Mesh/Archive/2011-10-10

From Second Life Wiki
Jump to: navigation, search

Agenda

  1. SH-2374 (Rigged Mesh Deformer), Can you give us users, developers and creators any further information on internal ideas LL is considering for implementing your own fix for this issue? If so, can you tell us whether the options and/or prototypes you may be considering to internally develop solutions for this 1) involve utilizing Havok or other proprietary code; 2) if any LL solution is implemented now or in future can you assure TPV's and open source developers that LL solution will be available to them for use as open source code and 3) are your potential fixes for this server side, client side, or both? A little transparency and communication to users goes a long way. (Jusden Jonstone)
  2. An ongoing issue of importance and priority all along has been on the stability of Mesh on main grid which I think most all users can agree is important. Can you give us a update on current or ongoing Mesh Server and Client side stability issues that are taking precedence over tackling additional mesh fixes or feature implementation? (Jusden Jonstone)
  3. Due to other SL commitments, I shall be unable to attend further Mesh OH for the time being, The transcripts are not updated regularly, could the team please address this, so myself and other interested parties can remain a part of these proceedings? [Graham Dartmouth]
  4. Mesh HQ 2 is down (as are 5 of the 8 old mesh beta regions still on the map). Where will the meeting be? (Drongle McMahon)
  5. The mesh-development build was stopped at the version of 3.0.6. But the version of beta viewer is 3.1.0. Which viewer should we test? davep_shining-fixes or snowstorm_viewer-development? [Yuzuru Jewell]
  6. The mesh-development build was stopped. With which viewer should I make the tutorial video about uploading by new dialog? Or till when should I wait? [Yuzuru Jewell]
  7. OPEN-105 update [Wolfpup Lowenhar]

Meeting Notes

Transcript for Monday October 10, 2011

[11:58] davep er... did I do that again?
[11:58] davep or are you talking about that time a few years ago?
[11:58] Vincent Nacon flooded with what?
[11:58] davep water
[11:58] Koli Contepomi Heh...
[11:58] Vincent Nacon oh that?
[11:58] Vincent Nacon muhahahaha!
[11:58] Siana Gearz i don't know, you kind of seemed proud of the fact last time you mentioned.
[11:59] davep yeah, optimizing the moving code
[11:59] davep optimized out a necessary call that set the height of the water plane
[11:59] Siana Gearz ah, just that :)
[11:59] davep yeah....
[11:59] davep whoops
[11:59] Siana Gearz such things happen to us every few days :P
[12:00] Rain Edring Hello:)
[12:00] Wolfpup will the rest of the lindens be showing up?
[12:00] davep uhh
[12:00] davep crazy expensive game grid
[12:00] Siana Gearz well, that's what your costs do :P
[12:01] Koli Contepomi Yay sim lag
[12:01] Koli Contepomi Jesus Geenz
[12:01] Siana Gearz oh that's still oldskool grid thing...
[12:01] Koli Contepomi My account is so out of date you're not my friend on it.
[12:01] Vincent Nacon yeah there's something going on in the corner of this sim
[12:01] Vincent Nacon it's rezzing bunch of prims
[12:02] Geenz Spad :>
[12:02] Wolfpup Runitai i put and update on the agenda that should make you and nyx happy
[12:02] davep yay!
[12:02] Maxwell Graf ...HOLA!!...
[12:02] davep http://wiki.secondlife.com/wiki/Content_Creation/Mesh_Import_User_Group
[12:02] davep agenda link ^^
[12:03] davep once vir gets here we'll get started
[12:03] Wolfpup ok
[12:03] Wolfpup this might evben make vir smile
[12:04] Asha Hello all :)
[12:04] Drongle McMahon Hi Vir
[12:04] Wolfpup hey vir
[12:04] Yuzuru Hello Asha!
[12:04] Yuzuru I am awake
[12:04] Asha Asha waves to Yuzuru ^^

