that's another thing I saw in there
[11:18 AM PT] Andrew Hellershanks: Tess had something for us that may take a bit of explanation. Let's hold off on more discussion of viewer versions and blocks for the moment.
[11:18 AM PT] Joe Magarac: Right. I send lmajor, minor, patch build, version, and channel all derived from Github version info.
[11:18 AM PT] Tess JL: Oh, no you make me nervous, Andrew!
[11:18 AM PT] Andrew Hellershanks waits for a final comment from Ubit.
[11:19 AM PT] Andrew Hellershanks: Tess, don't mean to. Just want to keep us to one discussion at a time.
[11:19 AM PT] Joe Magarac: None of those resemble anything other viewers send. They're just the real bild info.
[11:19 AM PT] Ubit Umarov: if gavin adds Dayturn to the viewer channel info, think ther will be no match on current osgrid bans,
[11:19 AM PT] Ubit Umarov: ofc i can't see those.. only Dan
[11:20 AM PT] Andrew Hellershanks nods
[11:20 AM PT] Andrew Hellershanks: I tried to pass that info on to Dan but it said it was unable to send an instant message to him.
[11:20 AM PT] Joe Magarac: Banning is based on MAC, id0, and IP, as I understand it.
[11:20 AM PT] Arielle Popstar: Darkstorm one can spoof those things but it still is blocked
[11:20 AM PT] Andrew Hellershanks: Go ahead Tess.
[11:20 AM PT] Joe Magarac: Yes.
[11:20 AM PT] Tess JL: Ok
[11:21 AM PT] Ubit Umarov: all those 3 "channel version"
[11:21 AM PT] Tess JL waits a bit longer just in case they haven't finished yet
[11:22 AM PT] Andrew Hellershanks: I think they are. I don't see any signs of typing in progress from anyone.
[11:22 AM PT] Tess JL: I may be wasting your time here, not because what I want to bring up isn't important but becasue it may not be possible. But Andrew thought it was worth discussing at least
[11:23 AM PT] Tess JL: I love nature scenes with lush vegetation in opensim and I know I'm not the only one. But as we all know, that usually means a lot of lag. Mesh plants are often *very* heavy both on server, client and the connection between them
[11:23 AM PT] Tess JL: I don't know if Ubit remembers but a year or two he and I had a chat about vegetation on opensim and he said something that has been on my mind ever since. He said he wished people only used system vegetation since mesh plants are so heavy on the server and bandwidth.
[11:24 AM PT] Tess JL: He was absolutely right of course. Vegetation is a serious cause of all kinds of lag in Second Life and on opensim. And not only here, Unity's online manual actually mentions specifically how a single poorly made tree can lag an entire scene down to a standstill.
[11:24 AM PT] Tess JL: But
[11:24 AM PT] Tess JL: people aren't going to revert to system plants of course for several reasons. One is that most of the textures used are old and outdated, another is that the 26 different system plant variants we have aren't nearly enough to cover all the variety we need for the countless different environments that exist across the hypergrid. A third reason is that many of the plant shapes are also rather crude and oudated by today's standard.
[11:24 AM PT] Tess JL: There are also two reasons you may not be aware of.
[11:24 AM PT] Tess JL: (Am I posting too fast btw?)
[11:25 AM PT] Cuga.Rajal @rajal.org:9000: doing fine
[11:25 AM PT] Tess JL: One is that system vegetation on opensim probably is illegal. Crappy as they are, there is still somebody who owns the IP rights for the textures. It's not Linden Lab, they bought licenses for those textures from somebody else, and I find it very hard to believe that somebody has paid for the rights to use them on opensim.
[11:25 AM PT] Arielle Popstar: ok for me
[11:25 AM PT] Ubit Umarov: ( joe orbert is here with yr viewer and it sends libclient 0.1.0 like vicent told )
[11:25 AM PT] Tess JL: The last reason is that although system vegetation is light on the server and the bandwidth, it's horrendously inefficient on the client side.
[11:25 AM PT] Cuga.Rajal @rajal.org:9000: I have some feedback as a content creator
[11:26 AM PT] Tess JL: Ok?
[11:26 AM PT] Cuga.Rajal @rajal.org:9000: I think you hit it onthe head with the side comment on how the asset was made
[11:26 AM PT] Cuga.Rajal @rajal.org:9000: there ius a LOT of bad mesh out there
[11:26 AM PT] Tess JL: There is and I'll show one example soon
[11:27 AM PT] Tess JL: But first look at the pictures behind Andrew
[11:27 AM PT] Cuga.Rajal @rajal.org:9000: plants are not intrisically different than fuyrniture, fish, buildings, etc
[11:27 AM PT] Cuga.Rajal @rajal.org:9000: it has to with how they are made
[11:27 AM PT] Tess JL: To the left is a patch of system grass, you know it well already. 128 triangles. That's not much of course but all you get out of them are 32 flat sheets with a picture of some grass on them. Not very useful and not at all up to the visual standards people expect today.