[12:04] Asha hehe
[12:04] Asha I'm barely awake.
[12:05] davep ok, here we go
[12:05] davep first item:
[12:05] davep SH-2374 (Rigged Mesh Deformer), Can you give us users, developers and creators any further information on internal ideas LL is considering for implementing your own fix for this issue? If so, can you tell us whether the options and/or prototypes you may be considering to internally develop solutions for this 1) involve utilizing Havok or other proprietary code; 2) if any LL solution is implemented now or in future can you assure TPV's and open source developers that LL solution will be available to them for use as open source code and 3) are your potential fixes for this server side, client side, or both? A little transparency and communication to users goes a long way. (Jusden Jonstone)
[12:05] Vir Linden Howdy all
[12:06] Vir Linden We are just starting to look into it, so it's too early to give technical details.
[12:06] Motor Loon http://www.indiegogo.com/Mesh-Clothing-Parametric-Deformer-Project
[12:07] Vir Linden We are looking for something open source in all likelihood.
[12:07] davep right now the only public JIRA related to the issue is https://jira.secondlife.com/browse/SH-2324
[12:07] Wolfpup Open Source is a GOOD thing
[12:07] Vir Linden but that's the sort of decision that we would be making per-project, not something we can make a blanket policy statement on.
[12:08] Jusden Jonstone possibility of accepting or using community developers solution if it works
[12:08] Vincent Nacon I'm still wondering how would it work anyway? techincally side of it
[12:08] Geenz Spad there's a few ways it could work :>
[12:08] Vincent Nacon seem like it would be another nightmare in math
[12:09] davep whoops, that's not public yet
[12:09] davep too early to comment on the technical side
[12:09] Siana Gearz Vincent, they don't know yet either, they first grab up the dough and then they decide :)
[12:09] davep (haven't done the due dilligence yet to see if the approach will work)
[12:09] davep but we WILL be able to comment next week
[12:09] Vir Linden 3) server-side code needed? Don't know yet.
[12:09] Vincent Nacon yeah but, honestly I don't think it would work at all without manual input for each one of these "deforms"
[12:10] Vincent Nacon I'm not agaisnt it, just don't see how it could work as automatic
[12:10] Vincent Nacon maybe displacement map?
[12:10] Vincent Nacon but it wouldn't be auto fit to the avatar's
[12:11] davep we can tell you that we're NOT looking into automatic rigging
[12:11] Vincent Nacon I just want to remind everyone here that asking for this sort of feature isn't just that simple
[12:11] davep the state-of-the-art in automatic rigging still requires artist intervention and is best left to the external tool
[12:11] Geenz Spad I guess in theory, you could do some sort of retargetting thing based upon the avatar's shape verticies :o
[12:11] Wolfpup actualy if you base the deformer map of of the default join/bone map then it should
[12:11] Vincent Nacon that made no sense, Wolf
[12:12] Geenz Spad but I imagine that could be problematic when you take into consideration of having many avatars in view
[12:12] Vincent Nacon aye, it's too much to ask for something as simple it may sound
[12:12] Motor Loon Probably not any worse than avatar physics
[12:12] Geenz Spad I'd beg to differ on that :>
[12:13] Nadi Vemo hello
[12:13] Vincent Nacon worst part is dealing with skin weight after deform (somehow) is sorted
[12:13] Wolfpup it is simple you take the default bone/joint map and use it as the basis for the deformation ie stretching bones and moving joints
[12:13] Vincent Nacon I'm sorry, I think it's a bad idea to focus on this sort of feature
[12:13] Vincent Nacon we need something else
[12:13] davep yeah, we can go deeper on this next week
[12:13] Asha What something else?
[12:14] Vincent Nacon that's the same problem
[12:14] davep for now we've got an agenda to get through
[12:14] Vincent Nacon sorry, go ahead
[12:14] Wolfpup that is what SeaWolf has dome for thier large dragons
[12:14] davep An ongoing issue of importance and priority all along has been on the stability of Mesh on main grid which I think most all users can agree is important. Can you give us a update on current or ongoing Mesh Server and Client side stability issues that are taking precedence over tackling additional mesh fixes or feature implementation? (Jusden Jonstone)
[12:15] davep Most of the issues we're aware of are viewer side
[12:15] Vincent Nacon and graphic card drivers
[12:15] Vir Linden And most of the work that's taking precedence is non-mesh (we just have a couple of bug fixes still in the pipeline)
[12:15] Wolfpup you mean the rendering system in the viewer right
[12:16] davep and fixing viewer side stability is a top priority -- there are many crash fixes going into the next release, and work continues on driver compatibility/render stability in a side branch
[12:16] davep the main blocker for render stability getting into release is compatibility with GTX 460 and intel 915/945
[12:16] Vincent Nacon even 11.2+ Catalyst for ATI/AMD issues?
[12:17] davep I'm not aware of any outstanding stability issues specific to AMD hardware
[12:17] davep (that is, that haven't been addressed by the compatibility work in https://jira.secondlife.com/browse/SH-2240 )
[12:17] davep there ARE outstanding issues in release
[12:18] davep davep is running that build now and getting 35 fps with everything on.