[11:27 AM PT] Cuga.Rajal @rajal.org:9000: all content will increase bandwidth
[11:27 AM PT] Tess JL: To the right is 128 triangles worth of decent mesh grass. This is a quickie and could probably made a bit more out of it but I think it shows what I mean. Of course it's possible you want and need a bunch of scattered plants rather than a dense patch but if you do, you have to invest more geometry in them. Single flat sheets in the foreground simply don't cut it today.
[11:28 AM PT] Cuga.Rajal @rajal.org:9000: being conscious of the truiangles and visual accuracy while you are building is the best you can do
[11:28 AM PT] Cuga.Rajal @rajal.org:9000: on the platform
[11:28 AM PT] Tess JL: Another example, a 720(!) tri system plumeria and a 28 tri mesh holly
[11:28 AM PT] Cuga.Rajal @rajal.org:9000: you are mentioning the right things
[11:28 AM PT] Cuga.Rajal @rajal.org:9000: its up to builders to do it right, or best possible on platform
[11:29 AM PT] Vincent.Sylvester @hg.zetaworlds.com: I have a region that has about 8000 trees on it, mesh trees, but they are only 4 different meshes, once that's downloaded it shouldn't be a performance problem on the server side of things. Only on the viewer rendering all that stuff, but that's what LOD is for and object culling
[11:29 AM PT] Vincent.Sylvester @hg.zetaworlds.com: The only thing the system plants can do is they are special items that can interface with some module to plant themselves
[11:29 AM PT] Tess JL: Which one would you like? If it's the holly, grab a copy of it!
[11:30 AM PT] Tess JL: Reusing assets certainly help, Vincent
[11:30 AM PT] Arielle Popstar: is there a Best Practices wiki to point out what is?
[11:30 AM PT] Tess JL: There should be
[11:31 AM PT] Tess JL: Look at these three trees as a final example
[11:31 AM PT] Tess JL: To the left, the "Cypress 2" system tree. Calling it a cypress would bring any botanist to tears but that's beside the point. Species confusion aside, it's a decent midground/background filler tree even today.
[11:31 AM PT] Andrew Hellershanks: Arielle, there is a wiki page about creating and uploading mesh.
[11:31 AM PT] Tess JL: BUT IT'S 28,512 F***ING TRIANGLES!!!
[11:32 AM PT] Misterblue Waves: are these trees that mimic the SL embedded trees?
[11:32 AM PT] Cuga.Rajal @rajal.org:9000: There are examples out there of better use of trianbgles for decent visual display
[11:32 AM PT] Andrew Hellershanks: Vincent, I do recall some Tree module that was either to help planting trees, or could make them grow and change over time.
[11:32 AM PT] Tess JL: To put that into perspective, the n1120 trees 256x256 m forest I rezzed behind the screen is only a bit over 16,000
[11:33 AM PT] Cuga.Rajal @rajal.org:9000: so these are "system" trees that you developed
[11:33 AM PT] Tess JL: That's right Cuga. The not-a-cypress is an extreme case of poorly optimised system vegetation but they're all bad
[11:33 AM PT] Misterblue Waves: depending on the viewer, it will do instancing
[11:33 AM PT] Tess JL: No, that forest is just plain mesh
[11:33 AM PT] Cuga.Rajal @rajal.org:9000: ah ok
[11:33 AM PT] Cuga.Rajal @rajal.org:9000: what are you proposing?
[11:33 AM PT] Cuga.Rajal @rajal.org:9000: or just demoing
[11:34 AM PT] Arielle Popstar: get better content from SL?
[11:34 AM PT] Ubit Umarov: well system vegetation is mostly viewer side
[11:35 AM PT] Misterblue Waves: when using a VR headset, the trees/plants made from plains stand out and look odd
[11:35 AM PT] Cuga.Rajal @rajal.org:9000: you might lower LI by unlinking
[11:35 AM PT] Tess JL: The tree in the center picture is one of those garbage meshes Cuga mentioned. It's probably made with Tree[d] and is one of the plants the Moles used for the new Bellisseria Linden Homes. It's crap as I said but still only 1082 tris, much less than the system not-a-cypress
[11:35 AM PT] Ubit Umarov: not even sure we do have any of the textures is uses on dbs
[11:35 AM PT] Ubit Umarov: and viewers do it programaticly...
[11:35 AM PT] Tess JL: The tree to the right is a properly optimised one, 245 tris
[11:35 AM PT] Ubit Umarov: each vegetalve is defined by set of parameters
[11:35 AM PT] Joe Magarac: Those are the open simulator textures for Pine 1.
[11:36 AM PT] Tess JL: I can't say who made it since it would be rude to promote my opwn builds here ;-)
[11:36 AM PT] Ubit Umarov: then base on those parameters the viewer generates meshes etc
[11:36 AM PT] Tess JL: The system vegetation textures are all ported directly from SL, Joe with the same UUIDs and all
[11:36 AM PT] Joe Magarac: I noticed.
[11:36 AM PT] Tess JL: Anyway, fixing, or at least reducing, vegetation lag would be a huge improvement to opensim and I firmly believe it's one of the tasks that are crucial to solve before we can claim the platform is up to today's standards. The question is, can we do it?
[11:37 AM PT] Tess JL: I can think of two ways. One is to inform people about the consequences of poorly optimised mesh and how and where they can find better alternatives to those objects that lag down their sims. Polylist meshes will always be heavy on the bandwidth of course but as I showed you, it is perfectly possible to reduce the problem to a fraction of what it is today.
[11:37 AM PT] Ubit Umarov: each system vegetable is like 24 bytes on region communications...
[11:37 AM PT] Tess JL: The other solution is to develop a better system for procedural plant generation. I can think of at least three ways to do that.
[11:37 AM PT] Tess JL: Either solution would be very hard to implement and no matter how you look at them they only get harder. But it would be such a huge improvement to opensim I think it would be worth it if it is possible at all. What do you think?
[11:37 AM PT] Ubit Umarov: wilhe a normsl prim is like 512 meshes etc...
[11:37 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Telling people about mesh problems doesn't just apply to trees, that's an issue with all mesh
[11:38 AM PT] Tess JL: Yes, Ubit that's the one great thing about system vegetation but as it works now, it has so many disadvantages peopel aren't using it anyway., so that's something we need to fix
[11:38 AM PT] Cuga.Rajal @rajal.org:9000: most people don;t care and desperately want content for their region. They grab whatever looks good.
[11:38 AM PT] Arielle Popstar: well Opensim is comprised of amatuers and once they do get good enough, they go sell on kitely or s/l
[11:38 AM PT] Ubit Umarov: well viewers should improve the code and its model
[11:38 AM PT] Misterblue Waves: OS could send optimized mesh trees/bushes for the SL embedded trees
[11:38 AM PT] Tess JL: I'd go for both solutions: more and better info about how to make proper mesh and how to control lag on your sim and also a seriously updated system plant system
[11:39 AM PT] Cuga.Rajal @rajal.org:9000: would be great of Opensim could take a leap above SL with better system plants. But it's a question of who has the time
[11:39 AM PT] Ubit Umarov: sadly ll likes the mesh business...
[11:39 AM PT] Misterblue Waves: that would take more bandwidth but, if we can find out how the viewers do instancing, they might be able to be less load on the viewer
[11:39 AM PT] Joe Magarac: There are some easy things that could be done.
[11:39 AM PT] Vincent.Sylvester @hg.zetaworlds.com: I'm not sure if we really need to put more on the wiki beyond the fact that at the end of the day OpenSim doesn't use super computers to do anything special, it's still just as much a normal 3d renderer so all the conventions for creating content for videogames apply to it as well
[11:39 AM PT] Joe Magarac: First, materials.
[11:40 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Following those practices generally yields good content
[11:40 AM PT] Tess JL: I like it too. I should have been selfish enough not to bring this up since I make better optimised mesh trees than any other content creator on opensim or in SL. But even my plants aren't really efficient enough
[11:40 AM PT] Joe Magarac: Existing system trees have impostors.
[11:41 AM PT] Arielle Popstar: you sell them Tess?
[11:41 AM PT] Joe Magarac: That's the lower right quadrant of hte image I put up near the entrance.
[11:41 AM PT] Tess JL: A few. I never get around to updating my stores
[11:41 AM PT] Joe Magarac: Sharpview is currently just displaying those.
[11:41 AM PT] Joe Magarac: LIke this.
[11:41 AM PT] Joe Magarac: https://content.invisioncic.com/Mseclife/monthly_2023_06/frostartrees.thumb.jpg.737d7dde96bd1ccba80b525e66fa062a.jpg
[11:42 AM PT] Tess JL: System trees also have seriously crappy LOD. With that level of escessive triangle count they have to to be usable at all
[11:42 AM PT] Joe Magarac: Until you're closer than about 5-10 meters, impostor trees look OK.
[11:42 AM PT] Joe Magarac: Sometimes better than the generated ones.
[11:42 AM PT] Vincent.Sylvester @hg.zetaworlds.com: That would be a move forward, because I think those impostors don't actually work as they should in most viewers, at least I recall large vars with them causing single digit fps when they should be just like particles at those distances
[11:42 AM PT] Joe Magarac: For some reason, "Palm 1" seems to have the leaf pattern from "Pine 1".
[11:43 AM PT] Tess JL: I think the impostor function was diabled long ago, Vincent
[11:43 AM PT] Joe Magarac: I think so too. I can't get it to happen in SL with Firestorm.
[11:43 AM PT] Lyr Lobo: Is this a viewer problem, Tess?