[12:18] Koli Contepomi Koli Contepomi is also running that build.

[12:19] Nebadon Izumi what hardware davep?
[12:19] davep GTX 580
[12:19] Wolfpup i hope that work for my sustems when it coems in
[12:19] Nebadon Izumi ouch only 35?
[12:19] Maxwell Graf yay only crashed 20 times or so in a row.
[12:19] davep it's still got issues with NVIDIA 4xx series
[12:19] Nebadon Izumi hello Max
[12:19] Koli Contepomi I'm getting 22 with shadows off, but I'm CPU bound.
[12:19] Nebadon Izumi i tried TP'ing you a few times
[12:19] Nebadon Izumi i had trouble too
[12:19] Maxwell Graf insta crashy today
[12:20] Blue Alloy and I'm here on FS... the LL client won't even log in without dying horribly
[12:20] Nebadon Izumi 35fps for a 580GTX seems terrible to me
[12:20] Maxwell Graf same
[12:20] Maxwell Graf memory pool is low. lol
[12:20] davep that's with anti-aliasing, shadows, water reflections
[12:20] Nebadon Izumi still sounds horrible
[12:20] Wolfpup im on viewer-development tip and got here just fine
[12:20] Nebadon Izumi 580GTX is one of the fastest cards you can get?
[12:20] Nebadon Izumi is that the low end your aiming for with Frame rate?
[12:21] Blue Alloy I had 2 gig free... no way it should shit itself like that
[12:21] Koli Contepomi Fastest single GPU card from NVIDIA, currently.
[12:21] davep I can't say I'm proud of 35 fps as a target, but it IS an improvement over the current release viewer
[12:21] Nebadon Izumi ya i guess its not bad, but for that level of hardware, that doesnt sound great either though
[12:21] Wolfpup would love to see increased performance out of my linux laptop for the viewer
[12:22] Nebadon Izumi considering most modern games you would be getting 60fps or better with that hardware
[12:22] davep G[2011/10/10 I=z saus O
[12:22] Maxwell Graf id settle for any frame rate as long as i dont crash 20 times trying to log in.
[12:22] davep GPU-z says I'm at 40% GPU load
[12:22] davep so, cpu bound
[12:22] Nebadon Izumi what cpu?
[12:22] Wolfpup the main processor of the computer
[12:23] Koli Contepomi My GTX560 Ti is probably around the same.
[12:23] Nebadon Izumi know what cpu is he using
[12:23] Nebadon Izumi lol
[12:23] Koli Contepomi Since the fans are quiet.
[12:23] Nebadon Izumi y know what a cpu is
[12:23] davep AMD something or other
[12:23] Wolfpup Central Processing Unit
[12:23] davep ok, we should move on
[12:23] davep Due to other SL commitments, I shall be unable to attend further Mesh OH for the time being, The transcripts are not updated regularly, could the team please address this, so myself and other interested parties can remain a part of these proceedings? [2011/10/10 Graham Dartmouth]
[12:24] Vir Linden Charlie apologizes for that, resolves to do better...
[12:24] Drongle McMahon They are there now
[12:24] davep Mesh HQ 2 is down (as are 5 of the 8 old mesh beta regions still on the map). Where will the meeting be? (Drongle McMahon)
[12:24] Calla Cela The problem i find with viewer 2 is that it slows my poor computer to a crawl - just selecting a piece of my inventory can take 10 secnds or more. I cannot build using V2/v3
[12:25] davep from charlar: (agenda #4) mesh HQ regions will be back up after this usergroup and will remain as the meeting location. We're going to restart all of them with trunk.
[12:25] davep looks like there's some serious channel drift going on with the mesh regions and that will be cleaned up shortly
[12:25] davep The mesh-development build was stopped at the version of 3.0.6. But the version of beta viewer is 3.1.0. Which viewer should we test? davep_shining-fixes or snowstorm_viewer-development? [2011/10/10 Yuzuru Jewell]
[12:26] Maxwell Graf im afraid to look but did that auto prim spammer get deleted over behind us yet?
[12:26] Vir Linden mesh-dev builds haven't been stopped, but may be nothing new has gone in for a while.
[12:26] Wolfpup Runitai thgat might have been causer but the server kernal updates that have been going on the past week or so
[12:26] davep grab davep_shining-fixes if you want to test graphics issues
[12:27] Drongle McMahon Where are they?
[12:27] Vir Linden At this point mesh-dev only has a handful of fixes that aren't in viewer-dev.
[12:27] Wolfpup Wolfpup < has a more pressing project