[11:43 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Either that or it just broke one day and no one noticed, wouldn't be the first thing
[11:44 AM PT] Tess JL: Any solution will have to be implemented both server and client side, Lyr. That's another complication of course
[11:44 AM PT] Ubit Umarov: joe do you added system vegetation to your viewer already?
[11:44 AM PT] Joe Magarac: The generated lower LODs for system trees are awful. Trunks become pointy.
[11:44 AM PT] Joe Magarac: I have it in test but have not released it yet.
[11:45 AM PT] Joe Magarac: It just uses the impostor images.
[11:45 AM PT] Tess JL: The trunks are reduced to flat billboards even at mid LOD, Joe. They look like thin lines when seen from the wrong angle
[11:45 AM PT] Vincent.Sylvester @hg.zetaworlds.com: As Ubit said they are just parameters, remember everything you see is rendered by the viewer, OpenSim just sends data. So technically, yeah you could replace the system tree meshes if you can figure out how to make some that work with parameters that get sent to the viewer
[11:46 AM PT] Joe Magarac: Don't they have two crossed planes?
[11:46 AM PT] Tess JL: But can I repeat my main question: is this even something we can fix? It's not a small task and we have to accept we have very limited development resources
[11:46 AM PT] Tess JL: No Joe, jsut a single plane
[11:46 AM PT] Joe Magarac: Oh, I implemented it with two crossed planes and that looks better.
[11:47 AM PT] Arielle Popstar: sounds similar to the talk a few years ago about updating the system avatar but that turned out to be a non starter
[11:47 AM PT] Joe Magarac: At least straight trunks.
[11:47 AM PT] Vincent.Sylvester @hg.zetaworlds.com: The impostors should behave like particles, always facing the camera, that's some flag that gets applied to the faces iirc
[11:47 AM PT] Joe Magarac: (for) straight trunks.
[11:47 AM PT] Arielle Popstar: were even a couple who were working on it
[11:47 AM PT] Tess JL: Ada Radius is working on it now but it's slow going
[11:47 AM PT] Andrew Hellershanks: Arielle, It starts with a discussion but someone still has to step forward to do the work required.
[11:48 AM PT] Ubit Umarov: it has some conceptual relation to system avatars..
[11:48 AM PT] Arielle Popstar: thats been going a few years now then
[11:48 AM PT] Tess JL: I haven't done any serious math or programming for decades so my skills there are too rusty and outdated to be useful. If they weren't, I would have done this myself
[11:48 AM PT] Ubit Umarov: and smoe problem... LL loves the mesh business
[11:48 AM PT] Vincent.Sylvester @hg.zetaworlds.com: To answer your question, pretty sure OpenSim doesn't send meshes for the system trees, they are in the viewer, we just send the parameters for what they should look like. Replacing the meshes should be possible if you can figure out the mesh format
[11:48 AM PT] Ubit Umarov: so will not invest on improving those things
[11:48 AM PT] Joe Magarac: I think there is an easy fix here.
[11:49 AM PT] Joe Magarac: The tree texturs need materials.
[11:49 AM PT] Tess JL: Ubit, if we find a solution to the vegetation problem or the avatar issue for that matter, it will not be compatible with SL
[11:49 AM PT] Cuga.Rajal @rajal.org:9000: but might be better'
[11:49 AM PT] Tess JL: No Joe. Nice as mateerials could be, it's not even on top ten on the prioprity list
[11:49 AM PT] Arielle Popstar: well firestorm has a specific version for us
[11:49 AM PT] Ubit Umarov: well guess avatar problem is a bit worse.. btu both viewer side mostly
[11:49 AM PT] Joe Magarac: Then bark looks better, leaves can be shiny when appropriate, and the trunk becomes rounded in the impostor.
[11:50 AM PT] Joe Magarac: It's backwards compatible; viewers not doing this just get flat-shading.
[11:50 AM PT] Cuga.Rajal @rajal.org:9000: fixes are being worked on =so fingers crossed
[11:50 AM PT] Joe Magarac: ?
[11:50 AM PT] Ubit Umarov: ( dont realy remember where the textures for system trees do came from.. think buried on all viewers? )
[11:51 AM PT] Vincent.Sylvester @hg.zetaworlds.com: You can supply your own models for each LOD level so you could make your own impostors as well, might even be possible to define for them to always face the camera, though not sure collada let's you do that
[11:51 AM PT] Arielle Popstar: its not that the talent isnt in Opensim for it but it getting everyone to workl on the same page
[11:51 AM PT] Joe Magarac: It's a file called "trees.xml".
[11:51 AM PT] Joe Magarac: In viewers.
[11:51 AM PT] Tess JL: Joe, there are 26 different system plants available, that's it. To be a useful resource we need thousands of species and we need to get the load they cause down - down on *all* parts of the rnder chain
[11:51 AM PT] Tess JL: Those two are only two of the many issues that are far mroe important than materials
[11:52 AM PT] Ubit Umarov: looking to trees.xml :)
[11:52 AM PT] Joe Magarac: If you're looking at trees.xml, it's the parameters for an L-system.
[11:52 AM PT] Ubit Umarov: ok we do have the textures.. somewhere
[11:53 AM PT] Tess JL: Of course! If not, they wouldn't have been rendered here ;-)
[11:53 AM PT] Vincent.Sylvester @hg.zetaworlds.com: We have the impostors, not the actual texture map that gets slapped on the tree mesh
[11:53 AM PT] Tess JL: The textures are stored server side
[11:54 AM PT] Joe Magarac: It's a texture atlas. Look near the door where I put up a panel.
[11:54 AM PT] Joe Magarac: Left side is bark.
[11:54 AM PT] Joe Magarac: Upper right is leaves.
[11:54 AM PT] Joe Magarac: Lower right is impostor.
[11:54 AM PT] Tess JL: The impostor texture and the others are all mapped onto the same texture atlas, Vincnet. Joe showed us an example here
[11:54 AM PT] Joe Magarac: That's it.
[11:54 AM PT] Joe Magarac: Only one leaf texture. It's really simple.
[11:54 AM PT] Tess JL: Oh, it's still up
[11:54 AM PT] Ubit Umarov: my box there has one ot the system trees
[11:55 AM PT] Ubit Umarov: ohh that that at right behind andrew
[11:55 AM PT] Tess JL: The trees.xml and grass.xml files only have the UUIDs for the textures. The textures themselves are stored on servers jsut lika all other textures
[11:55 AM PT] Joe Magarac: Right.
[11:55 AM PT] Joe Magarac: but they have no materials.
[11:56 AM PT] Joe Magarac: If those texture images had matching normals and speculars, they'd look better.
[11:56 AM PT] Tess JL: Hardly. You can't polish a turd
[11:56 AM PT] Vincent.Sylvester @hg.zetaworlds.com: Which they technically don't need to be. If you want to rework that stuff it makes sense to put them into the viewer along with the meshes, downloading content that isn't going to change anyways, because it's not exposed to users makes little sense
[11:56 AM PT] Lyr Lobo: Hehe
[11:56 AM PT] Joe Magarac: The impostor images are not that bad at distance.
[11:57 AM PT] Vincent.Sylvester @hg.zetaworlds.com: I'd say start with pulling a system tree apart figure out how they are setup and go from there
[11:57 AM PT] Joe Magarac: I have some images on Discord, and you'll be able to try it in Sharpview soon.
[11:57 AM PT] Joe Magarac: A system tree is a table of values in "trees.xml"
[11:57 AM PT] Tess JL: Yes, Vincent but the only way to get enough variety of plants to cover all the needs of opensim is to make it possible for users to create and distribute their own just like we can do with prim builds. That means no client-side sotrage
[11:58 AM PT] Ubit Umarov: yeha found some on our assets :)
[11:58 AM PT] Joe Magarac: do we want more system trees?
[11:58 AM PT] Ubit Umarov: we have them for ages.. supposed CC license
[11:58 AM PT] Joe Magarac: System trees are a hack from when SL didn't have mesh or even sculpts.
[11:58 AM PT] Orbert.Tatham @hg.zetaworlds.com: Could we simply distribute new trees.xml and grass.xml files?
[11:59 AM PT] Ubit Umarov: Joe think tess thing is more about quality
[11:59 AM PT] Tess JL: It depends on what you mean, Joe. No, we do not want more of the current system trees, we want less of them. But yes, we want more procedural trees and plants of all kinds ith low streaming cost *and* low render cost
[11:59 AM PT] Ubit Umarov: not enougt Orbert.Tatham because there is the code that does generate the things
[12:00 PM PT] Ubit Umarov: not sure it can do better quality
[12:00 PM PT] Tess JL: It's also about variety, Ubit. I mean, think of all the different environments we have here. 26 different types of plants is jsut a drop in the ocean
[12:00 PM PT] Lyr Lobo: Hi Andrew
[12:00 PM PT] Tess JL: WB Andrew
[12:00 PM PT] Andrew Hellershanks: Hi. My viewer went bye bye for a moment. :)
[12:00 PM PT] Joe Magarac: I'm not sure procedural trees are worth it. SL uses procedural trees because 1) they're kind of like prims 2) there was no mesh back then, and 3) they can move with the wind.
[12:00 PM PT] Joe Magarac: 3) has been turned off in most viewers.
[12:00 PM PT] Kayaker Magic: I've got to run, RL calls.
[12:01 PM PT] Andrew Hellershanks: ok, Kayaker. See you next week.