[12:27] Ever Student Maxwell: Fewer prims spamming, but also the person there is back from away so presumably is on it.
[12:27] Vir Linden since most of the mesh work has merged down.
[12:29] davep davep_shining-fixes builds are linked to SH-2240 (see description)
[12:29] davep #6. The mesh-development build was stopped. With which viewer should I make the tutorial video about uploading by new dialog? Or till when should I wait? [2011/10/10 Yuzuru Jewell]
[12:29] Vincent Nacon doesn't support in 11.7 (and up) Catalyst though
[12:30] Vir Linden mesh-development hasn't been stopped intentionally - I'll take a look at it.
[12:30] Vir Linden but as above, there's not much new there.
[12:31] davep Vincent: what about 11.7 catalyst?
[12:31] Yuzuru Thank you, vir. I will wait for the update of mesh-project viewer.
[12:32] davep #7. OPEN-105 update [2011/10/10 Wolfpup Lowenhar]
[12:32] Vir Linden I think the only pending change for mesh-dev is to pull latest viewer-development changes, which may or may not be anything you care about from a mesh testing perspective.
[12:32] Yuzuru I want a mesh-project viewer based on 3.1.
[12:33] Vir Linden I'll get it updated today
[12:33] Wolfpup i guess it is trime for me to talk
[12:33] Yuzuru Thank you!
[12:34] Wolfpup i have made some progress on getting hacd into llconvexdecompostion as an Open Source version
[12:34] Blue Alloy ...I'd settle for one that just doesn't crash immediately :¿
[12:35] Wolfpup yesterday i finaly got llconvexdecomposition converted to NOT useing static sln/make files and also fixed a build issue for the current building of the 'stub'
[12:35] Motor Loon convexdecomp. is that for mesh uploading on tpv's?
[12:35] Wolfpup it also builds 'outside' the source tree
[12:36] Vincent Nacon sorry, was distracted with few IMs. it's not supported. None of the viewers are supported in 11.7+
[12:36] Wolfpup yes it is motor and TPV'sd can not use the private lib
[12:36] Vir Linden Sounds encouraging
[12:36] Nebadon Izumi davep can I request that we revisit the SH-2374 topic before the end of the meeting
[12:36] Nebadon Izumi a few of us were not here when the meeting started
[12:36] Nebadon Izumi and missed a chance to talk about it
[12:36] Motor Loon indeed... most most annoying too
[12:36] Wolfpup Runitai you know how the stub will not build on windows right
[12:37] Wolfpup i founs and fix the issue localy but i have not published tit as yet
[12:38] Vir Linden We are fixing the llconvexdecomp build to also build stub.
[12:38] Vir Linden The separate stub build will be going away
[12:38] Vir Linden ie, there will be 2 builds, but only one repo behind them.
[12:38] Geenz Spad yay
[12:38] Wolfpup Vir i know and have been working on that for a wile now my self as im working on vgetting hacd incorperated as a Open Source version
[12:39] Wolfpup all my work is based off of the llconvexdecompostion lib repository
[12:39] Vir Linden OK
[12:39] Maxwell Graf while i realize you may not be able to offer an official opinion on the subject, could you offer a personal opinion on the mesh development project that I have initiated with Karl? We are over 60% of the way to funding it already, so looks like it will be going forward. How do you feel about this, whats your thoughts? any response will help.
[12:40] Wolfpup ther is already building on Tc 3p-hacd which will be built into the open source version of llconvexdecomposition
[12:40] Maxwell Graf we really are hoping you guys will take the code and use it when its done.