[12:01 PM PT] Tess JL: It's possible to improve the system trees by replacing the textures with better ones. Some of them are actually cutouts from 19th century hand colored drawings
[12:01 PM PT] Orbert.Tatham @hg.zetaworlds.com: Peace, Kayaker
[12:01 PM PT] Lyr Lobo: bye Kayaker
[12:01 PM PT] Motoko.Karu @login.digiworldz.com:8002: cya kayaker
[12:01 PM PT] Joe Magarac: Right. I'm pretty sure that "Palm 1" is one of those.
[12:01 PM PT] Ubit Umarov: well replace textures we can do...
[12:02 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Pretty sure we don't filter the incoming data on those either way, so if the viewer defines new types then they should just end up stored on the database.
[12:02 PM PT] Tess JL: Not unless you work with the viewer developers, Ubit. Remember the UUIDs are stored client side
[12:02 PM PT] Ubit Umarov: ( but osme viewers may include some on their basic cache )
[12:03 PM PT] Vincent.Sylvester @hg.zetaworlds.com: The thing is that us adding even more stuff to send to the viewer for things that are meant to be the same across clients is kinda pointless unless you specifically want the ability to change textures of trees
[12:03 PM PT] Ubit Umarov: yes but we can change the contents of the assets..
[12:03 PM PT] Tess JL: Creating new textures with the same UUIDs is not an option. That would cause havoc for those who sue the same viewer and cache here and in SL
[12:03 PM PT] Ubit Umarov: ( ofc should avoid .. :p )
[12:03 PM PT] Orbert.Tatham @hg.zetaworlds.com: Ok, question: I know there was research done on "genetic" plant production, where they used fractal algorithms and parameterization to produce some pretty realistic plants. That would have to be viewer side in all likelihood, but it seemed to be pretty space- and data-efficient.
[12:03 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Adding the textures and additional data directly to the viewer so it just needs to render x tree at coordinates seems easier to implement and reduces the data overhead
[12:03 PM PT] Ubit Umarov: yeah but we do have a few cases
[12:03 PM PT] Tess JL: THank you Orbert, that's something I completely forgot to mention!
[12:04 PM PT] Joe Magarac: I'd like to have materials for the existing textures, and then update "trees.xml". If materials aren't present, drop back to the current flat texures. Viewers that don't support this get the flat textures.
[12:04 PM PT] Ubit Umarov: and if you where right about texture rights we would need to have our own with same uuid
[12:04 PM PT] Tess JL: Take a look at this:
[12:04 PM PT] Ubit Umarov: like we do have sounds..
[12:04 PM PT] Tess JL: https://www.youtube.com/watch?v=ZMVz3oUCagw
[12:04 PM PT] Tess JL: Or if you want to run it on your own computer: https://benchmark.unigine.com/valley
[12:04 PM PT] Cuga.Rajal @rajal.org:9000: might want to use sprites if there are many more trees
[12:04 PM PT] Tess JL: This is all procedural vegetation
[12:05 PM PT] Tess JL: It's amazingly light on the client. My copmuter can easily run at 300 fps there
[12:05 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Like for some reason we also keep sending the blank texture over for the viewer to download and cache, which is so damn pointless. A known piece of content that is meant to be the same for everyone should just be on the viewer end. Sure we can add overrides, but then that should be forcefully pushed, otherwise viewers should assume it's the same thing
[12:05 PM PT] Tess JL: and it's gorgeous
[12:06 PM PT] Ubit Umarov: what is yr gpu?
[12:06 PM PT] Joe Magarac: Yes, The Valley (2013) is a great demo. You can download that program from Unreal and run it.
[12:06 PM PT] Joe Magarac: Most machines which can run SL can run The Valley.
[12:06 PM PT] Tess JL: Just a plain GeForce RTX 3090, Ubit
[12:07 PM PT] Ubit Umarov: lol "just"
[12:07 PM PT] Arielle Popstar: hehe
[12:07 PM PT] Lyr Lobo grins
[12:07 PM PT] Ubit Umarov: that was the gmaing top 2 years ago or so
[12:07 PM PT] Andrew Hellershanks: RTX 3090?? wow.
[12:07 PM PT] Tess JL: I've actually been to places in SL where the vegetation lag was high enough my gpu couldn't even manage 2 fps
[12:08 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I somewhat doubt that was solely lack of gpu power, the renderer isn't exactly pushing before it falls apart
[12:08 PM PT] Joe Magarac: That's why I'd like to use impostors more.
[12:08 PM PT] Motoko.Karu @login.digiworldz.com:8002: hehe... still chugging along w/ a GTX950
[12:08 PM PT] Joe Magarac: A huge number of instances of the same texture is cheap.
[12:09 PM PT] Cuga.Rajal @rajal.org:9000: same can be said with well-produced mesh
[12:09 PM PT] Ubit Umarov: well sl target is still to run on CPU embeded GPUS
[12:09 PM PT] Lyr Lobo: RTX 3080 here with 128GB of RAM
[12:09 PM PT] Lyr Lobo: not on the gpu, on my pc *grins*
[12:10 PM PT] Arielle Popstar: i'm getting 200 fps on my 1070 while here too
[12:11 PM PT] Ubit Umarov: im geting 60
[12:11 PM PT] Motoko.Karu @login.digiworldz.com:8002: 45-50 here
[12:11 PM PT] Ubit Umarov: because it makes no sense to have more than that
[12:11 PM PT] Ubit Umarov: so i do set FS to limit it to 60
[12:11 PM PT] Andrew Hellershanks: I'mproving visual quality is definitely worth while. Additional variety would be good but a limit would need to be placed on how many gets added. Perhaps x number per type of region (underwater, desert, tropics, etc.)
[12:12 PM PT] Tess JL: I've limited my fps to 60 here
[12:12 PM PT] Andrew Hellershanks: We don't want to increase the standard library to an excessive amount.
[12:12 PM PT] Ubit Umarov: yeah no point of having more than the monitor vertical freq
[12:12 PM PT] Lyr Lobo: ah, sorry to have to run. Have a great day!
[12:12 PM PT] Tess JL: What I'm thinking of, Andrew, is something similar in principle to prims. A fixed system with multiple user configurable paramteres
[12:12 PM PT] Ubit Umarov: we are not high action game..
[12:12 PM PT] Andrew Hellershanks: ok, Lry. Thanks for coming.
[12:13 PM PT] Joe Magarac: We will be.
[12:13 PM PT] Joe Magarac: Gotta go, another meeting.
[12:13 PM PT] Ubit Umarov: ofc can also use vsync..
[12:13 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Beyond the textures that's all viewer side though and technically it's not that hard either to look for the textures on viewer end first before asking the server for them. I don't think it makes sense to send all that from our end, when all users are meant to see it the same way anyways
[12:13 PM PT] Ubit Umarov: that will mean 60fps in mmy case
[12:13 PM PT] Andrew Hellershanks: ok, Joe. See you again next week.
[12:13 PM PT] Ubit Umarov: of fs limitation is better
[12:13 PM PT] Orbert.Tatham @hg.zetaworlds.com: Peace, Joe
[12:13 PM PT] Vincent.Sylvester @hg.zetaworlds.com: What would be a good start is just documenting how the current trees are built though, that's useful information to have regardless
[12:14 PM PT] Andrew Hellershanks: True, Vincent.
[12:14 PM PT] Tess JL: https://omrg.org/w/index.php?title=System_vegetation
[12:14 PM PT] Ubit Umarov: if i do select other program the fps goes down to 20
[12:14 PM PT] Tess JL: Still a lot of holes to fill in but I'm working on it there
[12:14 PM PT] Vincent.Sylvester @hg.zetaworlds.com: We already have a similar system for the script syntax, use local and then whatever the server sends, if server sends nothing you just get what the viewer ships with, which is fine for most
[12:14 PM PT] Ubit Umarov: gpu power use here with 60 fps is about 90Watt
[12:15 PM PT] Ubit Umarov: unselecting fs, goes down to 35Watt
[12:15 PM PT] Ubit Umarov: ( rxt 3070 ti with UNDERvoltage :P )
[12:15 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Think by default FS has some internal limit to 80 or 90 fps anyways. Ideally you want it to match or exceed the refresh rate of your monitor. I get 70 here, but with a refresh rate of 120 I see screen tearing unless I force vsync
[12:16 PM PT] Vincent.Sylvester @hg.zetaworlds.com: 90 fps and 90 Hz is minimum recommended for VR these days, but higher is better for motion sickness avoidance
[12:16 PM PT] Ubit Umarov: let me see fs without limits
[12:16 PM PT] Ubit Umarov: 200fps, 150Watt
[12:16 PM PT] Ubit Umarov: see the nonsense?
[12:18 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I'm fine with anything above 60 means at least half the time I get a full frame every two refreshes, so long as those numbers match to not render in between vsync doesn't have to work as hard. Of course there are also things like Gsync and freesync to sync the refresh rate to the fps, which avoids screen tearing too
[12:18 PM PT] Andrew Hellershanks: I think the best card I had was an GT 8600 until it died. The performance of the GTX9500 and, now, GTX1050, don't seem to be as good. It is likely the viewers are wanting/expecting more which could be part of the issue.
[12:19 PM PT] Ubit Umarov: vincent.Sylvester there is dual buffering if nto tripple
[12:19 PM PT] Andrew Hellershanks: How do we break down the tasks involved with improving the stock system flora and fauna?