[12:41] Nal yeah... there is a lot of interest on that subject'
[12:41] Nebadon Izumi we saw in log before we go here you were hoping for an open source option
[12:41] Vir Linden There are a couple of aspects here. One is we are still discussing whether it makes sense to try to do something internally - see start of the meeting.
[12:41] Maxwell Graf yes.
[12:42] Nebadon Izumi wouldnt it make more sense though with the popularity of the TPV's to make something everyone could use? when you say internally would that require usage of any propritary Havok libraries?
[12:42] Wolfpup Vir when open-105 is finished llconvexdecompostion should not need to build the stub any more as there will be the private(havok based) and open(HACD based version of the lib
[12:42] Vir Linden If we get a request to incorporate code from an external project, we are certainly prepared to consider it, but can't commit without having seen the code.
[12:42] Nebadon Izumi ok thats fair
[12:43] Nebadon Izumi were pretty confident the code will be very high quality coming from Karl
[12:43] Vir Linden As we also mentioned at the start, an internal solution would probably not be based on proprietary havok libs, although we haven't ruled it out completely.
[12:44] davep at the end of the day, it's just like any other external project
[12:44] Maxwell Graf thank you for commenting. It was not started with the goal of going around you guys, but working in parallel, as we realize you have a pretty full plate. Not trying to work against you, but with you. :)
[12:44] davep we look at the value of the change, quality of the code, contributor agreements still apply, all that
[12:44] Nebadon Izumi sure, we really just wanted to know that you were atleast willing to give the code a chance
[12:44] Vir Linden Maxwell, We appreciate that :-)
[12:44] Nebadon Izumi that is very reassuring
[12:44] davep and if it's valuable and we can help out, we help out
[12:44] Skygirl Kline )
[12:44] Maxwell Graf karl did mention he would def be willing to sign agreements, btw.
[12:44] Nebadon Izumi awesome
[12:44] Asha That's good to hear.
[12:45] Maxwell Graf thanks you guys.
[12:45] Jusden Jonstone ty
[12:46] davep as for sticking to only open source solutions, the general policy still applies
[12:46] Wolfpup is there some one else also working on an open source verion of llconvedecomposition?
[12:46] Blue Alloy so... I've got a question, if we run out of agenda items...
[12:46] davep if there's a better technology that's not open source, we won't let the requirement of providing an open source viewer get in the way of integrating the best technology available
[12:46] Vir Linden Wolfpup, I haven't heard. Might want to catch Nyx next time.
[12:47] Wolfpup ok i guess i was reading the chat wrong
[12:47] davep we won't not use a good technology just because it's closed source, but we won't intentionally hamstring open source viewers with said technology
[12:48] Wolfpup i know i have been working on it for a while now
[12:48] davep i.e., with mesh, we were careful to make sure open source viewers could stream and view meshes
[12:48] Siana Gearz davep, that's very official and very nothing-promising :)
[12:48] Jusden Jonstone nods
[12:49] Siana Gearz davep, i'd say convex decomposition is a great example, cause... there was no good reason, from our (user, TPV dev) POV, to use a closed source lib in the first place.
[12:49] Siana Gearz ratcliff's lib has been around for ages.