[12:19 PM PT] Vincent.Sylvester @hg.zetaworlds.com: In regards to mesh though, to get back on topic. The thing that needs to be hammered into people is that at the end of the day all the principles for making videogame content apply here as well. Proper LOD, conservative geometry and employing tricks to use textures to mask lack of geometry etc. You can do a lot with very little if you know how
[12:20 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Andrew, first would be documenting how that stuff currently works so that one might be able to attempt to edit or create their own version
[12:20 PM PT] Arielle Popstar: if you know how
[12:20 PM PT] Tess JL: Could we bring it up again later? Lots of people have left already
[12:20 PM PT] Orbert.Tatham @hg.zetaworlds.com: Tess does seem to be making some headway on the docs
[12:20 PM PT] Andrew Hellershanks: Vincent, that assumes that people know what are the principles for making video game content. Not everyone one will.
[12:20 PM PT] Andrew Hellershanks: Most won't
[12:20 PM PT] Arielle Popstar: or cant
[12:21 PM PT] Selby.Evans @grid.kitely.com:8002: bye all
[12:21 PM PT] Jamie.Jordan @grid.kitely.com:8002: Have a great week yall
[12:21 PM PT] Andrew Hellershanks: Bye, Selby.
[12:21 PM PT] Motoko.Karu @login.digiworldz.com:8002: bye Selby
[12:21 PM PT] Tess JL: I'm only working on documenting the existing system tree system, Orbert and to be honest, I do not really believe it's suitable as the basis for an upgraded one. It's probably better to build a brand new one from scratch
[12:21 PM PT] Orbert.Tatham @hg.zetaworlds.com: That is why I brought up the fractal stuff
[12:22 PM PT] Ubit Umarov: ok rl calls
[12:22 PM PT] Ubit Umarov: cya ppl
[12:22 PM PT] Orbert.Tatham @hg.zetaworlds.com: Laterz, Ubit
[12:22 PM PT] Cuga.Rajal @rajal.org:9000: tc Ubit
[12:22 PM PT] Motoko.Karu @login.digiworldz.com:8002: cya ubit
[12:22 PM PT] Vincent.Sylvester @hg.zetaworlds.com: You have to remember we have very little checking on the server end for data ingress, if you define a tree with ten times the parameters, multiple textures etc. it'll likely just enter it to the database as normal
[12:23 PM PT] Andrew Hellershanks: A new one would be better as the one we have now was mainly inherited from SL and LL.
[12:23 PM PT] Vincent.Sylvester @hg.zetaworlds.com: The point of that tree system or any system should be to reduce the load on the server end ideally, hence not having to send meshes and textures over, rather just parameters as we do now, since it is assumed everyone should see the same tree anyways
[12:23 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Especially since with hypergrid if a grid overrides things and you go somewhere else suddenly all the trees are unicorns
[12:24 PM PT] Cuga.Rajal @rajal.org:9000: I need to head out. Take care everybody
[12:24 PM PT] Vincent.Sylvester @hg.zetaworlds.com: I'd try to just replace one of the current trees with another one, as you would do with any other video game mod for starters, see what can be built from there
[12:24 PM PT] Orbert.Tatham @hg.zetaworlds.com: Peace Cuga
[12:24 PM PT] Tess JL: Exactly, Vincent. But how about something roughly similar to prims, but with shapes and twist parameters more suitable for plants?
[12:25 PM PT] Vincent.Sylvester @hg.zetaworlds.com: Like I said, the tree data probably has no specific serialization unique to it so whatever the viewer sends just ends up in the database anyways
[12:25 PM PT] Misterblue Waves: see ya, Cuga
[12:25 PM PT] Andrew Hellershanks: It would be nice to have newbies starting out having better items to use as a starting point when they begin to build places for themselves if we can
[12:25 PM PT] Andrew Hellershanks: Bye, Cuga.
[12:25 PM PT] Ubit Umarov: ok again.. rl calls
[12:25 PM PT] Ubit Umarov: have fun ppç
[12:26 PM PT] Ubit Umarov: and ppl
[12:26 PM PT] Andrew Hellershanks: ok. I think we will have to table this discussion for this week. We can visit it again next.
[12:26 PM PT] Tess JL: Ok
[12:26 PM PT] Tess JL: Have a great week everybody!
[12:26 PM PT] Vincent.Sylvester @hg.zetaworlds.com: There is always IRC
[12:27 PM PT] Andrew Hellershanks: Tess, IRC and the mailing lists.
[12:27 PM PT] Andrew Hellershanks: You too, Tess.
[12:27 PM PT] Orbert.Tatham @hg.zetaworlds.com: OMDG discord
[12:27 PM PT] Tess JL: Is mIRC still the client to use? I haven't been on IRC for ages
[12:27 PM PT] Vincent.Sylvester @hg.zetaworlds.com: That or hexchat yes
[12:27 PM PT] Andrew Hellershanks: On that note, I will draw this meeting to a close. Not many people are still here. :)
[12:27 PM PT] Tess JL: okie
[12:27 PM PT] Andrew Hellershanks: Thank you all for coming. See you again next week.
[12:27 PM PT] Vincent.Sylvester @hg.zetaworlds.com: We are on libera now, all that info is on the wiki