[12:49] davep hacd didn't work as well as the havok lib
[12:49] davep in our testing
[12:49] Geenz Spad I can understand the reasoning (LL licenses Havok, Havok supports convex decomp generation, so therefore use Havok)
[12:49] Wolfpup Siana Gearz open-105 is for provieding an Open Source version of the lib
[12:49] davep it fell over for any model with more than a few thousand triangles
[12:50] davep same with qhull
[12:50] Siana Gearz i see, ok.
[12:50] davep we DID look at these libraries, havok's version was just better
[12:50] Siana Gearz >.< means we'll have to put some work into it...
[12:50] Nebadon Izumi that is not suprising
[12:50] Nebadon Izumi though Kirstens did a pretty good job
[12:51] Nebadon Izumi we did some pretty complex uploads with it
[12:51] Nebadon Izumi ah
[12:51] Motor Loon Havent had issues using Kirstens S21 here
[12:51] Nebadon Izumi i never got much into it
[12:51] Wolfpup and that is quite a few monts old
[12:52] davep anything else?
[12:52] Blue Alloy I have a quick question
[12:53] Wolfpup hope fuly we can get HACD working @ close to the same level as Havok
[12:53] davep Blue: ?
[12:53] Vir Linden Fire away.
[12:53] Blue Alloy what's ththe project Prep's mentioned in this thread:
[12:54] davep yeah, that's the not-public-but-public jira I linked to earlier
[12:54] Vir Linden possible approach to deformable clothing
[12:54] Blue Alloy ah
[12:54] Blue Alloy unfortunately, I missed the earlier parts of the meeting
[12:54] davep basically, if you look at the raycast display and you see those joints we use for avatar raycasts
[12:55] davep those are scaled based on some of the parametric avatar deformers
[12:55] davep and you can apparently weight your mesh to them if you want to, say, make the belly bigger
[12:55] Blue Alloy ...those aren't the ones that show up when you do Show Collision Skeleton, are they?
[12:55] davep yeah, those are the ones
[12:55] Blue Alloy oh!
[12:55] Blue Alloy ...I think I see
[12:56] Blue Alloy ...this might lead to actual length recognition, then?
[12:56] davep er... scale, actually
[12:57] davep in *theory* weighting to those bones *might* let you make pants that auto-fit any avatar you put them on
[12:57] Blue Alloy where smaller mesh avs can fit into places they appear they should be able to?
[12:57] Vir Linden which is one of the things we're going to be investigating this week
[12:58] Blue Alloy and another (hopefully quick) dumb question: when using offset joints, where does the client put the camera for mouselook?
[12:58] davep and some sample content would go a LONG way to proving that hteory
[12:58] davep Blue: good question
[12:59] Blue Alloy is that base of mSkull?
[12:59] davep I believe it is
[12:59] davep davep looks

[12:59] Blue Alloy or between mEye*?
[13:00] Wolfpup i would think it should put it @ mNose as that would be centered between the eyes
[13:00] Blue Alloy (it's these kinds of questions that rarely get asked, but would be important for helping make mesh avatars much nicer)
[13:00] Vir Linden We're about out of time, and I have to run to another meeting. Thanks, everyone.
[13:01] davep yeah, I don't know
[13:01] Nadi Vemo Nadi Vemo waves

[13:01] Blue Alloy see ya
[13:01] Skygirl Kline byes
[13:01] Ever Student Bye
[13:01] Jusden Jonstone bye ty
[13:01] Yuzuru Thank you, Vir and Runitai.
[13:01] Asha Thanks for the meeting.
[13:01] Blue Alloy ....but please, can we have an LL client that's not an insta-crasher?
[13:02] Blue Alloy I'd like to be able to upload more mesh... LOL!